MZ@ !L!This program cannot be run in DOS mode. $Ns8 k k kj#kkjkj kjkjk~jkej k kkj kj kj kRich kPEdUh" , [ `P X x ( % `$ @.text `.rdataab@@.data; f @.pdata( ~ @@.reloc  @BHL$USWATAUAVAWHHpE3NL}HHӠ HEML}M,MMaLeMQIMtHHHHEHIAHt$hHuIM+L}LuXH^ MA@fHHK|.HHtJH98tBHAHHuHGMH;-L}PuELMPH\$ LHUHLuPHHtIEHUH+HHHDHuH /HxLH 4A|HHt4HHAHA;HCHHuHEH]H;t0@ffH AHH u^HH;uAH uqIELEHJDHLuXIILEL;EEHELL+I뮃u!H =L LHDH ;AA;H HLLH IIOtpHL9>tRL;~HHIHwHtHHUHH+HHHDH iHu)HL9;uL;EL LHUIcAHt$hELEHIMHLD$0HQIHH?L ǻHHHDMHD$(HHL$ L H LIHH L}HMtI/u I}L 3AH څ3HpA_A^A]A\_[]LILEHMuLILEHHE@H HHP LHHHD$  HHu1HtH+u HL 3AH #FH+u HLHMHHt H)uIHpA_A^A]A\_[]@USWATAUAVAWHl$HH1 H3HEE3HMLMLL}J L}H HEL,MMaLeMXIMtHHHHEIAH$HuHM+L}LMHh MAHHK|)HHtJH98tBHAHHuHGMH;mL}uELMH\$ LHUHLuHHtIEHUH+HHHDͷuH oHLH tA|HHt4HHAHBA;HCHHuHEH]H;t0@ffH AHH ubHH;uAH0uxIELELMHJDIILEL;ELuEHELL+I몃u!H yLLHH wA|A;tH KHLLH MLuIIttHLmL9>tRL;~HHIHHtHHUHH+HHHDͷ Hu)HL9;uL;EL LHUIAH$EH]IMHLD$0HIH$H?L HHHDMHD$(HHL$ LH #LIHH 'L}H,MtI/u ILH 3A 3HHH]HuHTHH]IWLuE HHuHMtsI.umI0bH HUHGLIHH H]HMtI.u IH/u HHu H<LH B3A-LHMHt H)uIHMH3HİA_A^A]A\_[]@USWATAUAWHl$HH#| H3HE3LMHMHK HEH HEH HEWH}MH}N,EMyMyL}MbIMt6Ht&HtH[HBHHEHBHHEHHHEIAH$L$LuO4HEM+H]LML4@IIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUHoL}HHtIEHUH+HHHDuH HhH 'mIHt4HHAHHCHHuHEH]I;t"H AHH u`HI;uދH]ǃu|IELLMHJDL}HIH]I;} HELL+I뫃u!H :LLHAH 8{H LLHmH IHC{ILmI9>tYL;~HHIH\HtHHUHH+HHHD NHu)HH9;uL;}L MHUIHEL$H$9}LMMuL ILMI(J9|tII| Ld$0H HL$(L HD$ QIHHILd$0HH fL LHILMH YHL$(HD$ H LHBH HfHLHt H)uHH|LH '3A% Z{3mHBHHEHBHHEHHLMHEMuL ILMLEHUHMHHHLHt H)u}HH|HHMH3HA_A]A\_[]@USVWATAUAWHl$HHv H3HE3LMHMH HEH߈ HEHđ HEWH}MH}N,EM^MyL}MGIMt6Ht&HtH@HBHHEHBHHEHHHEIAL$LuO4HEM+H]LMLfffIIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUH?L}HHtIEHUH+HHHDuH H8H mIHt4HHAHHCHHuHEH]I;t"H AHH u`HI;uދH]ǃu|IEL_LMHJDL}HIH]I;}HELL+I뫃u!H L LHH {H LLH=H IHttILmI9>tRL;~HHIH7HtHHUHH+HHHD )Hu)HH9;uL;}L JMHUI#EL$9}LEMuLILEI"J9|tII| Ld$0H HL$(L uHD$ QIHHILd$0HH PL 6HILMH CHL$(HD$ H jLkH,H kHLHt H)uHH|L 3AH {Fv31HBHHEHBHHEHHLEHEMuLwILELMHMHEHHPHHHOHD$pH; |$hHD$`HK HD$XH D$PHD$HH+ HL$@H ?D$8HD$0LL$(LD$ $ HHu4HtH+u HL 3AH zEuHH+u HHLHt H)uHH|HHMH3HA_A]A\_^[]@USVWATAUAWHl$HH"q H3HEH HMMHE3H HEWH}N M,EMM|$L}M\IMt%HtHHBHHEHHHEID$L$LuMM+H}LeHALHEIIJt(HHtJH90tBHAHHuHFLH;9H}uELMH\$ MHUH苵L}HHtIEHUH+HHHDͷuH ;HLH @AwIHttRL;~HHIHHtHHUHH+HHHDͷ {Hu)HH9;uL;DL <MHUIuAL$EH9}IMHHLD$0H IL צH?HHHMLLIH ϦHL$(H HD$ H HLHt H)uHH|L '3AH @p3.HBHHEHHHHEuHC HHELEMuL ILELMHEHHPHHH } HHD$pH |$hHL$`HD$X|$PHL$HH HD$@H D$8HD$0LL$(LD$ ȋ HHu4HtH+u HL -3A?H FoHH+uH]DHLHt H)uBHH|HHMH3HA_A]A\_^[]HL$USWATAUAVAWHl$HE3NL}oH HEML}M,MMaLeMOIMtHHHHEoIAH$HuIM+L}LuHMADHHK|.HHtJH98tBHAHHuHGMH; L}wuELMwH\$ LHUH_LuwHHtIEHUH+HHHDouH HXLH A|HHt4HHAHA;HCHHuHEH]H;t0@ffH AHH u^HH;uAHluqIELEHJDoLuIILEL;EEHELL+I뮃u!H LLH$H AA;H H_LLH II/tpHL9>tRL;~HHIHWHtHHUHH+HHHDo IHu)HL9;uL;EL ZLHUICAH$ELEoIMHLD$0H.IHϡH?L HHHDMHD$(HHL$ LH LIHH L}oMtI/u IZL 3AEH Dk3LILEoMuLILEoHEgH HHPHLHHHD$pD|$hHD$`HD$XD|$PHD$HHD$@D|$8HD$0HD$(D|$ ۆ HHu1HtH+u HL @3AkH jH+u HsLHMoHt H)u[IHİA_A^A]A\_[]HL$USWATAUAWHHx3NH]@Hց HEMH]M$MMiLmMWIMtHHHHE@IAHt$pLt$hLuMM+H]L}PH=LAfffIIKt<HHtJH90tBHAHHuHFLH;-H]HuELMHH|$ MHUHL}HHHtI$HUH+HHHD@uH /HxLH 4A{IHt4HHAHA;HGHHuHEH}I;t0@ffHAHH uZHI;uދH=unI$LEHJD@L}PIILEL;EHELL+I뱃u!H ALZLHHH ?AA;H HLLH IHStpII9tRL;~HHIH{HtHHUHH+HHHD@ mHu)HH9uL;DL MHUIgAALt$hHt$pHU@IMHLD$0HNIHH?L ĜHHHDMHD$(HHL$ LH LIHH H]@H=HtH+u HsLH 3Aqf3HxA_A]A\_[]HHHU@HuHHHU@HM8~HM@HHt H)uHHxA_A]A\_[]@USWATAUAWHl$HHb H3HEHy LMHEN,H2w HMHE3H}Hm} HEHu HEWWMEMMMyL}M'IMtGHt7Ht&HtHHBHHEHBHHEHBHHEHHHEIAH$L$LuO4HEM+H]LML<"@IIKt)HHtJH90tBHAHHuHFLH;}H}uELD$ LMMHUHϦL}HHtIEHUH+HHHDuH HșH mIHt4HHAHSHCHHuHEH]I;t"H AHH u`HI;uދH]u|IEL!LMHJDL}HIH]I;} HELL+I뫃u!H L LHH {H fL LH͘H dIH{ILmI9>tYL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;}L MHUIEL$H$9}H9}uH#HHEH9}uHHHEHEHuHi HHEIvJ9|tII|aLd$0H KHL$(L ʗHD$ oIHHHHILd$0H HHLL mHD$(MHiLMHL$ H LHcH HHLHt H)u"HH|LH w3Aza3HBHHEHBHHEHBHHEHHHEH9}uHHHEH9}uHHHEHEHuHh HHELMLEHUHMHD$ ϒHHLHt H)u^HH|HHMH3HA_A]A\_[]@USVWATAUAVAWHl$HH\ H3HE3LMHMHs HEH=u HEHw HEWH}MH}N,EMVMyL}M?IMt6Ht&HtH8HBHHEHBHHEHHHEIALuO4HEM+H]LMLQ+IIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUH/L}HHtIEHUH+HHHDuH H(H mIHt4HHAHHCHHuHEH]I;t"H AHH ru`HI;uދH]ǃu|IEL'*LMHJDL}HIH]I;} HELL+I뫃u!H L)LHH {H L)LH-H IH{ILmI9>tYL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;}L )MHUIE9}LEMuLILEI+J9|tII|Ld$0H HL$(L [HD$ QIHHILd$0HH 6L HILMH )HL$(HD$ H PL)(HH QHZfHLHt H)uHH|LH '3A *\3eHBHHEHBHHEHHLEHEMuL[ILELMHMHEH|$pHHPHHH$i HD$hH8u HD$`Hp D$XHD$PH(o HL$HH LD$@HD$8LL$0Lˉ|$(D$ w LHu#L5NAHt@H+u:H/H+u HIIHHu)L5 AMH E3ZHHtH.u HhHLHt H)uRHH|HHMH3HA_A^A]A\_^[]@USVWATAUAVAWHl$HHV H3HEHrl HMMHE3Hq HEWH}N M,EMM|$L}MRIMt%HtHHBHHEHHHEID$LuMM+H}LeH/AL@fHEIIJt(HHtJH90tBHAHHuHFLH;H}uELMH\$ MHUH+L}HHtIEHUH+HHHDͷuH H$LH AwIHttRL;~HHIH)HtHHUHH+HHHDͷ Hu)HH9;uL;DL \-MHUIAEH9}IMHHLD$0H IL H?HHHML,LIH wHL$(H HD$ H HHLHt H)u0HH|LH }3AV3cHBHHEHHHHEuHo HHELEMuLILELMHEH|$pHHP HH ~c Ho HL$hHD$`|$XHL$PH HD$HHh D$@ HD$8LL$0Lˉ|$(D$ zq LHu#L5AgHt@H+u:H)/H+u HIIHHu)L5nAlMH |E3\UHHtH.uHDHLHt H)uHH|HHMH3HA_A^A]A\_^[]@USVWATAUAVAWHl$HH Q H3HEHBb LM3HEHMHk HEWH}MN,EMEMyL}M.IMt%HtH-HBHHEHHHEIALuO4HEM+H]LML5IIKt)HHtJH90tBHAHHuHFLH;MH}uELD$ LMMHUH蟕L}HHtIEHUH+HHHDuH OHH WmIHt4HHAH#HCHHuHEH]I;t"H AHH u`HI;uދH]u|IEL4LMHJDL}HIH]I;} HELL+I뫃u!H jLk4LHqH h{H 6L74LHH 4IHs{ILmI9>tYL;~HHIHHtHHUHH+HHHD ~Hu)HH9;uL;}L 3MHUIxE9}LEMuLILEI0J9|tII|Ld$0HLHD$(L ˆHD$ [IHHILd$0HH HLHHHL$(L uHIHD$ LMH L2HxH HHLHt H)u2HH|LH w3AnP3KHBHHEHHLEHEMuLILELMHEH|$pHHP HH ] Hi HL$hHD$`|$XHL$PH ҩHD$HH] D$@HD$8LL$0Lˉ|$(D$ k LHu#L5ȷAHt@H+u:H>/H+u H-II HHu)L5AMH vE3qOHHtH.u HHLHt H)uHH|HHMH3,HA_A^A]A\_^[]@USVWATAUAVAWHl$HH@K H3HEHc LM3HEHMHf HEWH}MN,EMEMyL}M.IMt%HtH-HBHHEHHHEIALuO4HEM+H]LML7IIKt)HHtJH90tBHAHHuHFLH;mH}uELD$ LMMHUH迏L}HHtIEHUH+HHHDuH oHH wmIHt4HHAHCHCHHuHEH]I;t"H AHH u`HI;uދH]u|IELg6LMHJDL}HIH]I;} HELL+I뫃u!H L6LHH {H VL5LHH TIH{ILmI9>tYL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;}L G5MHUI蘯E9}LEMuL ILEI0J9|tII|Ld$0Hl~HD$(L HD$ [IHHILd$0H.~H ҀHLHHHL$(L HIHD$ LMH L_4HH HHLHt H)uRHH|LH /r3AJ3KHBHHEHHLEHEMuLILELMHEH|$pHHP HH W Hc HL$hHD$`|$XHL$PH BHD$HH_ D$@HD$8LL$0Lˉ|$(D$ e LHu#L5AHt@H+u:H^/H+u HMII*HHu)L5A MH qE3IHHtH.u HHLHt H)uHH|HHMH3LHA_A^A]A\_^[]@USWATAUAWHl$HHcE H3HEH\ HMHEJH\ H]HE3H}H%] HEH"` HEWWMMEMMMyL}MpIMtGHt7Ht&HtHcHBHHEHBHHEHBHHEHHHEIAH$L$LuO4L+H}LmL AALfHEIIHtHHtJH90tBHAHHuHFLH;9H}uDLL$ HULMMH苉L}HHtHHUH+HHHDuH <H|MH AAxIHt=HfHAHA;HCHHuHEH]I;t$fH AHH ucHI;uދH]u}HL @LEHJDL}IHLEH]M;DHELL+I먃u!H HLa@LHO{H FAzA;H L/@LH}{H IHStpII9>tRL;~HHIH{HtHHUHH+HHHD mHu)HH9;uL;DL ?MHUIgAL$H$EHEHuHHHEI6J9|tII|Ld$0H zHL$(L zHD$ QIHHILd$0HH zL lzHILMH yzHL$(HD$ H L>HbzH HfHLHt H)u"HH|LH /l3A zD3HBHHEHBHHEHBHHEHHHEHEHuHHHELMLEHUHMHD$ *H@HLHt H)uHH|HHMH3HA_A]A\_[]HL$USWATAUAVAWHHpE3HV L}HMHEML}HN$M IYH]HfIMtHHHHEHIAHt$hHuJ4M+L}LmXL ?AM@ffHEXHHI|HHtGH98t?HAHHuHGH;LL}PuELL$ HULMPLH螄L}PHHtI$HUH+HHHDHuH NHwMH SAzHHt?HfffHAHA;HCHHuHEH]H;t$fH AHH Ӿu\HH;u3H]؃usI$L x>LEHJDHE3IILEL;HELL+I믃u!H _L(>LHfvH ]AA;H 0L=LHvH .IImtyHL9>t[LeDL;~HHIHHtHHUHH+HHHDH ~Hu)HL9;uL;EL g=LHUIxAEHt$hEI}J|HtII|H}@QItAH GH]sLt$0L uHD$(L<HuHD$ H ''HHHEHHMH:HHuPHtEHHMHHt H)uLH b3A3?3HpA_A^A]A\_[]HHFHMHHHt H)u6HHpA_A^A]A\_[]@USWATAUAWHl$HH; H3HEHL LMHEN,HV HMHE3H}HU HEHT HEWWMEMMMyL}M'IMtGHt7Ht&HtHHBHHEHBHHEHBHHEHHHEIAH$L$LuO4HEM+H]LML j@IIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUHL}HHtIEHUH+HHHDuH HrH mIHt4HHAHsHCHHuHEH]I;t"H AHH 2u`HI;uދH]u|IELhLMHJDL}HIH]I;} HELL+I뫃u!H LhLHqH {H LkhLHqH IHõ{ILmI9>tYL;~HHIHܴHtHHUHH+HHHD ηHu)HH9;uL;}L gMHUIȟEL$H$9}H9}uHCHHEH9}uH϶HHEHEHuHHHEIvJ9|tII|aLd$0H knHL$(L pHD$ oIHHHHILd$0H nHpHLL pHD$(MHpLMHL$ H LfHpH ²HˢHLHt H)uBHH|LH \3AW:3HBHHEHBHHEHBHHEHHHEH9}uHôHHEH9}uHOHHEHEHuHHHELMLEHUHMHD$ HHLHt H)u~HH|HHMH3ޠHA_A]A\_[]@USVWATAUAVAWHl$HH5 H3HEHrN LM3HEHMHP HEWH}MN,EMEMyL}M.IMt%HtH-HBHHEHHHEIALuO4HEM+H]LMLpIIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUHozL}HHtIEHUH+HHHDuH HhmH 'mIHt4HHAHHCHHuHEH]I;t"H AHH u`HI;uދH]u|IELoLMHJDL}HIH]I;} HELL+I뫃u!H :L+oLHAlH 8{H LnLHmlH IHC{ILmI9>tYL;~HHIH\HtHHUHH+HHHD NHu)HH9;uL;}L _nMHUIHE9}LEMuLаILEI0J9|tII|Ld$0HiHD$(L kHD$ [IHHILd$0HhH kHLHgkHHL$(L EkHIHD$ LMH LwmHHkH HHLHt H)uHH|LH ?]3AZ53KHBHHEHHLEHEMuLILELMHEH|$pHHP HH cB H|N HL$hHD$`|$XHL$PH rHD$HHI D$@HD$8LL$0Lˉ|$(D$ _P LHu#L5AHt@H+u:H/H+u HIIڰHHu)L5SAMH #\E3A4HHtH.u HHLHt H)uHH|HHMH3HA_A^A]A\_^[]@USWATAUAWHl$HH0 H3HE3LMHMHF HEHB HEHJ HEHrB HEHK HEWH}WH}MN,EMMMyL}M'IH {BkHHB HHEHBHHEHBHHEHBHHEHHHEIAH$L$LuO4AHEM+H]LMLjffIIKt)HHtJH90tBHAHHuHFLH;ͬH}uELMLT$ MHUHtL}HHtIEHUH+HHHDͷu!H ϬHgMH ԩyIHt>HffHHHCHAHuHEH]I;t$fH HH YupHAI;uދH]IEL$LMAHJDL}HIH]I;}HELL+I롃u!H ѫL~LHeH ϨqH L~LHfH IHکtvILmI9>tTfL;~HHIHHtHHUHH+HHHDͷ Hu)HH9;uL;}L ~MHUIEL$H$9}H9}uHcHHEH9}uHI HHEHMHuH `I HHMIaJ9|tII|LL eMIHHHHL dL;MHdLMHdLd$0HD$(L}HL$ HdH H H @HLHt H)uHH|LH V3A.3HB HHEHBHHEHBHHEHBHHEHHHEH9}uHHHEH9}uHG HHEHMHuH G HHMHELMLEHUHL$(HMHD$ 蛲HHLHt H)uHH|HHMH3HA_A]A\_[]Ðffffff@USVWATAUAWHl$HH* H3HE3LMHMH*A HEH/C HEHD HEWH}MH}HN,EMMyL}MsIMt6Ht&HtHdHBHHEHBHHEHHHEIAL$LuO4HEM+H]LML8IIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUHOnL}HHtIEHUH+HHHDuH HHaH mIHt4HHAHӨHCHHuHEH]I;t"H AHH u`HI;uދH]ǃu|IELLMHJDL}HIH]I;} HELL+I뫃u!H LLH!`H {H LLHM`H IH#{ILmI9>tYL;~HHIH<HtHHUHH+HHHD .Hu)HH9;uL;}L MHUI(EL$9}H9}uHHHEI}fDJ9|tII|HuLd$0H _HL$(L w_HD$ H LXH_H IHtMHttYL;~HHIH,HtHHUHH+HHHDͿ Hu)HH9;uL;}L OMHUIEL$H$9}LEMuLILEI-J9|tII|Ld$0HWHD$(L [ZHD$ [IHHILd$0HWH BZHLH'ZHHL$(L ZHIHD$ LMH FLWHZH GHPHLHt H)uHH|LH L3A?$3^HBHHEHHLEHEMuLYILEHUHMHHLHt H)uLHH|HHMH3謊HĸA_A]A\_[]@USWATAUAWHl$HH H3HE3HMH}H36 HEHP4 HEH: HEWH}J L,MEMMyL}MIMt6Ht&HtHHBHHEHBHHEHHHEIAH$L$LuLM+H}LeHALfHEIIJt(HHtJH90tBHAHHuHFLH;H}uELMH\$ MHUH dL}HHtIEHUH+HHHDuH HWLH AwIHttRL;~HHIH HtHHUHH+HHHD Hu)HH9;uL;DL dMHUIAL$H$EH9}IMHHHLD$0H _UIL EUH?HbUHHMLЖLIH =UHL$(H iHD$ H sH|@HLHt H)uHH|LH A3A-J3HBHHEHBHHEHHHHEuH8 HHEH9}uH8 HHELMMuL TILMLEHUHMĽHؐHLHt H)uBHH|HHMH3袅HA_A]A\_[]H XH%@SH HAXHHu9HAHHHt{HCXHuH [HHHH [HHCXH [H(HHDyHLAXHQXMtI(u Iz3H(@SH HAHHHuHIH љHCHHuH [HHCHH [H(LHt3HBt#HHIHIPHHt H)u3H(H *HWH H(HAPHHAPH(LHt3HBt#HHIPIPPHt H)u3H(H HWH zH(@SH HA@HHu0HC@HuH [HHC@H [H(LHu!H}WH FH  H(HB u H|WHHI@IP@Ht H)u3H(HL$USWATAUAWHHx3HJ5 H]@MHEMH]N,MIyH}H;IMtHHHHE@IAHt$pLt$hLuO4M+H]LePL LDIIKt,HHtJH90tBHAHHuHFLH;H]HuELL$ HULMHMH\LeHHHtIEHUH+HHHD@u!H HOMH yIHt>HffHAHvHGHHuHEH}I;t$fHAHH 3u]HI;uދH}ЃuvIEL HLEHJD@LePIILEL;HELL+I뮃u!H LLHNH 끃t!H LʕLHNH IHɒttII9tVLmАL;~HHIHHtHHUHH+HHHD@ ޔHu(HH9uL;L 8MHUI|Lt$hHt$pxiIJ|@tII|IH HKL|$0L PNHD$(LĔHMNHD$ H H]@H=HtH+u HLH A3Ae3HxA_A]A\_[]HHHE@HU@HM8~HM@HHt H)uHHxA_A]A\_[]HA`HHA`HaHHWHuHHH%HQhHHHEHH\$WH HHHuSHH eHSAH HHHHt H)u͐3H\$0H _H;tHBuH HqRH H\$0H _@SVWH`H1 H3HD$PIHHMt&IZHx~HH Ey3MHHOH$Hu HHHHuNHL$0HT$(H , E3IH\$(H+Hu HŏHWH\$8D$@QHHuH+H-}uoHvdH5- HT$8HFLIHH + H|$@H+Hu H0H.u H!Hu"H-}LH 8D3t3H/u HHH$HL$PH3K|H`_^[@SH HHHu1H9t!;y3H [HHH [HbHH [H\$WH HHHuSH<H HPAH ߍHHHHt H)u3H\$0H _H;tHB uH 8HQPH H\$0H _@SH HHHu1H9t!:y3H [HHH [HbHH [H(LHt?H;Et6HB u!H HEPH TH(H3HIHt H)u 3H(@SH HHHu͋HuH [HHH [@SVWH`Hq H3HD$PIHHMt&IHx~HH u3MHHOH$Hu HHHHuNHL$0HT$(H W* E3IH\$(׌H+Hu HHWH\$8D$@HHuH+H-1zuoHdHu) HT$8HFLIHH ) H|$@BH+Hu HpH.u HaHu"H-yLH v4D33H/u H+HH$HL$PH3xH`_^[@VH0HHHt HH0^H\$HHl$PH|$XH-" HHHELHHE33D$ H)H N H+Hu HvHthHOHHHHtH/Hu H?Ht1HHH tFHAHHu HHH2 H|HkHHHu3 HH HHl$PH\$HH|$XH0^@USWATAUAWHl$HHc H3HE3LMHMH" HEH HEH5' HEWH}MH}N,EMMyL}MrIMt6Ht&HtHuHBHHEHBHHEHHHEIAH$L$LuO4HEM+H]LMLD@IIKt)HHtJH90tBHAHHuHFLH;]H}uELD$ LMMHUHPL}HHtIEHUH+HHHDuH _HCH gmIHt4HHAH3HCHHuHEH]I;t"H AHH u`HI;uދH]ǃu|IELLMHJDL}HIH]I;} HELL+I뫃u!H zLãLHBH x{H FLLHBH DIH{ILmI9>tYL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;}L MHUIpEL$H$9}H9}uHHHELMMuL ILMIMJ9|tII|5Ld$0H HAHLL hAHD$(MHdALMHL$ H LH^AH HsHLHt H)u HH|LH e.3Ax 3HBHHEHBHHEHHHEH9}uHHHELMMuL ILMLEHUHMŷHfHLHt H)uHH|HHMH3qHA_A]A\_[]H\$WH H3HIpHtH{pH)u0HK HtH{ H)uHK@HtH{@H)uHKHHtH{HH)uHKPHtH{PH)ũHKXHtH{XH)uHK`HtH{`H)uHKhHtH{hH)uHK8H{8Ht H)uhHHtHH)uIHHtHH)u*HHtHH)u HHtHH)uHKxHtH{xH)uӂH\$83H _@SH HqH{(t HqH9HH [H%BH\$Ht$WH HIHIpHHt IօHK Ht HօHK@Ht HօHKXHt HօHK`Ht HօutHK8Ht HօubHHt HօuMHHt Hօu8HHt Hօu#HKxHtHHH\$0Ht$8H _H3H\$0Ht$8H _HQPLH DH%HL$USWATAUAVAWHl$HE3NL}oH HEML}M,MMaLeMOIMtHHHHEoIAH$HuIM+L}LuHMADHHK|.HHtJH98tBHAHHuHGMH;=L}wuELMwH\$ LHUHHLuwHHtIEHUH+HHHDouH ?H;LH D~A|HHt4HHAHA;HCHHuHEH]H;t0@ffH AHH Âu^HH;uAHduqIELEHJDoLuIILEL;EEHELL+I뮃u!H MLLHT:H K}AA;H H:LLH  }II_~tpHL9>tRL;~HHIH}HtHHUHH+HHHDo yHu)HL9;uL;EL RLHUIshAH$ELEoIMHLD$0H^7IH9H?L 9HHHDMHD$(H9HL$ LťH ~LIH9H |L}oMtI/u I}L +l3A{H &3LILEoMuL2~ILEoHEgH THHPHLHH~HD$pD|$hHD$`HD$XD|$PHD$HHD$@D|$8HD$0HD$(D|$   HHu1HtH+u H|L pk3AH A&,H+u H|LHMoHt H)u|IHİA_A^A]A\_[]Hl$Ht$ WH0HA0IHHHtLLBHHt*H}t#MHl$ LH`Hl$PHt$XH0_E3Hl$PHt$XH0_H$H\$H<u}LBHκzHHt:3HzHu.H+u H{H |H?LGPH y3:LLHH"H+HuHo{HQLL!HHH\$HHl$PHt$XH0_@USVWATAUAWHl$HHH3HE3HMH}H HEH? HEH HEWH}J L$MEMMyL}M{IMt6Ht&HtHHBHHEHBHHEHHHEIAL$LuLM+H}LmHҵALfHEIIJt HHtJH90tBHAHHuHFLH;zH}uELMH\$ MHUH+BL}HHtI$HUH+HHHDuH zH$5LH wAwIHttRL;~HHIH)wHtHHUHH+HHHD zHu)HH9;uL;DL MHUIbAL$EH9}IMHHHLD$0H 3IL m3H?H3HHMLLIH e3HL$(H xHD$ H uHLHt H)u"wHH|L e3AH v3QHBHHEHBHHEHHHHEuH HHEH9}uH HHELEMuLwILELULMHEHHPHHH P HQwHD$pH] |$hHL$`HD$XH D$PHD$HLL$@Lˉ|$8HL$0H 3SLT$(D$ 8 HHu4HtH+u HuL d3A'H YHH+uHuDHLHt H)uuHH|HHMH3cHA_A]A\_^[]@USVWATAUAWHl$HH2H3HEH LM3HEHMH HEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LMLcfffIIKt)HHtJH90tBHAHHuHFLH;MuH}uELD$ LMMHUH<L}HHtIEHUH+HHHDͿuH OuH/H WrmIHt4HHAH#wHCHHuHEH]I;t"H AHH vu`HI;uދH]u|IEL/LMHJDL}HIH]I;}HELL+I뫃u!H jtLLHq.H hq{H 6tLLH.H 4qIHsrttILmI9>tRL;~HHIHqHtHHUHH+HHHDͿ tHu)HH9;uL;}L MHUI\EL$9}LEMuLsILEI'J9|tII|Ld$0HV+HD$(L -HD$ [IHHILd$0H+H -HLH-HHL$(L -HIHD$ LMH rL1H-H ofHLHt H)uBqHH|L _3A-H 3HBHHEHHLEHEMuLqILELMHEHHPHHH  HqHD$pH |$hHL$`HD$X|$PHL$HH RHD$@H) D$8HD$0LL$(LD$  HHu4HtH+u HRpL ^3AyH HH+u H#pHLHt H)u pHH|HHMH3m]HA_A]A\_^[]@USVWATAUAWHl$HHH3HE3LMHMH: HEH HEHT HEWH}MH}N,AEMwMyL}M`IMt3I+t$I+tI;cHBHHEHBHHEHHHEIAL$LuO4HEM+H]LMLWIIKt)HHtJH90tBHAHHuHFLH;}oH}uELD$ LMMHUH6L}HHtIEHUH+HHHDuH oH)H lmIHt4HHAHSqHCHHuHEH]I;t"H AHH qu`HI;uދH]ǃu|IELVLMHJDL}HIH]I;} HELL+I뫃u!H nL+VLH(H k{H fnLULH(H dkIHl{ILmI9>tYL;~HHIHkHtHHUHH+HHHD nHu)HH9;uL;}L _UMHUIVEL$9}H9}uHK HHELEMuLmILEIGJ9|tII|2Ld$0H g%HL$(L 'HD$ ]II+I+I;M;Ld$0H #%H'HLL 'HD$(LMH 'LMLD$ H lL`TH'H iHLHt H)uRkHH|L Y3AH C3DHBHHEHBHHEHHHEH9}uH HHELEMuLkILELMHMHEAHHPHHHkHD$pH|$hHD$`H HD$XH DT$PHD$HH HL$@H ADT$8HD$0LL$(LD$ p HHu4HtH+u H4jL X3AH .HH+u HjHLHt H)uiHH|HHMH3OWHA_A]A\_^[]HL$USWATAUAWHHx3HZ H]@MHEMH]N,MIyH}H;IMtHHHHE@IAHt$pLt$hLuO4M+H]LePL LDIIKt,HHtJH90tBHAHHuHFLH;iH]HuELL$ HULMHMH1LeHHHtIEHUH+HHHD@u!H iH$MH fyIHt>HffHAHkHGHHuHEH}I;t$fHAHH Cku]HI;uދH}ЃuvIEL LEHJD@LePIILEL;HELL+I뮃u!H hLLH"H e끃t!H hLRLH#H eIHfttII9tVLmАL;~HHIHeHtHHUHH+HHHD@ hHu(HH9uL;L MHUIPLt$hHt$pxiIJ|@tII|IH gHL|$0L `"HD$(LLH]"HD$ H dH]@H=THtH+u HfLH J3AQu3HxA_A]A\_[]HHHE@HU@HM8.HM@HHt H)ueHHxA_A]A\_[]@USVWATAUAWHl$HHRH3HE3LMHMHrHEH_HEH$ HEWH}MH}N,EM^MyL}MGIMt6Ht&HtH@HBHHEHBHHEHHHEIAL$LuO4HEM+H]LMLfffIIKt)HHtJH90tBHAHHuHFLH;MeH}uELD$ LMMHUH,L}HHtIEHUH+HHHDuH OeHH WbmIHt4HHAH#gHCHHuHEH]I;t"H AHH fu`HI;uދH]ǃu|IEL LMHJDL}HIH]I;}HELL+I뫃u!H jdL LHqH ha{H 6dL[ LHH 4aIHsbttILmI9>tRL;~HHIHaHtHHUHH+HHHD dHu)HH9;uL;}L MHUILEL$9}LEMuLcILEI"J9|tII| Ld$0H HL$(L HD$ QIHHILd$0HH L HILMH HL$(HD$ H bL HH _HLHt H)uRaHH|L O3AH k 31HBHHEHBHHEHHLEHEMuLaILELMHMHEHHPHHHaHD$pH|$hHD$`H HD$XHD$PHD$HHHL$@H 7D$8HD$0LL$(LD$  HHu4HtH+u HH`L N3A,H j HH+u H`HLHt H)u`HH|HHMH3bMHA_A]A\_^[]@USVWATAUAWHl$HHH3HEHLMHEN,HHMHE3H}HHEHAHEWWMEMMoMyL}MXIMtGHt7Ht&HtHKHBHHEHBHHEHBHHEHHHEIAL$LuO4HEM+H]LML;fffIIKt)HHtJH90tBHAHHuHFLH;]_H}uELD$ LMMHUH&L}HHtIEHUH+HHHDͰuH __HH g\mIHt4HHAH3aHCHHuHEH]I;t"H AHH `u`HI;uދH]u|IELLMHJDL}HIH]I;}HELL+I뫃u!H z^LLHH x[{H F^LLHH D[IH\ttILmI9>tRL;~HHIH[HtHHUHH+HHHDͰ ^Hu)HH9;uL;}L MHUIFEL$9}LEMuL]ILEI-J9|tII|Ld$0H HL$(L HD$ QIHHILd$0HH L HILMH HL$(HD$ H \LHH YHLHt H)ub[HH|L I3A1H 3AHBHHEHBHHEHBHHEHHLEHEMuL[ILELULMHMHEHHPHHH[HD$pHDD$hHD$`HHL$XH L6D$PHD$HHLL$@LD$8HD$0LT$(D$ HHu4HtH+u HLZL H3AH HH+uHZDHLHt H)uZHH|HHMH3bGHA_A]A\_^[]H(u H03HZE3H*L8AHHuH(H4HAHaZHAHVZHHLZHH>ZHHH(@USVWATAUAWHl$HHH3HEH LM3HEHMHHEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LML3fffIIKt)HHtJH90tBHAHHuHFLH; YH}uELD$ LMMHUH_ L}HHtIEHUH+HHHDͿuH YHXH VmIHt4HHAHZHCHHuHEH]I;t"H AHH Zu`HI;uދH]u|IELLMHJDL}HIH]I;}HELL+I뫃u!H *XLLH1H (U{H WLwLH]H TIH3VttILmI9>tRL;~HHIHWUHtHHUHH+HHHDͿ IXHu)HH9;uL;}L MHUIC@EL$9}LEMuLVILEI'J9|tII|Ld$0HHD$(L HD$ [IHHILd$0HH |HLHaHHL$(L ?HIHD$ LMH VLHBH SfHLHt H)uUHH|L C3AH sV3HBHHEHHLEHEMuLUILELMHEHHPHHH fHgUHD$pHs|$hHL$`HD$X|$PHL$HH *HD$@HAD$8HD$0LL$(LD$ NHHu4HtH+u HTL B3AH oHH+u HSHLHt H)uSHH|HHMH3-AHA_A]A\_^[]@USVWATAUAWHl$HHBH3HE3LMHMHRHEHHEHHEWH}MH}N,EM^MyL}MGIMt6Ht&HtH@HBHHEHBHHEHHHEIAL$LuO4HEM+H]LMLfffIIKt)HHtJH90tBHAHHuHFLH;=SH}uELD$ LMMHUHL}HHtIEHUH+HHHDuH ?SH H GPmIHt4HHAHUHCHHuHEH]I;t"H AHH Tu`HI;uދH]ǃu|IELLMHJDL}HIH]I;}HELL+I뫃u!H ZRLLHa H XO{H &RL_LH H $OIHcPttILmI9>tRL;~HHIHOHtHHUHH+HHHD yRHu)HH9;uL;}L MHUIs:EL$9}LEMuLPILEI"J9|tII| Ld$0H HL$(L HD$ QIHHILd$0HH L HILMH HL$(HD$ H PLH| H MHLHt H)uBOHH|L =3AH 31HBHHEHBHHEHHLEHEMuLOILELMHMHEHHPHHHOHD$pH|$hHD$`HHD$XHD$PHD$HHsHL$@H (D$8HD$0LL$(LD$ tHHu4HtH+u H8NL <3A"H HH+u H NHLHt H)uMHH|HHMH3R;HA_A]A\_^[]HL$USWATAUAWHHx3NH]@HHEMH]M$MMiLmMWIMtHHHHE@IAHt$pLt$hLuMM+H]L}PH=LAfffIIKt<HHtJH90tBHAHHuHFLH;MH]HuELMHH|$ MHUHL}HHHtI$HUH+HHHD@uH MHLH JA{IHt4HHAHOA;HGHHuHEH}I;t0@ffHAHH COuZHI;uދH=]unI$LEHJD@L}PIILEL;EHELL+I뱃u!H LLLHH IAA;H LHLLH IIHJtpII9tRL;~HHIH JHtHHUHH+HHHD@ LHu)HH9uL;DL NMHUI4AALt$hHt$pHU@IMHLD$0HIHH?L THHHDMHD$(HIHL$ LH ~KLIHCH HH]@H=8HtH+u HJLH i3Ad3HxA_A]A\_[]HHHU@HuHJHHU@HM8~HM@HHt H)uIHHxA_A]A\_[]HL$USWATAUAVAWHl$HE3NL}oHHEML}M,MMaLeMOIMtHHHHEoIAH$HuIM+L}LuHv!MADHHK|.HHtJH98tBHAHHuHGMH;}IL}wuELMwH\$ LHUHLuwHHtIEHUH+HHHDouH IHLH FA|HHt4HHAHRKA;HCHHuHEH]H;t0@ffH AHH Ku^HH;uAH< uqIELEHJDoLuIILEL;EEHELL+I뮃u!H HLLHH EAA;H ^HHLLH `EIIFtpHL9>tRL;~HHIHEHtHHUHH+HHHDo HHu)HL9;uL;EL *LHUI0AH$ELEoIMHLD$0HIH?H?L HHHDMHD$(H HL$ LH >GLIHH BDL}oMtI/u IEL k43A'H '3LILEoMuLrFILEoHEgHHPHLHH GHHFHD$pHTD|$hHL$`HD$XD|$PHL$HHD$@D|$8HL$0H 'HD$(D|$ =HHu1HtH+u HEL 33AoH ^H+u HDLHMoHt H)uDIHİA_A^A]A\_[]@USVWATAUAWHl$HHBH3HEH\LM3HEHMHHEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LML-fffIIKt)HHtJH90tBHAHHuHFLH;]DH}uELD$ LMMHUH L}HHtIEHUH+HHHDͿuH _DHH gAmIHt4HHAH3FHCHHuHEH]I;t"H AHH Eu`HI;uދH]u|IEL,LMHJDL}HIH]I;}HELL+I뫃u!H zCLk,LHH x@{H FCL7,LHH D@IHAttILmI9>tRL;~HHIH@HtHHUHH+HHHDͿ CHu)HH9;uL;}L +MHUI+EL$9}LEMuLBILEI'J9|tII|Ld$0HfHD$(L HD$ [IHHILd$0H(H HLHHHL$(L HIHD$ LMH AL*HH >fHLHt H)uR@HH|L .3ArH [3HBHHEHHLEHEMuL@ILELMHEHHPHHH H@HD$pH|$hHL$`HD$X|$PHL$HH u$HD$@HD$8HD$0LL$(LD$ HHu4HtH+u Hb?L .3AH tHH+u H3?HLHt H)u?HH|HHMH3},HA_A]A\_^[]@USVWATAUAWHl$HHH3HE3LMHMHHEHHEHdHEWH}MH}N,EM^MyL}MGIMt6Ht&HtH@HBHHEHBHHEHHHEIAL$LuO4HEM+H]LMLK5fffIIKt)HHtJH90tBHAHHuHFLH;>H}uELD$ LMMHUHL}HHtIEHUH+HHHDuH >HH ;mIHt4HHAHc@HCHHuHEH]I;t"H AHH "@u`HI;uދH]ǃu|IEL4LMHJDL}HIH]I;}HELL+I뫃u!H =L3LHH :{H v=L3LHH t:IH;ttILmI9>tRL;~HHIH:HtHHUHH+HHHD =Hu)HH9;uL;}L 3MHUI%EL$9}LEMuLC<ILEI"J9|tII| Ld$0H AHL$(L HD$ QIHHILd$0HH L HILMH HL$(HD$ H <L#2HH  9HLHt H)u:HH|L *)3AH 31HBHHEHBHHEHHLEHEMuL;ILELMHMHEHHP HH:HD$pH|$hHD$`HHD$XHOD$PHD$HHHL$@H |$8HD$0LL$(LD$ HHu4HtH+u H9L -(3A9 H HH+uH]9DHLHt H)uB9HH|HHMH3&HA_A]A\_^[]H\$WH H oHAH9uHHtHH='@H=HLG7HH'HHH HHtHB:H HHHu%:HuH 9HLH 6Hu H=&KHCHHMLMtAf;HHHHHu9H=&Hu H8LH 3Ai3H\$0H _Hu H7HH\$0H _HL$USWATAUAWHHx3HH]@MHEMH]N,MIyH}H;IMtHHHHE@IAHt$pLt$hLuO4M+H]LePL P4LDIIKt,HHtJH90tBHAHHuHFLH;7H]HuELL$ HULMHMHLeHHHtIEHUH+HHHD@u!H 7HMH 4yIHt>HffHAH9HGHHuHEH}I;t$fHAHH S9u]HI;uދH}ЃuvIEL 3LEHJD@LePIILEL;HELL+I뮃u!H 6L2LHH 3끃t!H 6L2LHH 3IH4ttII9tVLmАL;~HHIH 4HtHHUHH+HHHD@ 6Hu(HH9uL;L 2MHUILt$hHt$pxiIJ|@tII|IH 5HL|$0L pHD$(L1HmHD$ H 2H]@H="HtH+u H$4LH 3A腺3HxA_A]A\_[]HHHE@HU@HM8莱HM@HHt H)u3HHxA_A]A\_[]@USVWATAUAWHl$HHbH3HEHLM3HEHMH?HEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LML@fffIIKt)HHtJH90tBHAHHuHFLH;}3H}uELD$ LMMHUHL}HHtIEHUH+HHHDͿuH 3HH 0mIHt4HHAHS5HCHHuHEH]I;t"H AHH 5u`HI;uދH]u|IEL>LMHJDL}HIH]I;}HELL+I뫃u!H 2L>LHH /{H f2LW>LHH d/IH0ttILmI9>tRL;~HHIH/HtHHUHH+HHHDͿ 2Hu)HH9;uL;}L =MHUIEL$9}LEMuL31ILEI'J9|tII|Ld$0HHD$(L HD$ [IHHILd$0HHH HLHHHL$(L HIHD$ LMH 0L<HH -fHLHt H)ur/HH|L 3A> H Ƶ3HBHHEHHLEHEMuL0ILELMHEHHPHHH H/HD$pH|$hHL$`HD$X|$PHL$HH  HD$@HD$8HD$0LL$(LD$ HHu4HtH+u H.L #3A H ߴHH+u HS.HLHt H)u=.HH|HHMH3HA_A]A\_^[]H\$Ht$WH IHHMH,H~sHCu Hj-tdH,HHtSHLHHܬH/Hu H-H+u Hs-HH\$0Ht$@H _J,HHuH\$03Ht$@H _LD$HL$USVWATAUAWHHpHZHj3HEHuHMHuHMI+HELH"HHHtHHGHHEHID$Lt$hLuM4LL-fHEPIIJ|HHtJH98tBHAHHuHGLH;9-HuXuILMXLl$ MHUHL}XHHtJHUH+HHHDHu!H 7-HMH <*qIHt2HfHAH/twHCHHuHEH]I;t(fDH AHH .uJHI;uދƃucJHJDHL}IM;HELL+IuH a,HrLMH c)똃H 4,HLMH 6)IHu*toII96tQL;~fHIH)HHtHHUHH+HHHLH ,Hu$HH93uL;~L MHUILt$hHuHHuHHHHH{H\$0HHH?L HLAHHDHHD$(HHL$ LIH +HH (H#HMHHt H)u)LH 3AHpA_A]A\_^[]HwHHuHHuH5<*HHuHHM@H葢HMHHt H)u2)HpA_A]A\_^[]@USVWATAUAWHl$HH«H3HEHLM3HEHMHHEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LMLCfffIIKt)HHtJH90tBHAHHuHFLH;(H}uELD$ LMMHUH/L}HHtIEHUH+HHHDͿuH (H(H %mIHt4HHAH*HCHHuHEH]I;t"H AHH r*u`HI;uދH]u|IELBLMHJDL}HIH]I;}HELL+I뫃u!H 'LSBLHH ${H 'LBLH-H $IH&ttILmI9>tRL;~HHIH'%HtHHUHH+HHHDͿ (Hu)HH9;uL;}L AMHUIEL$9}LEMuL&ILEI'J9|tII|Ld$0HHD$(L eHD$ [IHHILd$0HH LHLH1HHL$(L HIHD$ LMH P&L@HH Q#fHLHt H)u$HH|L j3A H c&3HBHHEHHLEHEMuLb%ILELMHEHHPHHH 6H7%HD$pHC|$hHL$`HD$X|$PHL$HH 5 HD$@HD$8HD$0LL$(LD$ HHu4HtH+u H#L 3A H |?HH+u H#HLHt H)u#HH|HHMH3HA_A]A\_^[]H\$Ht$WH IHHMHN"H~sHCu H"tdH%"HHtSHLHHH/Hu H"H+u H"HH\$0Ht$@H _!HHuH\$03Ht$@H _@SH HAHHt(3%uHCH H9H0u HU%u_H%HKHtHCH)u3"HHtHǃH)u"HCHH@H [HH [@USVWATAUAWHl$HHH3HEHLM3HEHMHoHEWH}MN,EMMMyL}M6IMt%HtH5HBHHEHHHEIAL$LuO4HEM+H]LMLKfffIIKt)HHtJH90tBHAHHuHFLH;!H}uELD$ LMMHUHL}HHtIEHUH+HHHDͿuH !HH mIHt4HHAH#HCHHuHEH]I;t"H AHH B#u`HI;uދH]u|IELILMHJDL}HIH]I;}HELL+I뫃u!H LILHH {H L_ILHH IHttILmI9>tRL;~HHIHHtHHUHH+HHHDͿ  Hu)HH9;uL;}L HMHUIEL$9}LEMuLcILEI'J9|tII|Ld$0HHD$(L 5HD$ [IHHILd$0HxH HLHHHL$(L HIHD$ LMH LGHH !fHLHt H)uHH|L : 3A H [3HBHHEHHLEHEMuL2ILELMHEHHPHHH HHD$pH|$hHL$`HD$X|$PHL$HH HD$@HD$8HD$0LL$(LD$ HHu4HtH+u HL S 3Au H tHH+u HHLHt H)umHH|HHMH3 HA_A]A\_^[]{H\$Ht$WH HIHIHHt Iօu&HHtHHH\$0Ht$8H _H3H\$0Ht$8H _@SH HHHIHCHHHt H)uHnHHHYHHt H)ue3H [HAHHAH\$WH HHHt$HHIH)u H{3H\$0H _HHHIH)uHHC3H\$0H _W@USVWATAUAWHl$HHrH3HE3HMH}HHEHϷHEHDHEWH}J L$MEMMyL}M{IMt6Ht&HtHHBHHEHBHHEHHHEIAL$LuLM+H}LmHQALfHEIIJt HHtJH90tBHAHHuHFLH;iH}uELMH\$ MHUHL}HHtI$HUH+HHHDuH kHLH pAwIHttRL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;DL |OMHUIAL$EH9}IMHHHLD$0H IL H?HHHMLNLIH HL$(H !HD$ H +HLHt H)uHH|L J3Az H 3SHBHHEHBHHEHHHHEuHSHHEH9}uHGHHELEMuLILELMHMHEHHP HHHD$pHԩ|$hHD$`HHD$XH(D$PHD$HḪHL$@H h|$8HD$0LL$(LD$ HHu4HtH+u HL &3A H oHH+u HVHLHt H)u@HH|HHMH3HA_A]A\_^[]@SVHxH—H3HD$`HM~DH +HLD$0L 3HD$(LLHt$ H H 3Mt!IAHx~IH L3HWH|$pD$PH\$HHHuH=Ht|H+uvH<kH HT$HH߬LHGIHHL$PH GHHtH+u HH/u HHu H=?LH 3A03H|$pHHL$`H3Hx^[HL$USWATAUAVAWHHE3HLePMHEMLeHN,M IYH]HBIMtHHHHEPIAHt$xHuJ4M+LeL}XL KMEHffHHK|/HHtJH98tBHAHHuHGMH;LeuELL$ HULMLHL}HHtIEHUH+HHHDPu!H HMH zHHt>HffHAHVHCHHuHEH]H;t$fH AHH ubHH;uAH]Ѓu}IEL ILEHJDPL}XIILEL;DeHHELL+I몃u!H LbILHH zH eL.ILHH cIItuHL9&tWLmL;~HHIHHtHHUHH+HHHDP Hu*HL9#uL;DeHL HLHUIEHHt$xD9eH|dI}N9dPtII|H}@ItzH HLt$0L 0HD$(L4HH-HD$ H cLePMtI,$u IL 3AH G3HHHEPHHUHEPE3H )IHEH}QHHtH/u HzHuL 3AH wҕH+u HIL%"I$HMPHt H)u)IHĀA_A^A]A\_[]@USVWATAUAWHl$HHH3HE3HMH}HҧHEHHEHHEWH}J L$MEMMyL}M{IMt6Ht&HtHHBHHEHBHHEHHHEIAL$LuLM+H}LmHVYALfHEIIJt HHtJH90tBHAHHuHFLH;H}uELMH\$ MHUHL}HHtI$HUH+HHHDuH HLH  AwIHttRL;~HHIH HtHHUHH+HHHD  Hu)HH9;uL;DL WMHUIAL$EH9}IMHHHLD$0H WIL =H?HZHHML|VLIH 5HL$(H a HD$ H k HLHt H)u HH|L 3A H F3SHBHHEHBHHEHHHHEuHHHEH9}uHHHELEMuLP ILELMHMHEHHP HH( HD$pH|$hHD$`H$HD$XHhD$PHD$HH HL$@H X|$8HD$0LL$(LD$ HHu4HtH+u H L f3AO H ׵"HH+u H HLHt H)u HH|HHMH3HA_A]A\_^[]HHM~FH HALD$0L -HD$(L]H*HD$ H `3HHMt!IAHx~IH _]R3HHHH钝@USVWATAUAWHl$HHrH3HE3HMH}HHEHϥHEHDHEWH}J L$MEMMyL}M{IMt6Ht&HtHHBHHEHBHHEHHHEIAL$LuLM+H}LmHhALfHEIIJt HHtJH90tBHAHHuHFLH;iH}uELMH\$ MHUHL}HHtI$HUH+HHHDuH kHLH pAwIHttRL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;DL fMHUIAL$EH9}IMHHHLD$0H IL H?HHHMLfLIH HL$(H !HD$ H +HLHt H)uHH|L J3AT H 3SHBHHEHBHHEHHHHEuHSHHEH9}uHGHHELEMuLILELMHMHEHHP HHHD$pHԗ|$hHD$`HHD$XH(D$PHD$HH̜HL$@H ȱ|$8HD$0LL$(LD$ HHu4HtH+u HL &3A H HH+u HVHLHt H)u@HH|HHMH3HA_A]A\_^[]@USVWATAUAWHl$HHH3HE3HMH}HJHEHHEHHEWH}J L$MEMMyL}M{IMt6Ht&HtHHBHHEHBHHEHHHEIAL$LuLM+H}LmHuALfHEIIJt HHtJH90tBHAHHuHFLH;H}uELMH\$ MHUHL}HHtI$HUH+HHHDuH HLH AwIHttRL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;DL rMHUIAL$EH9}IMHHHLD$0H WIL =H?HZHHML(rLIH 5HL$(H aHD$ H kHLHt H)uHH|L 3A H SF3SHBHHEHBHHEHHHHEuHHHEH9}uHHHELEMuLPILELMHMHEHHPHHH(HD$pH|$hHD$`H$HD$XH؛D$PHD$HHHL$@H |$8HD$0LL$(LD$ HHu4HtH+u HL f3A H /"HH+u HHLHt H)uHH|HHMH3HA_A]A\_^[]HL$USWATAUAWHHx3NH]@HvHEMH]M$MMiLmMWIMtHHHHE@IAHt$pLt$hLuMM+H]L}PH=rLAfffIIKt<HHtJH90tBHAHHuHFLH;=H]HuELMHH|$ MHUHL}HHHtI$HUH+HHHD@uH ?HLH DA{IHt4HHAHA;HGHHuHEH}I;t0@ffHAHH uZHI;uދH=punI$LEHJD@L}PIILEL;EHELL+I뱃u!H QLpLHXH OAA;H "HLLH $IHctpII9tRL;~HHIHHtHHUHH+HHHD@ }Hu)HH9uL;DL oMHUIwAALt$hHt$pHU@IMHLD$0H^IHH?L ԵHHHDMHD$(HɵHL$ L=oH LIHõH H]@H=HtH+u HLH A3A3HxA_A]A\_[]HHHU@HuH&HHU@HM8ސHM@HHt H)uHHxA_A]A\_[]@USVWATAUAWHl$HH{H3HEH$HMMHE3HHEWH}N M,EMM|$L}M\IMt%HtHHBHHEHHHEID$L$LuMM+H}LeHwALHEIIJt(HHtJH90tBHAHHuHFLH;H}uELMH\$ MHUHL}HHtIEHUH+HHHDͷuH HLH AwIHttRL;~HHIHHtHHUHH+HHHDͷ  Hu)HH9;uL;DL uMHUIAL$EH9}IMHHLD$0H IL gH?HHHMLuLIH _HL$(H HD$ H HLHt H)uHH|L 3A H s{3.HBHHEHHHHEuHӔHHELEMuLILELMHEHHP HH pHqHD$pH}|$hHL$`HD$X|$PHL$HH HD$@HD$8HD$0LL$(LD$ XHHu4HtH+u HL 3Af H yzHH+uHDHLHt H)uHH|HHMH32HA_A]A\_^[]HL$USWATAUAWHHx3NH]@H6HEMH]M$MMiLmMWIMtHHHHE@IAHt$pLt$hLuMM+H]L}PH=wLAfffIIKt<HHtJH90tBHAHHuHFLH;H]HuELMHH|$ MHUHL}HHHtI$HUH+HHHD@uH HLH A{IHt4HHAHrA;HGHHuHEH}I;t0@ffHAHH #uZHI;uދH=evunI$LEHJD@L}PIILEL;EHELL+I뱃u!H L vLHH AA;H HLLH IHtpII9tRL;~HHIHHtHHUHH+HHHD@ Hu)HH9uL;DL VuMHUIAALt$hHt$pHU@IMHLD$0HIH_H?L 4HHHDMHD$(H)HL$ LtH ^LIH#H bH]@H=gHtH+u HLH ɖ3ADv3HxA_A]A\_[]HHHU@HuH&HHU@HM8辉HM@HHt H)u~HHxA_A]A\_[]@VH0HHHHtEHRqHuHEqHyHt-HH0^H;tH H%H 3H0^H\$@HhHl$HHH|$(Lt$  HHH H/Hu HHtbHHHtBH HHHtULHHH/Du H^EyEH+u HJ3H|$(Hl$HH\$@Lt$ H0^H 'H tHHLHHt0LHHH/Du HEuH H [HcHHHt0LH<HH/Du HqEH _H NPHH3HHt6H;tLHH:H/u Hy#H H H@USWAUAVHl$HHuoH3HEpHgwE333Ht/H;u3XcH HqH H5cHH$L$L$H wHxHUbHH BH5bHH QHrxH bHHALRxH vHLxaHL=%= t_LAHL$@HMT$8HD$0L|$(D$ DHU3ra3HwHZa3H HwH;a3H iHwHaH OwʯaDHtL55L5V5H`H|HL5julH`LLAZI37HH`H0uL_E3HHvHt`9(vt#LHXH tSI`LH7`IHHuLtII  `H >v)_H *v赪_H vs_H v_H u9_H š\LH_L HLtH;D_L ILtH_L ILtH^I)6u IH Du^H 3ٜLHu ^HMH uy WI)7u I|H 3膜LHu ^HMH tEy mWI)7u I)H{H蒻LHu 6^H }E3I.HHu WI)7u IHG{HߟLHuL5(!WH{MH ty VI)7u I{H)3u HmH .3wHHu {]HLH s6yL5VH)3u HH IHaLHHH"oH `yL5> BH)3u H{HbH JLfE3HD$0H_HD$(H6nHD$ T{HHu XIH aLHHHrH X` yL5h aBH)3u HHaH ZJLdE3HD$0H_HD$(HmHD$ zHHu HHp`LHHHTnH _pyL5 AH)3u HEHfaH PLhcE3HD$0H^HD$(HmHD$ zHHu z "HH_LHHHijH "_yL52z +AH)3u HH`H DPLbE3HD$0H]HD$(HelHD$ yHHu GHZ_LHHHhH ^:yL5 @H)3u HH@`H QLBbE3HD$0HN]HD$(HkHD$ xHHu T FH^LHHHiH ]yL5T ?H)3u HtH_H VQLaE3HD$0H\HD$(H/kHD$ MxHHu QFH$^LHHH0iH Q]yL5a Z?H)3u HH_H PL4bE3HD$0H\HD$(HjHD$ wHHu  EHy]LHHH=lH \iyL5ƽ >H)3u H>H^H HQLabE3HD$0H}[HD$(HiHD$ wHHu k EH\LHHHZnH \ξyL5+k $>H)3u HH]H PLaE3HD$0HZHD$(H^iHD$ |vHHu DH3\LHHHmH [3yL5 =H)3u HHa]H ;Ls^E3HD$0HGZHD$(HhHD$ uHHu % CH[LHHHbH Z蘽yL5% <H)3u HmH\H ';L^E3HD$0HYHD$(H(hHD$ FuHHu JCHZLHHHAgH JZyL5Z S<H)3u HH;\H :L^E3HD$0HYHD$(HgHD$ tHHu BHrZLHHHhH YbyL5;H)3u H7H[H 1:Lz_E3HD$0HvXHD$(HfHD$ tHHu nBHYLHHHkH YǻyL5$n;H)3u HH[H 9L\E3HD$0HWHD$(HWfHD$ usHHu yAH,YLHHHPbH yX,yL5:H)3u HHZH ;9L\E3HD$0H@WHD$(HeHD$ rHHu @HXLHHHbH W葺yL5 9H)3u HfHYH 9L[E3HD$0HVHD$(H!eHD$ ?rHHu C@HWLHHH2cH CWyL5SL9H)3u HH\YH 8L>[E3HD$0H VHD$(HdHD$ qHHu ?HWLHHHWcH V[yL58H)3u H0HXH 8LZE3HD$0HoUHD$(HcHD$ qHHu  ?HVLHHHbH VyL58H)3u HH6XH 7L@YE3HD$0HTHD$(HPcHD$ npHHu ?r>H%VLHHH9^H rU%yL5?{7H)3u HHWH 7L}ZE3HD$0H9THD$(HbHD$ oHHu =HBeLH T蛷yL56H)3u HpH!WH =LCYE3HD$0HSHD$(H+bHD$ IoHHu QM=HbLH ^TyL5nQg6H)3u HH/THUWE3IHH=THE蜠H)7Hu H3Hu <HbLH R{yL55H)3u HXHRHHH9 IuHIHtH+H tbvH fbLIHIӴHHu '<HCHHZLMtA HHuL55.5H)3u HHZLH Ry c;H)7u HzHQHHH9 IuH=IHtH+H auH aLHHHHHu I;HGHHYLMtA.HHu :H)7u HHYLH -Q3yL5"4H)3u HHPHHH9 PHuHOHHtH+H `uH `L)HHHHHu i:HCHH]YLMtANHHuL5wp3H)3u HHYLH FPy 9H)7u HHPHHH9 GuH=GHtH+H _3tH _L[GHLG7HHu 9HGHHXLMtApHHu 9H)7u HHXLH oO3yL5d]2H)3u HH=OHHH9 FuHFHtH+H ^SsH ^LFH|FWHHu 8HCHHWLMtAHHuL51H)3u H1HWLH Ny 7H)7u HH_NHHH9 EuH=EHtH+H ^urH ^LEHEyHHu 7HGHHWLMtAHHu C7H)7u HZHcWLH M3AyL50H)3u HHMHHH9 EuHEHtH+H :]qH ,]LDHD虯HHu 6HCHHAWLMtAHHuL5/H)3u HsHVLH L\y )6H)7u H@HLHHH9 FDuH=EDHtH+H \\pH N\LDHD軮HHu 6HGHHsVLMtAHHu 5H)7u HH5VLH K3yL5.H)3u H`HKHHH9 vCuHuCHtH+H |[oH n[LOCH@CۭHHu /5HCHHULMtAHHuL5=6.H)3u HHULH Ky k4H)7u HHJHHH9 BuH=BHtH+H ZnH ZLBHrBHHu Q4HGHHUULMtA6HHu 3H)7u H޽HULH 5J3żyL5*#-H)3u HHJHHH9 AuHAHtH+H YnH YLAHAHHu q3HCHHUTLMtAVHHuL5x,H)3u HHTLH NIy 2H)7u HļH%IHHH9 AuH= AHtH+H X;mH XL@H@?HHu 2HGHHSLMtAxHHu  2H)7u H HySLH wH3yL5le+H)3u HHEHHHH9 :@uH9@HtH+H X[lH WL@H@_HHu 1HCHHRLMtAHHuL5*H)3u H9HRLH G"y 0H)7u HHgGHHH9 l?uH=k?HtH+H "W}kH WLE?H6?聩HHu 0HGHH)SLMtAHHu K0H)7u HbHRLH F3IyL5)H)3u H&HFHHH9 >uH>HtH+H BVjH 4VLu>Hf>表HHu /HCHHRLMtAڼHHuL5(H)3u H{HRLH Edy 1/H)7u HHHEHHH9 =uH==HtH+H dUiH VUL=H=çHHu /HGHHRLMtAHHu .H)7u HHQLH D3yL5'H)3u HhHDHHH9 =uH=HtH+H ThH vTL<H<HHu 7.HCHH3QLMtAHHuL5E>'H)3u HHPLH Dy s-H)7u HHCHHH9 P<uH=O<HtH+H ShH SL)<H<HHu Y-HGHHQLMtA>HHu ,H)7u HHPLH =C3͵yL52+&H)3u HH CHHH9 ;uH;HtH+H R!gH RLY;HJ;%HHu y,HCHH5PLMtA^HHuL5%H)3u HHOLH VBy +H)7u H̵H-BHHH9 :uH=:HtH+H QCfH QL:H:GHHu +HGHHOLMtAHHu +H)7u H(HaOLH A3yL5tm$H)3u HHMAHHH9 :uH:HtH+H QceH PL9H9gHHu *HCHHOLMtAHHuL5ɢ#H)3u HAHNLH @*y )H)7u HHo@HHH9 49uH=39HtH+H *PdH PL 9H8艢HHu )HGHH9NLMtA¶HHu S)H)7u HjHMLH ?3QyL5"H)3u H.H?HHH9 d8uHc8HtH+H JOcH ly 9(H)7u HPH>HHH9 7uH=7HtH+H lNbH ^NLo7H`7ˠHHu (HGHH[MLMtAHHu 'H)7u HHMLH >3yL5 H)3u HpH=HHH9 6uH6HtH+H MaH ~ML6H6HHu ?'HCHHLLMtA$HHuL5MF H)3u HŰHNLLH =y {&H)7u HH<HHH9 5uH=5HtH+H L aH LL5H5 HHu a&HGHHKLMtAFHHu %H)7u HHKLH E<3ծyL5:3H)3u HH<HHH9 (5uH'5HtH+H K)`H KL5H4-HHu %HCHH%KLMtAfHHuL5H)3u HHJLH ^;y $H)7u HԮH5;HHH9 Z4uH=Y4HtH+H JK_H JL34H$4OHHu $HGHHJLMtAHHu $H)7u H0HiJLH :3yL5|uH)3u HHU:HHH9 3uH3HtH+H Jk^H JLc3HT3oHHu #HCHHILMtAHHuL5ћH)3u HIHILH 92y "H)7u HHw9HHH9 2uH=2HtH+H 2I]H $IL2H2葛HHu "HGHHILMtAʯHHu ["H)7u HrHHLH 83YyL5H)3u H6H8HHH9 1uH1HtH+H RH\H DHL1H1豚HHu "HCHHAHLMtAHHuL5 H)3u HHGLH 7ty A!H)7u HXH7HHH9 1uH=1HtH+H tG[H fGL0H0әHHu '!HGHHkGLMtA HHu  H)7u HH-GLH 73yL5H)3u HxH6HHH9 N0uHM0HtH+H FZH FL'0H0HHu G HCHHFLMtA,HHuL5UNH)3u HͩHVFLH $6y H)7u HH5HHH9 /uH=/HtH+H EZH ELY/HJ/HHu iHGHHELMtANHHu H)7u HHELH M53ݧyL5B;H)3u HH5HHH9 .uH.HtH+H D1YH DL.Hz.5HHu HCHHELMtAnHHuL5H)3u HHELH f4y H)7u HܧH=4HHH9 -uH=-HtH+H CSXH CL-H-WHHu HGHHELMtAHHu !H)7u H8HDLH 33yL5}H)3u HH]3HHH9 "-uH)-HtH+H CsWH CL-H,wHHu HCHHDLMtAHHuL5ٔH)3u HQHBDLH 2:y H)7u HH2HHH9 t,uH=s,HtH+H :BVH ,BLM,H>,虔HHu HGHHCLMtAҨHHu cH)7u HzH{CLH 13ayL5ƓH)3u H>H1HHH9 +uH+HtH+H ZAUH LAL+H~+蹓HHu  HCHHBLMtAHHuL5H)3u HHBLH 0|y IH)7u H`H0HHH9 *uH=*HtH+H |@TH n@L*H*ےHHu /HGHHBLMtAHHu H)7u HHALH 03yL5H)3u HH/HHH9 *uH*HtH+H ?SH ?L)H)HHu OHCHHKALMtA4HHuL5]VH)3u HբHALH ,/y H)7u HH/HHH9 H)uH=G)HtH+H >SH >L!)H)HHu qHGHHALMtAVHHu H)7u HH@LH U.3yL5JCH)3u H¡H#.HHH9 x(uHw(HtH+H =9RH =LQ(HB(=HHu HCHH}@LMtAvHHuL5H)3u HH8@LH n-y H)7u HHE-HHH9 'uH='HtH+H =[QH <L'Ht'_HHu HGHH?LMtAHHu )H)7u H@H?LH ,3'yL5H)3u HHe,HHH9 &uH&HtH+H <{PH <L&H&HHu HCHH?LMtAHHuL5H)3u HYH>LH +By H)7u H&H+HHH9 &uH= &HtH+H B;OH 4;L%H%衍HHu HGHHa>LMtAڡHHu kH)7u HH#>LH *3iyL5Ό H)3u HFH*HHH9 <%uH;%HtH+H b:NH T:L%H%HHu HCHH=LMtAHHuL5# H)3u HH|=LH )y QH)7u HhH!-H L:E3HD$03H)HD$(H&8HD$ DEHHu HH*LHHHW:H `)y H)7u H֜H,H L3E3HD$03H)HD$(H7HD$ DHHu H3LH (qy >H)7u HUH,H /L9E3HD$03H(HD$(H7HD$ 1DHHu 5Hp9LH ^(y H)7u HԛH+H L8E3HD$03H(HD$(H6HD$ CHHu H8LH 'oy <H)7u HSH,+H L7E3HD$03H'HD$(H6HD$ /CHHu 3H7LH \'y H)7u HҚH:H$Hd:H$H5:H$H:H$H9H$H`9H$H8H$xH8H$pHk8H$hHL8H$`H-8H$XH7H$PH7H$HHx7H$@HQ7H$8H*7H$0H6H$(H6H$ HM6H$H.6H$H6H$H5H$H5H$H5H$HC5H$H5H$H4H$H4H$H3H$H3H$H3H$H:3H$H2H$H2H$H2H$H1H$H1H$H01H$H0HD$xH/HD$pH/L b-5LF-H(HD$hH/HD$`Hg/HD$XH;/HD$PH.HD$HH.HD$@H'.HD$8H-HD$0Ho-HD$(H#-HD$ akHHu 'H,LH .$y ' H)7u HH$HHH9 uH=HtH+H 4HH 4LHtHHu _s HGHL2H0LMtAIy _ H)7u HH^#HHH9 uH=HtH+H 3tGH {3LHxHHu ` HGHLx1HI0LMtAy `? H)7u HVH"HHH9 |uH={HtH+H j3FH \3LUHFфHHu a% HGHL0H/LMtAy a H)7u HH"HHH9 uH=HtH+H ,&FH ,LH*HHu b~ HGHL*0H.LMtATy b H)7u HHi!HHH9 NuH=MHtH+H L2EH >2L'H胃HHu c HGHL/HT.LMtAy cJ H)7u HaH HHH9 uH=HtH+H }(DH o(LH܂HHu f0 D$@H!D$8E3ɉt$0E3HD$(HHHD$ PrHHu f H)7u HHCH;5tSH;$tJHIHHuL5fH@LMuL5fHE3E3HH)3uH.fDMu7HGL;wH;u HGJJ\HHIHAHH]Hk,LH 1ÑH)3u HHHHH9 #uH"HtH+H ,$CH ,LH(HHL-HH ALHH)3u H$H+MH {3 4I)7u IHVHHH9 uH HtH+H Y+lBH K+LdHUpHHE3H; rAH)1uERHHHH9 uL=MtI+H *AH *LHLMtHIGIL+H*LMtA)xI)7IiihxgM)/L50u,I!L5ggGL5fHt=L)+u8H-Ht6HcHHbufL5~H|$DH)7u H.H E3H)HALMtA2y k"H E3H{)HALMtAy k-HHL,HH"_L,HH!m@L%HH, N!LO-HH!/L-HH"La+HH'!L$HHpҍLK$HHqL-HH!gL*HH uHL*HH V)L*HH 7 LX,HH L(HHL ,HH ڌL$HHLT$HHoL-(HHC}PL"HH ^1L'HH?L0+HH LQ+HHLr,HH LS(HHËLd,HHwL}(HHXL%HHf9L#HHmGLX%HH~(LI%HHo L(HHhLk+HHˊL*HHL HH`L%HHnAL'HHUO"L0+HH0LI"HHL"HH(L#$HHIӉL$HHJL$HHxlL HHHzxQL(HH_x6L%HH:DxH(LH )yL5xL)/uNHCH)7HI)6u IHXHGL5;xH=tHH=!tEtMH (CD3H _Ht;H)HKu*{"HuH oHBH oHL$HL$H$HMpH3vHİA^A]_[]@USVWATAUAWHl$HH H3HE3LMHMHJ"HEH&HEH&HEWH}MH}N,EM^MyL}MGIMt6Ht&HtH@HBHHEHBHHEHHHEIAL$LuO4HEM+H]LML7fffIIKt)HHtJH90tBHAHHuHFLH;H}uELD$ LMMHUHOL}HHtIEHUH+HHHDuH HBH mIHt4HHAHHCHHuHEH]I;t"H AHH Bu`HI;uދH]ǃu|IELLMHJDL}HIH]I;}HELL+I뫃u!H ʇLLHAH Ȅ{H L{LHAH IHӅttILmI9>tRL;~HHIHHtHHUHH+HHHD Hu)HH9;uL;}L MHUIoEL$9}LEMuLcILEI"J9|tII| Ld$0H aAHL$(L 5AHD$ QIHHILd$0HH AL @HILMH AHL$(HD$ H *LH@H +HLHt H)uHH|L Js3Ak H s1 31HBHHEHBHHEHHLEHEMuL7ILELMHMHEHHPHHHHD$pH|$hHD$`H $HD$XHO!D$PHD$HHkHL$@H lD$8HD$0LL$(LD$ %HHu4HtH+u HL Ir3A H r0 HH+u HyHLHt H)ubHH|HHMH3pHA_A]A\_^[]@USWATAUAWHl$HHH3HEHHMHEJHH]HE3H}H HEH HEWWMMEMMMyL}MpIMtGHt7Ht&HtHcHBHHEHBHHEHBHHEHHHEIAH$L$LuO4L+H}LmL ALfHEIIHtHHtJH90tBHAHHuHFLH;H}uDLL$ HULMMH JL}HHtHHUH+HHHDuH H=MH AxIHt=HfHAHA;HCHHuHEH]I;t$fH AHH CucHI;uދH]u}HL LEHJDL}IHLEH]M;DHELL+I먃u!H ȁLQLH;H ~AzA;H LLH;H ~IHtpII9>tRL;~HHIH~HtHHUHH+HHHD Hu)HH9;uL;DL MHUIiAL$H$EHEHuH`HHEI6J9|tII|Ld$0H W;HL$(L +;HD$ QIHHILd$0HH ;L :HILMH :HL$(HD$ H LH:H !}H*mfHLHt H)u~HH|LH +3A 3HBHHEHBHHEHBHHEHHHEHEHuH HHELMLEHUHMHD$ GH@HLHt H)u~HH|HHMH3bkHA_A]A\_[]HL$USWATAUAWHHx3HH]@MHEMH]N,MIyH}H;IMtHHHHE@IAHt$pLt$hLuO4M+H]LePL (LDIIKt,HHtJH90tBHAHHuHFLH;}H]HuELL$ HULMHMH/ELeHHHtIEHUH+HHHD@u!H }H(8MH zyIHt>HffHAHHGHHuHEH}I;t$fHAHH cu]HI;uދH}ЃuvIEL LEHJD@LePIILEL;HELL+I뮃u!H |LLH6H y끃t!H |LjLH(7H yIHzttII9tVLmАL;~HHIHzHtHHUHH+HHHD@ }Hu(HH9uL;L MHUI eLt$hHt$pxiIJ|@tII|IH {H3L|$0L 6HD$(LdH}6HD$ H xH]@H=hHtH+u H4zLH J!3A;3HxA_A]A\_[]HHHE@HU@HM8.HM@HHt H)uyHHxA_A]A\_[]H iH%y̋Q @tÃuÃDLL$ DD$SVATAUHHAL{Hl$p3H|$@LLt$8DL|$0tAHHt.vHvx;}HHD9DuHHH$Im`DM}hIMupH$Iu`IuhIup{HHuDHtHmu HxMtI/u IxMI.IIM`MehI}pIm`M}hMupHt H)ufxMtI,$u IQxHtH/u H=xD$H=vEHu>{HHM?@1D`HHEHwLcD;}-MMF9du J JHH)w;u0Dp@HIcHzHHtdHD5HcLI;~#HHHDAHHIAI;IHDdL_ID$LsE3HIwHHt HD`(wvHtH+u HvHtH/u HvL|$0Lt$8H|$@Hl$pHHA]A\^[@USWHl$HHouHHuH_[]HHUwL LHEwHEo! EsHMoH|$(HD$ H$BH3HxHHUL(LHEHHEH+EoB Es`HMoH|$(HD$ rBHHHHULLHEHHEH+EoA Es HMoH|$(HD$ BHHHHULLHEHvEoB Es`HMoH|$(HD$ LMAH2H_HHUL LHEHP HEHHEHoEob EsaHMoH|$(HD$ NAHHHHULLHEHHEHHEHHEH)Eo Es;HMoH|$(HD$ @HaH~HHUL(LHEH`HEH@HEHEob EsHMoH|$(HD$ m@HHHHULLHEH`HEHEoB EsHMoH|$(HD$ @HHHHULxLHEHHEHpHEH`HEH Eo EsHMoH|$(HD$ ?H8H=HHUL( LHEHHEH`HEHEoc EsHMoH|$(HD$ ,?HHHHULLHEH`HEHMEoB EsEbHMoH|$(HD$ >H}HrHHUL`LHEH`HEH HEH8HEH@HEH0 HEH1`Eo EsqHMoH|$(HD$ @>HHHHULLHEH HEH HEH`HEH HEH HEH HEH HEHHEHdEo% Es@HMoH|$(HD$ =HTH9HHULLHEHh HEH HEHEob Es3HMoH|$(HD$ (=HHHHUL LHEHHEH`HEHHEHHEHhHEH HEHhHEH@HEHHEH HEH0HEHHEHHEH@ HEH HEH HEH` HEHXHEH HEHHE'HHE/HHE7HRYEoEsWvHMoH|$(HD$ ;HHHHULLHEH HEH HEH`HEH HEHHEHHEH HEH HEH HEHXHEH@HEH]Eo Es-d#HMoH|$(HD$ ;HHHHULLHEH(HEHEoAEsHMoH|$(HD$ :HHYHHULLHEH(HEHEoAEsDHMoH|$(HD$ S:H<HHHULLHEH HEH HEH`HEHEo Es HMoH|$(HD$ 9HHHHULLHEH`HEHNEoB Es{%HMoH|$(HD$ }9HvH#HHUL@LHEH HEHHEH`HEHEo EsHMoH|$(HD$ 9HHHHULLHEH0HEH`HEHEoc Es-FHMoH|$(HD$ 8HHBHHUL LHEH0HEHHEH`HEHEo EsFHMoH|$(HD$ &8H7HHHUL8 LHEH HEH HEH`HEHYEo EsHMoH|$(HD$ 7HHVHHUL0LHEH HEH HEHHEH`HEHhEo Es1GHMoH|$(HD$ /7HPHHHUL LHEH HEH`HEHEEoc Es'HMoH|$(HD$ 6HHjHHUL LHEH HEHHEH`HEH7Eo EsGHMoH|$(HD$ N6HHHHULLHEH`HEHEoB Es'(HMoH|$(HD$ 5H'HHHULLHEH HEH`HEHEoc EsrHHMoH|$(HD$ 5HH)HHULLHEHp HEH HEH`HEHvEo Es(HMoH|$(HD$ 5HVHHHULLHEHHEH`HEH Eoc Es> HMoH|$(HD$ 4HHHHHUL@LHEHHEH`HEHظEoc Es)HMoH|$(HD$ 74HH HHULxLHEHHEH`HEHEoc Es HMoH|$(HD$ 3H-Hr HHUL( LHEH HEHHEH`HEHEo EszHMoH|$(HD$ V3HH HHUL LHEH HEHHEH`HEH1Eo Es؊HMoH|$(HD$ 2HQH HHUL LHEH HEHHEH`HEH{Eo EsTHMoH|$(HD$ j2HH HHUL LHEH HEHPHEH`HEHmEo EsHMoH|$(HD$ 1HuH HHULhLHEHHEH`HEHEoc EslHMoH|$(HD$ 1HH/ HHUL LHEH HEHHEH`HEHԨEo Esk HMoH|$(HD$ 1HH HHULxLHEH8 HEHH HEHHEH`HEH HEHh HEHx HEH HEHHEHHEHHEHvIEo Es ,HMoH|$(HD$ E0HH HHULLHEH HEHHEH`HEH HEHP HEH HEH0 HEH@ HEH8HEH@HEH HEHHEH HEHEo Es%<HMoH|$(HD$ a/HH HHUL LHEH HEHHEH`HEHHEHqEo EsHMoH|$(HD$ .HHHHULHLHEH HEH`HEHHEHEo EsHMoH|$(HD$ j.HHHHULLHEHHEH`HEHHEHEo EsnNHMoH|$(HD$ -HHHHUL LHEHHEH`HEHHEHEo EsnHMoH|$(HD$ ~-H?H$HHUL LHEH HEH HEHXHEH`HEH HEHHEHHEHHEH HEH HEH HEHHEHHEo Es #HMoH|$(HD$ ,HnHKHHUL LHEHHEH`HEHHEHpEo EsOHMoH|$(HD$ /,HHHHUL LHEH HEH` HEH`HEHHEHXHEH(HEH0HEHH HEHXHEHHEHHEHHEHHEHEEo Es4HMoH|$(HD$ K+H$HHHUL LHEH HEHHEH`HEH HEH0 HEH0HEHHEHHEH HEH8 HEH@HEH@ HEHHEH` HEH0HEHx HEHEo$Esp.HMoH|$(HD$ F*H'HHHUL LHEHHEH`HEH HEHpHEH0 HEH HEHHEHHEHHEHHEHHEH HEH0HEH HEHEo Es?&HMoH|$(HD$ W)H@HHHUL8LHEHXHEH0 HEH HEH HEHHEH HEHhHEHHEHHEH EoB Esԑ9HMoH|$(HD$ (HHEHHUL(LHEHXHEH0HEH@ HEHZIEo EsQHMoH|$(HD$ )(H"HHHULLHEHEoA Es2HMoH|$(HD$ LM'HHvL HZxLHUwEo EsrHMoH|$(HEwHD$ ~'HH$HHULLHEHXHEHwEoA EsrHMoH|$(HD$ 'H/HH(H5wLHULHEH Eo@<Es3HMoH|$(HEHt$ &HHthH(HULPLHEH Eo@<Es 3HMoH|$(HEHt$ i&HHtH/u HY3H/u HYH$H_[]H\$Hl$Ht$WH0HAIHHDPHpAAAAtAtH TZHeH [3MtIWHHCHu HSLEL pH ZHHD$ MH  W3MtIWHu[HCHu3zLEL H YHHD$ MH V3QMHHDMt7IIWHt)LEL H jYHcMH oV3HHH\$@Hl$HHt$PH0_H\$Hl$Ht$H|$ AVH0LIH YIZHHu 33Hx(LpHXHtHHppHD$`HtHHC HD$hH{@H{HHEHC`HkPH{XH{8HHD$pHtHHChH{xHHHHHANtdtXtJzt<t+H +XH<H XH+u HVH2H=H=?H=H=HH{0hYHH\$@Hl$HHt$PH|$XH0A^@SH LYMHE3L#Ћ$I[<u5Is)H WL OMHuH T3H [AEt&At3H [H HIMHH [HHIMHH [HH\$WH0LQHHY8L#M$E3Iz<u:Is.H VL MHH S3H\$@H0_AEtAt 3H\$@H0_H HIHIML\$ LHH\$@H0_@SH0MHL#LHIME3AHY$<u5Is)LL H VHH S3H0[AMt0I{t)LL aH UHH R3H0[EtAu;H IIHMt2M@L H UHuLT$ MH R3H0[3HH0[H@SH0MHL#LHIME3AHY$<u5Is)LL H UHH R3H0[AMt0I{t)LL aH THH Q3H0[EtAu@H HIIHIt2M@L H THpLT$ MH Q3H0[HHH0[H@SH HHHHu H [H@HHHA HHH)uR3H [H\$WH LQ`HyhHYpHQ`LAhLIpMtI*u IRHtH/u HnRHtH+u HZRH\$8H _H\$ UVWATAWH H.IIM3NMHHtHI,$HTLHuH\$hH A_A\_^]H Lt$XQLHHHLl$PQLHIHPHHSHLIITHHHtHtHHHHxH)uUQHtELIIwPHHt]HH;t-H.u H QMcD$HHH=y@Ht9(H(u3HP(HHMcD$H=yH+u HP3I.Ll$Pu IPI/Lt$Xu IPHH\$hH A_A\_^]H\$ WH HBIt,~+HHuhRHuHHHH\$HH _qSHHl$0MRH QHHHH`H;t8Ht!LBAt!uHl$03H\$HH _HM`3H}pHt$@HuhHE`HEhHEpHt H)uOHtH.u HOHt$@HtH/u HzOHHHl$0H\$HH _H\$WH HHH T*HHu(>QHuH JPHLH oMHH\$0H _H\$Hl$Ht$WH0HHEHAH;bPH;MPu@H¹H?EDtHHGHÃ|$`t H;GHDHhpHphHt?H}t8HLHHt|LEHHAH+Hu H7NHHH~EtFHyAHHt9HxHHHFHH OH Ou3yOHFHHeHùH?EDtHHOH˃|$`t.H;Or(HKHHu3HHMHGHHH\$@Hl$HHt$PH0_@SH HcHNMHHuH [3H|$8HLHHu"OHuH .NH H NH+u HLHH|$8H [H\$Hl$VWAVH0AHL3KHHua.H GHtDHLLD$ IXLHHuH YMH @Nt.FN33HuHLLljt$ ILHH/u HLHl$XHH\$PH0A^_^H\$Hl$Ht$ WAVAWH0LHK3LHHH&HHttH`H&HHt7H;5MNj׋H;5QLH;5L u HeNuH+u HBKHHtH.u H+KHu(MIH+u H KI.uIJMHtLIHHt,H:E3L|$ HJHH+u HJMt@Ht;HzJHHtH/u'HJLMHH%HHHH\$PHl$XHt$hH0A_A^_LD$HT$SUVWATAUAVAWH(IxAE3H|$pIHHL;"Mh H.HCHIULL;MuAE3LLMtALLHKH JHHHH`H;t&HtwLBAttVHM`HuhH}pL}`L}hL}pHt H)uDIHtH.u H0IHtH/u HIH|$pH+u HIIIIL;H$Hl$xHt!HH(A_A^A]A\_^][H$Hl$xJHtJL9fuH IH% LLH FzM3HHHHHtuH KHHt8HHJH VIHLLH FH/u H!HH+u HHMtI/u IG3@SUWH HAHHLMtAJHHH HH IIHFHHJHHt~H;HHt$HIHHtHHLt$PHILHtHGHI.u I!GH.Lt$Pu H GH/Ht$Hu HFHuH GHLH :EHH _][H\$Hl$Ht$ ATAVAWH@LIH: ILILHIHEHHHHHEH΅uHHGH|$`H8EIHEH\$(H LHl$ MH{DH|$`\HEHHtKI.u IEH.u HE3QHFIH&EMH LHDI.u IEHtH.u HEH\$hHl$pHt$xH@A_A^A\H\$Hl$Ht$WHPHIIIHHHHH|"H jFH{LLH lCLH(L@ MtH$3HHHHEL;LLKH;s)H FHMHD$(LLH|$ H  Cs$u7L;H EHHD$0LLD$(LH|$ H B1uRL;vMLD$0LH|$(L3Ht$ 3Cy&H+u H(D3H\$`Hl$hHt$pHP_HH\$Hl$Ht$ ATAVAWH@LIHzILGLHIHBHHHHH+CH΅uHHDH|$`H8CIHBH\$(HWLHl$ MHAH|$`\HBHHtKI.u I4CH.u H%C3QH"DIHfBMHLHXAI.u IBHtH.u HBH\$hHl$pHt$xH@A_A^A\H\$Ht$WH H8 HH HHHu.DHu#H CHLH @H= H=Hu-H0HH\$0Ht$8H _HPH sHHHu.CHu#H CHLH .@H= H=Hu-H"0HC,H\$0Ht$8H _H8(HHu-H/HH\$0Ht$8H _HHHu-H/H-H\$0Ht$8H _HXHOHu-HS/Ht]H\$0Ht$8H _HYHHu-H/H/H\$0Ht$8H _HHHu-H.HH\$0Ht$8H _HHHHu-H.H,H\$0Ht$8H _HH[Hu-H?.H`I0H\$0Ht$8H _HEHHu-H-HH\$0Ht$8H _HHHu-H-HH\$0Ht$8H _H\$03Ht$8H _H(LUH9=H|Hue+LH=HPHu1_LHS=H$Hu+HA?LL =HHuLHj<HHu L"H<HHubTH >LHR=H;Hu4bLHdV<HGHu.LD>H"<HHuLH;HHu$L>H;HHud.H=;HHu7eH>`;HyHu 8LN=Hb,;HMHuH=L=L:HHuWL <LH:HHud-H<L:HHu 4eL L'Hh<R:HHu -HC<H |L:HHu"fH)HH(3H(H\$Ht$H|$AVH H<HHHH`HHL5C DH t7A@t7=6CHPAtAE3O=&= SH'=HHHz=HHHH;uW8HH8HH8HHt~8HHtd3G8HHtP08HHt9E3H J3B8HHt7HHt3H 'H H\$0Ht$8H|$@H A^LH;}I;txHXHt/LR3M~!HJ@HI;tQI;tLHHI;|3HHI;t-HuH9L;tHI;t HuL;t3øH;tMLXMt#MA3M~IIH9t-HHI;|3HH;tHu3H;B9ø@VH t:HHH`HH-9H u H ^HN`3H\$8H^pH|$@H~hHFhHF`HFpHt H)u7HtH/u H7H|$@HtH+u H7H\$83H ^H\$Hl$Ht$H|$ ATAVAWH HiMMHHHu:HHIIAHH@H H;iuwLQL;SumQ DDC AAAD;uM t@0IDHHIHA tA@0IDHHSHEM23ɅuuHOHHoI;fDHHH;iLYL;[DA ADK A;ugA t$A@A0MDLH+HI?LLQHA tA@0IDHHKHDHMID13Ʌu.HI;P3H\$@Hl$HHt$PH|$XH A_A^A\H 7H)LD$`LH 4HL$SWATAUAWHPH3HPHH$HtHHuLcHl$HHt$@Lt$8J 8LAHHCH$I;VHK H$H H6H!5HH3Hb4LHu"6HuH 6HH 6H+u H4MH3MIHuqHH4HHu3D3H3HHu"c6HuH 5H8H O6H/u H(4IIDI;tHtpHHI;hH$H$AHIcH$H;$5I773Lt$8Ht$@Hl$HHPA_A]A\_[L$HL$IcMMLH 4M@MIH 1I6LL$ LD$HL$SUVWATAVAWHHHHE3HEE1HHL$HtHH+HHHHGHDHuH$LH1LHbH$$Ht|H 95LH03 %)MLH v'u%HH\$8H _H (HH &H+u H'HH\$8H _)HuH (HH ~)HH\$8H _H\$Hl$Ht$WH HAIHHHHu*SH &t3>LHHH&Hu"(HuH #(HH (HH\$0Hl$8Ht$@H _H\$Ht$WH HAH@ Hpt3HYH Y;&t3H\$0Ht$8H _HHH&Hu"a(HuH 'H.H M(Ht$8HH\$0H _H\$Hl$H|$AVH HIIILLH#u=HL}'HfHI}tZIC@tP3IHHu2L:'H#HI:tIC@t IIiICH;u0tH@8MMtE3LIIA+HuHLILLII%H\$0Hl$8H|$@H A^H(HALL;'uAE3H(H%'Ht$HMtA'HHHl$@&H %HHHH`H;t*HLBAttgHM`3H\$8H]pH|$ H}hHEhHE`HEpHt H)u$HtH/u H$H|$ HtH+u H#H\$8Hl$@HHt$HH(@UWHHHHHQLBpHBhMtM@Mt HHH_]IH\$hHt$pHHHx=H$3Lt$@H9GuwHOHtgHAHtLHHw4HH_GHH k_GHH HYH!HKGHy HcHEHIH.Hu H< HH x!HM@H |3H\$HHl$PHt$XH A_A^_E3H; n!AAH; H; L!AA uH%"@SH HHu H [3H|$8H;!@H;b H;  u Hv"H+u HUH|$8H [H\$Hl$Ht$ WH 3AHHH;H H9AH9FG uHH!F uH1!LWL;VHGHNH;tHt HO DDF AAAD;0AH t@ADHHOHA t A@IDHHVHAu DAu DDA;u|It{EMCuieLI;uGI;u;DHH)!HHu*HH+u H ËH\$8Hl$@Ht$HH _H\$HT$HL$UVWATAUAVAWH@ID;IAFًLHwwAAAEFwEg@ AH t@0ADIHT$0 IFHHD$0AIIL;H$3HH$HC u HH$AHH{HtsIH+H;|C  t@0IDHHSHA;uAHHLHD$0IHXE3H|$ LHIUH$HHAHH;@I(H H]H I.u I3H$H@A_A^A]A\_^]HT$HL$SUVWATAVHHL$II3Mt$Lt$0K HHHu3HHA^A\_^][LHt^Hr0H HQHGHHH;wH;sLHH;(H+J HI+ffHHHIHuIHHuH53HHA^A\_^][L$L$L|$@LD$ HT$(H\$(ILH MH(MtL9A t LH3H(H\$Ht$H|$ UAVAWHH@HA`E3HEIHAhLE HELHApHULypHLy`ILyhHMHE HL9{`HE Ht%HMH HE HtHHE HMHt HHMHE HUHtHHMHE HUIHHHCxHHHMHt H)uHM Ht H)uHtH+u H3MHMM>L>L?Ht H)uHMHt H)ulHM Ht H)uWH\$hHt$pH|$xH@A_A^]H\$Ht$WH IHLAHHH z<L mMQLHHtHH\$0Ht$8H _H FHHHu%HuH H]LH Ht$8HH\$0H _H\$Hl$Ht$H|$ AVH LqHIAH`HHt"HPHHuH$MHHtH #HH 7HH\$0Hl$8Ht$@H|$HH A^H\$Hl$Ht$ WH0HIHHuWHH0Ht:HLLHD$ HHHuH(uqH\$HHl$PHt$XH0_@SH H 5HHL HLذHQTL HLH</L HLH' _L HLH:L vHLdHL QHLGHL ,HL*HvL HL HQL HLH,H+u H H & HHTL HLH4L HLHL HLqH*L &HLTH]pL YHL7HpKL |HLH&xL HLHx^L HLHx=L iHLϮHxH+u Hx 3H [HtH+u H\ H [H\$Hl$VWAVH@HH L HHL5?HIOHHHH|&H / LMH6H - HH H@(HtH;HLHHs1H LHT$(MH HD$ H  8HvKHL$0L |HD$(L3Lt$ 3 yH.u H* HG8Hw8H+u H H # HHD$(LA Hl$ HH`HG@HH+u H H ź HHSD$(LĺA Hl$ HHHGHH"H+u HN H 1 HHH~Hm HHHH|'H M L NL?HPH GPHH H@(HtH;HLHH sOHL$(L H  LH/HD$ H H.u H HGP>HʹHwPH HHHH|'H L LHH SHH H@(HtH;HLHHH sOHL$(L QH B L3HlHD$ H H 2H.u HHGX{HHwXH HHHH|'H L LHH SHH H@(HtH;HLHH0sOHL$(L H  LpHHD$ 0H oH.u HHG`HdHw`H' HHHH|'H  L 0LH H PHH H@(HtH;HLHH`sOHL$(L H LHHD$ `H H.u H@HGhD$(LAHl$ HZHwhHHGpHD$(LrAHl$ H HhHGxHD$(LDAHl$ HH2HHOD$(LAHl$ HHHHD$(LAHl$ HxHHHD$(LAHl$ H?HHHD$(LAHl$ HHNHHkD$(LgAHl$ H͵HHH2D$(L>AHl$ HHܿHHD$(LAHl$ H[H裿HHHH3HHHH|'H L LHH  SHH H@(HtH;HLHHsRHL$(L {H LHHD$ H H.u HIHLJHH+u H#H $HHD$(LsA`Hl$ HHyHHD$(LA@Hl$ HH@HHtaHLH-HtMD$(LAHl$ HoHHHtH+u HD3HtH+u H,H\$`Hl$pH@A^_^Ht$WH HH\$0HH=HHHkHHeHH_HHDHPuHHuHH HtPHQH;~BL H DLAGMu M9 HHH;| ցHϋs%tHH uHH9uHH fE3H3HHHLHH HH Hu HjHLHH k3ɅAH\$0Ht$8H _H bH[MHL8H _+H >HM@H BHu HH\$0Ht$8H _H\$E3DZHEExIcHD;D~H\$E~-AA+yF IcHD;D}E~$EQE;|IcAIHD;DANɋH\$H\$AH\$Hl$VWAVH0H3LHHIH)uI~HHGH/LMtAHHu ADH.HH2L5HT$(E3H\$ IIIHHD$(UI.Hu I[Hu AE3E33HH/u H,AHHHHuHtA}EIF IMHE @ IFHIFHPHHuA?H(uHGHHLMtAHHu"AL H 3܄Hu&3IH)uJIHHH.u H*Hl$`HH\$XH0A^_^3APHAXHH@UVAVH0HrIMHH\$PHH|$XLd$`Ld$pLl$(Li HL|$ L|$xIfHIuHMIJ< HMHHMIHHH+H}L|$ Ll$(Ld$`H|$XH\$PHHH0A^^]H\$WAVAWH`HH3HD$PLE3H VHAH9ÝuHHtHH=~@H=uHLGHHHHH {H|HtHBH HHHu%wHuH HLH H$Hu sHHCHHSLMtADHHHH+u HHH9GuLHwIHH)u H3HL|$(HLt$0H?HT$(HAL+E3L HrHMtI/u IsH.u HdH 3H;@3H;%3H; u H7y$H+sH5)HH+u HtAIjHHuHt t,HuIHHD$@3HD$HLt$8;HHu)HvHMI.u{ITpH HT$8HLHCIHHL$HH ׎HMtI.u IH+u HHu$vH5RLH D3F3HH$HL$PH3*H$H`A_A^_HUVWATAUH`3HpDHp HpHJH$HHxfDH9HtH;u HIHuLt$PL|$HHtH;tHHLwIL HLLH H\$XLHu7H H H LHHII,$Hu IHH AH9Kt,H H͚H H+HY{3HH+HyuH6HgHuHH TH +5=v~H9HH/AHDH<L%UHH$HHI`H;HHB'H}H H\DAH qHu Ht HMtI.u IHtH/u HMtI/u IMH ۛD39}H$L$L$H$ytHHT$0E3Ht$(IHHH6HD$0DH+Hu HJHt#E3E33HHmu H%HL%H$L$H$HAxHH8Ht H)uMtI.u IMtI/u IHtH.u HH$Ht H)uMtImu IyHtHmu HdMH jD3{L|$HLt$PH\$XH`A]A\_^]LUWAUIHHwH3HHE3II[IsMcEHMH@MsEHMM{LMHT$HLD$`LmLl$hLmLmLmLl$pLl$xLmLl$@Ll$PLl$8Ll$XXLHuH=oHIHIOHߒHH=%/HHt-HMLDl$ HHH+u H Hu oT!I/u IHuHHEHHu o"HHMHHuH.u HH IHAH9f|uL5e|MtIH)@H HLCLHHHH |L5|MtIB7H H8LHu%"HuH .HLH SMuH=WrIFIHoLMtAHLHuH=rGI.u IH6H9Cu H{LcHI$H) u H^IHt$HHUHH}HHuH?HAE3L+IL HHtH/u HI,$u IHuH=brH.HH\$HuH 5~IMHAH9zuL=zMtIHW@HNHLCLH}HHH \zL=]zMtIBeH }HfLHu%PHuH \HݪLH MuH=s<IGIHLMtAHLHu s*I/u IHkH9Cu LcLsI$IH) u HIHt$`HUHLeHHu H?HAE3L+IL @HELMtI)<$u I=I)>Mu I,MMuH=sDH)>HL|$`uH]H;u%3HuH=OuHD$hgHKAՋrH|Ѕt3HqHHuH=wHH@HL$hHHH\$hHL$HHHALMtAxHHuH={XH6HHu {)H)>u HH;H=5HUE3LmIHHHSHEH/Hu HHuH=|E3E33H2H+u H{H=|IGIHLMtA}HHuH=}]H;HHu }.H.u HHAIGIH{LMtAHHuH=-}HFH;@uHFL I$H;u LfI$yHxpH@hHt$L9ot3HHt[HGHH6HtH@Ht 3HL.3HHtHH<H+Lu H Mu }H.u HIGIHvLMtAHHuH=(}Ll$8hHFH;6uHFLpIH;u Lv IHxpH@hHt+L9ot%HHHHGHHuDHLH.u HH|$HMHHLeHGLMt A!HoLHuLHuH=( ID$H;:uID$H0HH;u It$HHxpH@hHt$L9ot3HHteHGHI;HtH@Ht 3IH33HHuIHI.H+Hu HH|$HHuH=[Ll$8 I,$u IMGAIG MHL;AM9G tHIGJ4I@IGH.u HH]HUE3H]IL}IHH LHtH+u H>Mu,H=Y HIu+IMHGHHLMtALHuH=CLl$8o IGH;QuIGH8H|$@HH;,uIH|$@HHxpH@hHt3L9ot-3HHuLl$@HGHIEHtH@Ht3IHHD$@=3HHu ILl$@#HI$H+HHD$@u HHu 7I/u IHHH E3IHLHRI.HHEu I|H/Mu HjI.u I[HuH=v H}H]LHH oHAH9luL=lL|$@MtI^H }HH}HLC LHD$@HIoHHH >lL=?lMtIH_HD$@LMuH=DIGIHxLMtALHuH=I/u ILH vHUE3Ll$@IH}LeH/Hu HI,$u IHuH=cH/Ht$`u HHUIHHHlH9BuHZHzHHH) u HIHH](HHu0H?HU(HAL+E3L HNHHtH+u HOH/Mu H=MHuH=UHNAH; H; H/LHu H.u HIGIIIHLHIϿLHtIϿHD$@HtkIHt6H(u HuH HAH  y  I/I(I/u I跧uH;H=lHNHt=~'H /HAH )HxLvItsHILf I$HF(HD$@HH.u HwIHL$@AH]EH wHL$xHLd$pMLuydH=^HLHu 6HFL`I$Mu HFLxL|$@IM^ He~EHzyH='HzEHMyH=H=HUE3LmIHHHsHE0H/Hu H6HuH=QE3E33HH+u HH=j!H QjHAH9vguH5ugHtHTH |xCHsxHLCHH jHHH 1gH52gHtHH"HHuH=HFHHqLMtAtHLHuH=H.u HH riILl$@HAH9fuL fLL$PMtI^H wHHwHLCLHD$PHiHHH PfL QfMtIH1HD$PLMuH=IAIHrLMtA~HD$8HHuH=Ll$8HL$PH)uHL$xHoHALMtA#HD$PLHtHT$pH HD$XHtHL$PH)uH^Ll$PH9Cu$HCHsHD$8HHH) u H}IHD$xMHT$XL+H\$8E3HHHHH?HHHL HHD$@LHtH.u HHL$XH)uH+Ll$Xu HLl$8MuH=KM|HzH9Gu LgLwI$IH)u HIHt$`WLeXL}`HuhEp2HD$8HHH=LMtI.u IFMtI,$u I1Ht$@HtH.u HLl$8HL$PHt H)uMtImu IHL$XHt H)uE3HD$`H CL|$xLLeDLu3HD$`HD$HHD$HHEHEHD$pHD$pHEHEHD$hHD$hHEHE_HMH\$HH|$`Ht H)uOHL$hH$Ht H)u1MtI.u IL$MtI,$u IHML$Ht H)uHL$pHt H)uMtI/u IHML$Ht H)uHtH+u HH$HtH/u HlIHH3HA]_]HwuHUXHHHHFAL+LHHlHMpHF HHHHMxH?IL HMtI,$u II/Mu IH.Ll$@u HI.u IHuH=HH]I菷yH=THvDHMyH=ucH cHAH9auH5aHtHTH vCHvHLC,HHZcHHH `H5`HtHHpHHuH=)HFHH9vLMtALHu H.u HiH*dE3IHHuH=Ll$8I.u I*H+u HIH ybE3Ll$PL|$8HAH9_L5_MIH=xHU@E3Lm8IHHHkHE@葯H/Hu HHuH=E3E33HH+u HbH=H pCHoHLCdLHaHHH _L5_MtIH訅LMuH=aIFIHkLMtAHEHD$@HuH=MKI.u IH `MMHAH9q^uL p^LL$XMtI^H oHH oHLCtLHD$XH`HHH "^L #^MtIH賄HD$XLMuH=gHt$@IAIHqLMtAHHHuH=!HL$XH)uH?L|$XH9Cu LcH{I$HH) u HbIHD$pHUHHEPHHLeHHAL+H?L E3HLMtI,$u IH/Mu HMH`ILHuH=H@I.u IHT$xILHu6I/H= u IHD$PHD$8LHI/u IgHMH H9Au%HAHqHD$8HHH)u73ۋ Ht$@3ۺHL$8HL}HH?AL+HL LHLHE3HëHHD$8HtH(u HI.u IH.Lu HH\$@HuH=pI/H}u IpHT$HHHHuH=ÿ6H/H]u H6HMHHuH=HCLHnHLMtAy.H=KH.Ll$8rHLl$8_H.u HHLLuLH\$HL|$xH|$`HM*H HGM@H H=Ӿ@USVWATAUAVAWHl$H8HSH3HE(HbYE3HMM LMLD$xEH(AHhEUALl$hLl$pLl$XLl$@Ll$PLl$`HLl$(AD$ E3ILHu rHMLd$8I$Hu IAVMuHqHUE3LmIHHHtkHE3H+Hu H9Hu E3E33HµH/u H IH/XIN LAHjHHEAIFIHEpu HAHtHMHSpf/QAZHL$xHG[HI趕H ZHAH9XuL%XMt I$HhAHhHLCPLH~ZHHH sXL%tXMtI$BH \ZHLHu%HuH HsLH Mu uID$IHeLMtAHHHu :I,$u I]HL$xHcHALMtAnLHu  HH9Cu"L{HsIHH+u HIHYHUHE HHL}HLeH?AL+E3L H蜦HMtI/u II,$Ll$@u IH.Mu HvIHu 1 Hd؅y  H/u H=nH\$xHibHHCLMtACHHu  HzAH;HHu  H/u HHI迫Dy s H.u HEIHHUH iE3MH]Lm"HHtH+u HKHu  HEkE3HHHu  H.u HHI؅y  H/u HtH[iHHQiIH6iHH,iHD$pHlHUE3LmMHEHHkH+HIu HnHu ', E3E33HH/u H@HL$pIHL$p' HEH;uIHHHu ) LHu ) HxMHD$XH HEH@xfDH0HuHtH;u H@HuHt#H;tHHHFHEH]HD$xILmLmLl$xH UHAH9Su%HSHtHHwdHD$HHJHfdHHD$HL@DHHrUHHH SHSHtHFHT$HH LUHHu'HuH HfLD$HH HHCHHQ`LMtAHD$@HLHH+u HWIHIݺH9Au!HyLyHIH)u&IL|$@HEAHEL+HH}HE3H?IL HEHѡLHtH/u HHL$@IH)uMMt)HM<LHtQHHFHHD$@MtI,$u IlLl$HHtH/u HSLmMtI/u I;LmHtH+u H#LL$@H u3A,IML}LL$HILEHU6y-xHMHhH9AuHH LHHtLHMBHD$PHHt5RHD$`HHtHHxHHD$PHHut.L=ԴHUHMHRxLH MtI(u I>HMHt H)u)HL$xHt H)uHELd$HH}HD$@H.u HH+u HHMH|$XHt H)uHMHt H)uHL$HHt H)uMIGxHMHHHtH*u HtHMHt H)u_HL$xHtgH)uaIYHML`MLx Ht H)u'HtH.u HHL$xH|$XHtH)H|$Xu H|$XH OQIH\$@HAH9lOuH=kOHtHH5o_@H5f_HLFHHPHHH $OH=%OHtHB}H PH~HHu%hHuH tH}LH Hu/Ll$PLl$`HGHHcLMtA3HD$PHHu/Ll$`H/u HH 1PHAH9fNuH=eNHtHL=Y^@L=P^IMGHHOHHH NH=NHtHBgH OIhHHu%RHuH ^H|MH Hu/Ll$`HGHHI\LMtA"HD$`Hu /H/u HHpAH9Fu)H^HFH\$@HD$PHHL)>u HMHD$XHE3HEH]%HHu /%HXHUHt$`JHGAM+LHIL|$PHH?IL HuLHtH+u HH.Ll$@u HH/Ll$`u HI/Iu ILl$PMu /zI$IHL$hMHAQHI HEHJLAfnEAHw[HZcu 4HMHtHHHHELmHMȻHWDH葰HEHHHEDHkWMHdHHHHAAH9Cu(L{H{IHH) uHAIHHL}L+LmHHUH?HL E3H\LMtI/u I]H+Ll$Pu HILl$HIMu2L=8H.u H!HD$hIHD$hLl$PI,$u IMLmL}HEM~eHELeH]H}Lh`Hp JLDULLl$(LHLd$ 赟IIuH|$HE3Lt$8LeHuHEHBHME3HH.Hu HNHHD$h8HD$hIL=HD$XHD$XHD$pHD$pMtI,$u IHtH/u HHL$@Ht H)uپHtH.u HžHL$PHt H)uHL$`Ht H)uMH GmD3DH|$hL|$pHHt$XMt8'H+u H]H|$hHt$XLL|$pHHI.u I6HtH+u H"MtI/u IHtH.u HHtH/u HIHM(H3OH8A_A^A]A\_^[]LUVWIkH`Hh@H3HE3I[ McMkDMsDHMDM{HLLEH|$8H}H}H|$`H|$PeHEHu bHhF HhH(HH|$(D$ DDIHD$@HHu  HHHu HмH 1ILH\$8H|$@HAH9GuL%GMt I$HKWAHBWHLCLHHHHH 8GL%9GMtI$BXH HHYLHu%CHuH OHuLH tMu +ID$IHvPLMtAHELHu I,$u IH HHLH\$`HAH9|FuQL5{FMtIH7VH GHHD$H胖LHiHLD$HnHUHHD$hL@`LHGHHH EL5EMtIF HT$hH hG LHu'HuLD$hH H}tH $Mu IFIHSLMtAȽHD$hHHD$PHu I.uIeHL$hH LH9Au!HAHYHD$PHHH)u-HH;ZAHL$PL+HEE3HH]HL}H?L HEHԓLHtH+u HչHL$PH|$`H)uH|$PMu  HMHKH9AuHYLiHIEH)u{HH\$@HH]HLeH?AL+E3L IHEH(HHtH+u H)I,$H|$@u IImLu ILHu  H؅y  H.u HȸtH[HUE3H}IHHHMHEzH+Hu HHu  E3E33H H.u HRHa IGHEHEH;uIHMHHu 0 LHu  HpLHE HHEHHxffL9L|$HMtL;u HIHuMt%L;t IIMoLl$hIEjHD$xLH|$HH|$hH|$xH CHAH9OBuHNBHtHH5R@H5yRHLF\HHCHHH BHBHtHBH iCH HHu%HuH HpLH $Hu Ht$@HCHHgNLMtAȹHD$XHHHH+u HmHHHߺH9Au!LqHqIHH)u<HHt$XHEAHEL+HLuHE3H?HL HEHLMtI.u IHL$XLH)uԵHMt)HMRHHtgLHXH HD$pHtH.u HH|$@MtI.u IjH|$XMtI,$u IPH}HtH+u H8LL$pH d3H|$`AH|$PLL;H]LMHLD$@HT$X=yxHMHoH9AuHL SLMtLHMIHD$PLHt5YHD$`HHtHL`IĴHD$PLHu|L=ۢHEHL$HH@xHHHtH*u HDHL$hHt H)u.HL$xHt H)uHD$8HL$@Ht$XLeHD$8-Imu IH.u HܳHL$XLeHt H)u³HL$@Ht H)uHMHt H)uLHCxHL8Ht H)uyHL$hHt H)ucHL$xHtdH)u^MVL`LHp MtImu I+MtI/u IHL$xLuHtH)Luu LuL|$8H Q?LH|$@HAH9=uH5=HtHHqM@HhMHLCӱHH?HHH =H5=HtHBH >H耍HHu%jHuH vHkLH HuH|$`H|$P;HFHHQLMtA5HD$HLHD$PHuH|$`H.u HͱH .>HAH9<uH5<HtHHVL@HMLHLCHH=HHH <H5<HtHBdH =HeHHu%OHuH [HjLH HuH|$`%HFHHGLMtAHD$`HHu H.u HHL$HHHbH}H9AuLqLiIIEH)uHHEAHEL+HLuHH]H?E3L IHEH5LMtI.u I6H+H|$@u H"ImH|$`u I H|$PMu I$IHMLH|$xHAQHI HD$HH9LALuAHHEHEHt H)uHtH.u HMtI,$u IMtImu IwHL$`Ht H)uaMtI.u IMHL$PHt H)u7MH ZD31L|$8HuLuHMt9(H}H+u HHuLuHHHI/u IժL$@L$PL$XHtH+u HH$MtI.u IL$HHtH.u HqHHMH3ڗH`_^]H\$UVWATAUAVAWHl$H0H,H3HE(E3HMHLeLd$0EELd$HALd$PEHT$XHHEHuH5F LNH6L;tbIXHt.L@IMH@H9t9HHI;|IMtfHH;tHuH;8bIHH @HtrHD$0HHuH5 HCHHHLMtA/LHuH5X H+u HϨILd$0ʈ؅yH5% Imu IMHJHT$xE3Ld$pIHHH =HD$xCH+Lu HIMuH5 E3E33IˎImu IH5|M H4HFAH;ALHHBMtALHuH5  HGE3EI(}؅yH5 T M)eu IjE3HFHHELMtArLHuH5  I!؅yH5|  M)eu IH3E3t-H;5tYHuMH59  H3LVImIHH[DH% $ HFHHDHD$pHFLMtALHuH5 IEH;ɧuIEL0IH;u MuI}HppHHhHt$H9~t3kHHt_HFHI6HtHAHt 3IL-33HHtHIǥL)#Lu HHt$XMuH5VM)eu Il3IDyHEHu_HtH5M)&u I+HFHH];LMtA>LHuH5gIFIH/=LMtALHuH5)M)&u IILyHEHHuHtH5HM)eu I^H 0LHAH9/u"H/HD$0HtHH>EH>HLCAHHD$0Hj0HHH W/HX/HtHDH I0H~HD$0Hu%ѥHuH ݤH^]LH H\$0HuH5bHCHH9LMtAHHHuH5!HL$0L)!u6HHD$0HuH5H s/HAH9p.u"Ho.HD$HHtHH5=EH5=HLFHHD$HH!/HHH .H.HtHDH /H}HD$HHu%HuH H\LH LT$HMuH5IBIH:LMtAQHD$PLHuH5uHL$HL)!uHIH9CuLsH{IHL)#u H3H\$03IH]HELuYHD$HLHuH5^H7HUIBHAML+HHHHL}H?L LMtM)&u I13L)#Du HHL$PHt$0L)!u HL$HLHt$PL)!uL)'Ht$Hu HޠHMuH5?ILH;H]t$H{-HˌuH5iHCEH]H7HD$@HHD$hHHuH5ǎ(H6EH诏HD$HHLHuE3HȡIH9Cu/H{LSLLT$HHIH) uHLT$HIILH}ILmL+I?HUM HE3IyLMtI/u IE3H+ALd$Xu HIHD$HHEMu+L|$PHL$hH5͍H)1M&Imu I8ZH HHD$hHHxL1MtL;u HIHuMtL;tIIM~IŠLMMHD$@LELMHUHMHD$(HD$pHD$ IuàHHA誠HHAL H M33HD$`HD$@$LL$XHLEHT$`葋Ll$`HEH|$XHD$HLLIչHD$PHE3HHvH.Hu HHD$PH(u H؝3HL$PHtHH}H+u Hx+ ӟLD$HLHIJ3DHD$HH5HD$hH@xHL0Ht H)uXMtI/u IDMtI,$u I/Lt$@L|$PE3HT$0HtH*u H MtImu IMtI.u IHtH/u H̜HL$HHt H)uMtI/u ILH KD3#MtImu ItHL$HHt H)u^HtH/u HJHD$hH@xHL0Ht H)u*MtI/u IMI,$ IH(uE3MtI/u IכMtI.u IÛMtI,$u IHW)E3HDtH.Hu HH|H5E3L|$PECHHuH5!HW;AH;t`HGH;u 3H9_3H;uOWf.zt/3.EHKH3{؅yH5NAL)'u HÚtHHHFHH4LMtAÝHHuH5#HH:E3EHo؅xL)'u HI3WHFHHq0LMtARHHtH@HH8LMtA%HD$HLHHVL)'u H˙H ,&HAH9a%uH=`%HtHHT4@HK4HLCHH%HHH %H=%HtHBbH %HctHHu%MHuH YHRLH ~HHGHH4LMtA(LHaL)'u H֘HL$H3L)!uĘM)eH|$Hu II;H %HAH9M$uL-L$Mt IEH8AH8HLCLH$HHH $L-$MtIEB<H $H=sLHu%'HuH 3HQLH XMuH5\$IEIHl7LMtAHD$HLHuH5$uM)eu IzLHuH5,GL7IHP5ZyH5,H\$HMH%HoHHuH5$L)#u HM)eu IL)'u H3H I#E3HAH9"u"H"HD$HHtHHi1EH`1HLC˕HHD$HH"HHH I"HJ"HtHDrH "HsqHD$HHu%[HuH gHOLH LT$HMuH5.IBIH[,LMtA$HD$PHHHuH5E.HL$HL)!uH#HL$XqHD$HLHuH5.bH5IH9Cu(LkHsIEHH) uH_LT$H3ۋ3LLmILUL+I?HUM HE3HoHMtM)eu IHL$HLL)!uL)&H\$Hu HLHuH5K.HR+EH}HH]H(HD$@HuH5/pH*EHHD$HHLHuE3/HL$@H3I̋H9Cu/L{LSILT$HIIH) uH%LT$HHILL}IH}L+I?HUM HE3ImHHtH.u HؓE3H+ELd$Pu HIHD$HHEHu/HL$@:H/u HIH_HD$pHHxHH\$hHtH;u HIHuHt"H;tHHHCHEHHE Ld$hLeLeLeIIHMHt$XH HMHFHD$`HHEI=LL;HNH; nuH H\$`HLGHHALLd$XD3E{HL$pLL$PLEHT$X3|H|$XHEL|$PHD$HMLH׹LHHt$@E3HHEgH.Hu HImu I{MHtGHsnH+u HZx*zLD$HMHH;ILd$HMH5|/HD$pHT$hH@xHHHt H)uHMHt H)uHMHw356HtH/u HHL$HHt H)uMtI/u IHD$pHT$hH@xHHHt H)u[HMHt H)uFHMHC H)9 . HMHt H)uHL$hHt H)uHMHt H)uHt$@E3HH{eH.Hu HH H5!{E3E/H HAH9u"H~HD$PHtHHEHHLCHHD$PHHHH -H.HtHD.H H/gHD$PHu%HuH #HELH HL|$PMuH5Gz9IHa؃uH5#z9M)'u IE3H HAH9~uH=}HtHH+@H*HLC{HHHHH 6H=7HtHB'H H(fHHu%HuH HDLH CHuH5Gy;HGHHW*LMtAHLHuH5y;gL)'u H~HL$XH$HIHALMtAHHuH5x<HGH;|HHH|$0L)'u HH)LD$0HEIH$HELE1M@HMIDȺRlHHwH5'x<H H;u HAXH'H H;u HAXHHHHD$0HHH=H5w<E3*L)#u H9HIH9Fu$HFLnHD$HHIEH)u H3Hv+IHT$HHE3HE HUH}HD$0HHuH5:w;H (HUH&HHFAL+LHHHHM H?IL -HL$HLHt L)!ue3L)'H\$Hu HPL)&Hu H?M)eH\$0u I+LMuH5v;M)'u ILH]EH{H_wHEHuH5GvCH6EH/wHD$0HHuE3CHM5HB3I̋H9Cu*LkL{IIEIH) u He3HIL|$0LHUILmL+I?HUM HE3IaLHtH.u HE3H+ELd$`u HLd$0MuCHM~I/u IӆMHHD$pHHxfHH\$hHtH;u HIHuHt"H;tHHHCHEH\HE Ld$hLeLeLeIIHEH\$XH HD$@IH#1HKLH; ŇuHKLLFHHL$XA?HtH@Ht IHL;I HHuE3 HL$XHH+Lu H{H\$XMHCH;uHCJ4Ht$0HH;ԆuJtHt$0HHppH@hHt,L9nt&IHHnLFHHL$XAHHtH@HtIHHHD$0DIFHHu 3Ht$0%HL$XHτH+HHD$0u HH\$XHHCH;u$IHCJ NaIHlI9Guq._H$HHI/u I^_H`HHH9Cu*.Z_LHu8TH`HH*H9Cu-'_H$HHu8]H7H5@MH/u H^MtI.u I^MtI/u I^H$Ht H)u^H$Ht H)ug^L$H HLD3Mt;*LpLL$Hx Hp(Lx0HX8 HLLI.u I ^L|$8Lt$@Ll$HH|$PHt$XH\$`HtHmu H]IHhA\]HHHUAUHHX3HpHHRHHxDL`DLpDDLxD)pAHjH]y pt+HH]y pt HL H=gHT$8E3Hl$0IHHHKHD$86H/Hu H\Hu qE3E33HhCH+u H\qjAAAAACȅuH=HT$HE3Hl$@IHHHHD$H96H/Hu H?\Hu uE3E33HBH+u H\uH;]uH9kvwHCH8HH;r]uH9kvYH{HHzpHBhHt%H9ot3+ZHHtbLGHHALMtA/[HHu H~E3H{0LHu H/u HWH3YI9FuAv IDV(f.5YzuYHtH5EI.u IZWvXIID$HvLMtAgZLHu HE3I/HHuH5lEcI.u IVHL*؃uSXHtHLH/H5&Eu HVMtI/u IVMt\I.uVIVKH/u HrVHNMĉ^PHHALMtAvYy"H5DLH LD3 H-VHE(t$`L|$pLt$xH$H$H$MtI,$u IUL$MtImu IUHHĘA]]H\$UVWATAUAVAWHl$HHMH3HEE3HUH;PVAHMELuALuLHHIEHHE;XHELHH@LMuILEIIALHEMHZLmHMIEHHEWHEHH(H@LMuH LEIAHHMHLHH/VLuH9Cu(LcLkLeI$IEH) u HNTIIHLeHLuH?HUHAL+E3L I-HMHHt H)uSH+MLuu HSLuLuHXH/u HSUHTHEHHxL)MtL;u HIHuMtL;tIEIM}I`UH MMIHMH H]i HHEQILuHEH!L AH j3AHMLMLEHUX@H}Le߅y LELEHEMLHEH׹QHHHME3Hh+HEHEH(u HRH.u HRHEIHtOH2ЉEHEH(u HnRUx(TLEMHHILuM滤H@HEHEH@xHL(Ht H)uRMtI/u IQHMHt H)uQLmL}HtH/u HQHMHt H)uQMtI,$u IQMHEH@xHL(Ht H)uzQMtI/u IfQHtH+u HRQH MHAH9H=HyHHEH@xHL(Ht H)uQMtI/u IPHtH+u HPH]E3HzHj)H+Lu HPMtkI/u IPHEPH QHH RH>HEI/I]PLmL}"H LQHH `RL->MHtH/u HPMtI,$u IPMtI/u IOHtH.u HOHMHt H)uOMH D3/H]TH_@HVHLCNHHHHH H=HtHBmQH Hn*HHu%XQHuH dPHLH MHuL-=IHGHHRLMtA#RHHHuL-I=H/u HNH !HAH9&uH=%HtHL-I@L-@IMEMHHHHH H=HtHBWPH IX)HHu%BPHuH NOHMH sLHuL-w<HGHHLMtAQHEHHuL-5<H/u HMHUOAH9CuL{HsIHL)+u H}MMHEHE3HEL}LHHuL-;H}HUH]JHGAM+LHHIH]HH?L MLMtI/u ILH+Mu HLH/Luu HLH.Iu HLIMuL-;lI$IH]I$Hu ILHCMSHK HEHLAHELuH@HEL-IHHsH.OLHBH@LMuILHIALM(HEML-OIHHsHNHHH@LMuHLHHAHHHH5MMH9Gu&LgHMI$HH) u HWKIHHLeHLuH?HUHAL+E3L H%HMtImu IKH+Mu HJIHH.u HJIJHMHH HMLmJH}JHVE3IC#I/Hu IJHthH+u HuJILmL;mH]HLHtH+u HFJIHMH37H$HA_A^A]A\_^]ûH KIH LIL-e8I/IIH JIH KI}H\$Ht$ WHPHA3H;JHuHHAH JH;u HAXH'H JH;u HAXHHHIHHl$hHu H-71H qHT$(E3Ht$ IH|$("HHHH$IHHH/u HHHHD$8AC H\$0Ht$@@tAu DʋF @uuA ADDLCD HNHLHL$0*HHuYH-6H+u H/HHtH/u HHHtH.u HHLH 3Ah3!H+u HGH.u HGHHl$hH\$pHt$xHP_HLH L@HPUVHhHHIHXHxHL`LLpE3L;FHLxELuu:HII(HUouCGHTHHHGH/HEguGHME3H; yHHL$IMtHHULuHMHHUHHELuH}H H+Hu HFHuL55HLIL}MIIGLMtAIHHuL54 E3v3H;G@3H;G3H;G u H#Iy E3H+u HEH JHAH9OuHNHtHH=Z@H=QHLGDHHHHH HHtHBGH H HHu%kGHuH wFHLH CHu oHCHHLMtA@HHHu E3 H+u HDHE3HzHHu E3H/u HDH+u HDIHUH E3ML}HE9EHMtI/u IbDHu I.ILH]u=DH HAH9uHHtHH=@H=HLG(CHHVHHH {H|HtHBEH 5HHHu%EHuH DHLLH AHu HCHHKLMtAFHHu E3`H+u H8CAHEI{FHHu E3fH/u HBE3HA"y E3H+u HBtSI$ID$ HuwAHUoEHMgLHH\$0HD$(Dt$ nHH[ 9 H HAH9.uH=-HtHH@HHLCcAHHHHH H=HtHBDH pHHHu%CHuH CHLH +@Hu HGHHLMtADHHu E3H/u HsAEHIDHHu E3hH+u H@AH@!؅y E3}H/u HAtoI$ID$ HuwAHUoEHMgLHH|$0HD$(Dt$ HHuL56/E3 L)'u H@H H HAH9iuHhHtHH=,@H=#HLG?HHHHH !H"HtHB:BH H;HHu%%BHuH 1AHLH V>Hu )HCHHLMtABHHu E3 H+u H?EHIBHHu E3 H/u Hk?HIhy E3f H+u H>?tSI$ID$ HuwAHUoEHMgLHH\$0HD$(Dt$ HH  H HHAH9uH=HtHHp@HgHLC=HHHHH uH=vHtHB~@H HHHu%i@HuH u?HLH <Hu m HGHHLMtA>AHHu E3I H/u H=EHI+AHHu E3 H+u H=H؅y E3 H/u H=t]I$ID$ HuwAHUoEHMgLHH|$0HD$(Dt$ 4HHL5+ E3S H HAH9 uH HtHH=@H=HLG<HH=HHH HHtHB>H HHHu%>HuH =H3LH :Hu ! HCHHrLMtA{?HHu !E3G H+u H<EHIh?HHu !E3S H/u H;HIy !E3H+u H;tSI$ID$ HuwAHUoEHMgLHH\$0HD$(Dt$ .HHK")H HAH9^uH=]HtHTH CHHLCV:HHHHH H=HtH=HHHu #" HGHHLMtA=HHu #E3H/u H:EHI=HHu #E3H+u Hd:Hd؅y #E3H/u H::t]I$ID$ HuwAHUoEHMgLHH|$0HD$(Dt$ HH4L5V($E3H :HAH9uHHtHTH eCH=\HLG8HHHHH HHtHs;H HHu %HCHHKLMtAd<HHu %E30H+u H9EHIQ<HHu %E3<H/u H8HIy %E3H+u H8tSI$ID$ HuwAHUoEHMgLHH\$0HD$(Dt$ 'HH4&H HAH9guH=fHtHTH CHHLC?7HHmHHH "H=#HtH9HHHu ' HGHHLMtA:HHu 'E3H/u H7EHI:HHu 'E3uH+u HM7HM؅y 'E3H/u H#7t]I$ID$ HuwAHUoEHMgLHH|$0HD$(Dt$ HHL5?%(E3H #HAH9uHHtHTH NCH=EHLG5HHHHH HHtH\8HHHu )|HCHHDLMtAM9HHu )E3H+u H5EHI:9HHu )E3%H/u H5HIy )E3H+u H5I$ID$ HuwAHUoEHMgLHH\$0HD$(Dt$ HHu*L5#E3H[GL)#u H5H}H;55AuHUH;n6t H;Z6A3H EHAH9uL=MtIHm@HdHLC3LHHHH ҼL=ӼMtIB{6H H|LHu%f6HuH r5HLH 2Mu /jIGIHLMtA;7HHHuL5a"/^I/u I3H5E3IH9CuLsHsIHH+u H33HLuHH}H?HUHML+E3L H` HMtI.u Ia3H.u HR3Hu /HGHHLMtAV6HHtiH/u H3HEHL6HHu"/L5P!H+u H2Ht:)H+u H2H؅yyLEлLEyH$HU`E3H}XIHHHҸHE`H+Lu H$Mt#E3E33I2 I,$u Iz$E3HMлHMHML5Ll$`HL$xHFxHUHHHt H)u6$HM8Ht H)u!$HMHHt H)u $Ht$PH|$`FHM8LuHt H)u#HtH+u H#HMHHt H)u#HAI&HD$xHu IFH؅y 1FHL$xH)uh#3HL$xH HAH9u!HHEHtHH޽DHսHLC@"HHEHjHHH HHtHC$H IHHEHu%$HuH #H_LH !HuHu HEHFHHLMtA%HLHu EH.u HG"H#3HMH9Cu LkLcIEI$H) u H"3HUHHHH?HHLAE3L+IL HD$xMtImu I!E3I,$u I!LT$xE3Mu EDHL;tbIBH;"u 3I9Z"HD$xLHu AIBH;| uIBHHH;\ u IZHHXpH@hHt&H9{t 3HHtlHCHHL$x=HtH@Ht 3IH53HHuI HL$xHnH.Hu H<LT$xLHu @I*u I3Lt$xE3H]H;@AH XHAH9uL-Mt IEHAHvHLCLHHHH dL-eMtIEBH HLHu%wHuH HLH Mu ?IEIH3LMtAL HEHHHu ?Imu IHE3H9Cu!LsHsIHH) u H3HHUHHHH?HHLAE3L+HL _LMtI.u I`3H+HL$xu HJ3HMMu >HFL;taID$H;u 3I9L$@8H;uAD$f.zt.)AI8H y >3I,$u IE3E3E33II,$u I8=H HAH9u!HHEHtHHDHHLCHHEH?HHH ģHţHtHCH HHEHu%HuH H4LH H]HuH|$`=HCHH2LMtAsLLHuH|$`<H+u H3HMH mHAH9u"LLD$`MtIHEHHLCLHD$`HHHH LMtILH HHD$`LHu*HuH H LH LD$`Mu ;I@IHLMtAOHHHu ;HL$`H)uH RHAH9u"LLD$`MtIHuEHlHLCLHD$`HHHH LMtIL~H ߤHHD$`LHu*dHuH pHLH LD$`Mu :I@IHLMtA4HD$hLHu :HL$`H)uH}3ɺHL$`H9FuL~H~IHH)u H3H\$hHHLH?AL+HL E3HHHFHEHMtI/u ICE3H+u H13H/HL$hu H3Hu 9HFHHtLMtAHHu 9H.u HHm3HMI9Fu!I^MnHIEI)u I3H\$xHHHHH?HHAL+E3L I4LLHtH+u H23H/HL$xu H3Imu I E3Mu 8HME3HLuHI%H 6HAH9uYH=HtHH^H HHHurHugH H&LH KHHLCHHHHH gH=hHtH0HHHu 7HGHH8LMtA!HD$xHHu 7H/u HHMHHALMtAHHu C7H4H9Cu!LkHsIEHH) u H\3Ht$xHTHHHHLHHH?AL+E3L HLMtImu IE3H/u H3H.u HH|$xMu t6I؅y \6I,$u IE3H HAH9uH=HtHTH CH HLCwHHHHH zH={HtH#HHHu 5HGHHӰLMtAHLHu 5H/u HH 3HAH9uLLD$hMtI^H =HH4HLCLHD$hHȞHHH LMtIFHHD$hLMu 4I@IHɨLMtA2HLHu 4HL$hH)uHMH HALMtAHD$hLHu O4H@H9Cu'HCL{HEHIH) uHeLD$h3H]HLHHHHAL+H?L E3IHHtH+u H 3HMHL$hH)u3I/HL$hu IE3Hu 3HGHH4LMtALHu K3H/u HH-3H9Fu!H^LnHIEH)u HS3H\$xHHHLH?HHAL+E3L ILHtH+u H3I/HL$xu IE3Imu IE3Mu v2IAIIMHD$xHu J2H؅y 22HL$xH)ui3HL$xI΅t I$II IIHHHuHu3E3I,$u I HHHu H I.u IH. H(HhLuHIHD$(E3D$ E3HLHuE3L|$hY1HHHu HE3H}I.IuwE3H;=MEt$H1HuLd$h30HGHHwHjLMtASLHuLd$h30HE3AI؅yLd$h30I/u I HHU@E3Lu8IHHHHE@rH+Lu Hx MuLd$h30E3E33II/u IC E3Ld$h3/HGHHLMtABLHuLd$h3/HUAI6LHuLd$h3}/I/u I IM؅yLd$h3L/Imu I MHHUE3LuIHHHPHE/H+Lu H5 MuLd$h3.E3E33IImu I E3Ld$h3.HUHH?(HHH9 AuL=@MtI+H gRH YLH VLMuLd$h3-.IGIHLMtAHD$xHHuLd$h3-I/u I# ( LHuLd$h3-H H9Cu!LcHsI$HH) u H 3Ht$xHLHLH?HHAL+E3L HwLMtI,$u Iw E3I/u Ie E3H.u HS Ld$xMuLd$h3,I8؅yLd$h3,Imu I E3H?HUE3LeIHHHͥHEH+Lu H MuLd$h3\,E3E33ILH Mu 3IEIHkLMtAHLHu 3Imu I%H HAH9ۍuL-ڍMt IEH=AH=HLGLH=HHH L-MtIEBH HLHu%HuH H2LH Mu 3 IEIHGLMtAxHHu 3Imu IHH9CuLcL{I$IH)3u H3H]3H@HHL8}LHtHH8IEHAML+IHHHHHH?L +HD$xMtI,$u IaE3H/u HO3Imu I=E3I/u I+LT$xE3MuHEHD$XH|$`H|$h$H+HMLUuH}3HuLHL$x9HPHHL=CHܡL}HL%ΡH+LeuHL$py#3H|$XHT$`DHT$hD%$H;5@eHHHH )E3IHHHDžPH.LHD$xHHEu HLE3MuE3H|$X3L|$`L|$h#AD|$ E3HHȱHD$xHu(HM3H|$XDHT$`HT$hHMF#HHHEHLHtHL$xH)ubH+u HSH]XHXLUIE3HXML`HH LHtH+u HHD$xMuE3H|$X3L|$`L|$h"HIHHUpHL$x3HPE3HXH`TLHu 3H|$XHT$`DHT$h,"H HPHMIEMHH`H H.LLu H)3ImHL$xu IE3MuE3H|$X3Ll$`Ll$h!H ME33LuHAH9وuZL%؈Mt I$HkH HLHusHuhH H3LH LH&HLCLHHHH TL%UMtI$<HԮLMuHEHD$XH|$`H|$h ID$IHLMtAHD$hHHuHEHD$XH|$`t I,$u IHUE3H9Fu H~H^HHH)u H{3HHHhHLpH?HhHAL+E3L H"HD$xHtH/u H!3H+u HLT$xH|$hMuHEHD$XH|$`IMHDHHH9 u$LLD$hMtI0H jUH \LHYHD$hLMuHEHD$XH|$`$I@IHLMtAHHuHEHD$XLd$`HL$hH)uH\$x3HhHxHpHD$hLHuHEHD$XLd$`H HhHMI@HLHHxH HJLHtH+u HsE3H/u HaHL$h3H)uN3H+HL$hu H83HL$xMuHEHD$XHL$`I.ML}I H]XH\$xHHHHHEH3HmHD$hLHuE3H|$XE3HD$`3DHMHI@MIHH LLHtH+u H?3HL$xHL$hH)u'HD$hMuE3H|$X3HD$`ML}HLt$XII 3IDHT$`HD$XDHT$heHUAILHu I؅y I/u IaHHE3HDžxIHHHH3LH+HL$hu HMu E3E33II/u IE3H|$XE3L|$`3oHȘE3ILHu I؅y I/u IkE3HHE3LIHHHiH H+Lu HHD$hMu E3E33II/u IE3H|$XE3L|$`3xH;5H 3H\$hHAH9u"HHD$xHtHH50EH5'HLFHHD$xHHHH pHqHtHD9H H:HD$xHu%"HuH .HLH SLT$xMuE3H|$X3L|$`IBIHLMtAHHHuHEHD$XL|$`=HL$xH)uwHAHMHD$xLHtHH9Fu#H^H~HHH)uH%LT$x3HHHLH?HHAL+E3L HLHtH+u H3HL$hHL$xH)uE3H/Ld$xu H3MuHEHD$XH|$`7HUE3IHHuHEHD$XLd$`I/u I@3HD;؅yHEHD$XHt$`H/u HE3IH2HHHHHׄHH+Hu HHuHEHD$XHt$`LE3E33H,H/u HuHE3HD$XHt$`HlHH]HHHH IHHMHMH ׉H+Hu HHuHMHL$XHt$`I.HH}uH83HHH9 }uL=}MtI+H aLH SL}H}PLMuHMHL$XHt$` IGIH:LMtA{HD$hHHuHMHL$XH|$`I/u IHpHHH9 E}uL=D}MtI+H 膣H L}H}LMtIGIHLMtALHSI/u IzH#AH9Fu,LfH^Ld$xH\$hI$HL)6u H@E3HLU3LHLLHuH|$`H4HIGJAMM+HHIHLH?L iHMtI,$u I3ImHL$xu IE3I/u IxE3H+u HfLl$hHuLl$`HHH HE3H]hIH|$pHLH+Hu HLl$hHuLl$`L}E3LeIIH} !HH 3HHEHE3HE83HD$`؅} H/u H3i H]XIIH\$hHLH, H HE3HIHHHtH+u HI/HD$hu IE3H HEH}HHt H.uHAIHHR H؅@ H/u HtqHM 3D$@E3ɉD$8E3D$03HD$(HEHD$ HHgHuHLVH]9H/uHHuHy{3HHH9 \yu$L[yLD$hMtI0H 舟H L0yH!yHD$hLM: I@IHLMtAHLH HL$hH)unHH|$hH9Cu"L{LkIIEH) u H:3IHLHHH?HHAL+E3L IHMtI/u IE3H+u HE3H4 HMLH}HHt I.uADl$ E3HH蒞HHHIHEHLHH/u HPH+u IAHEHE3HH3HHLH%H 'HHMIGIHHH H.LLHEu H3I/u IE3MHMHtHD$`H(uyHxE3HHH9 vuH=vHtH+H H LvHvHHHGHHOLMtA0HD$hHHH/u HHAH9Fu&H^L~H\$hIHL)6u HE3H3LHL2HHCH HHqJHGDLM+HIHHH?HL LMtI/u I E3H/u H3H+u HH|$hMIUL `I;H;XILHOImu IIGIE33HHD$hHIϾHHIHHI)7ugI\IMH;MELD$hII;uHII} H%HHD$hHIEHxHHZImu IHU(HL$hHM(HtHE8H(u HHM0HHD$hH]0HtHEHH(uHHH CE3IHHDžH/Lu H<3MImu I!I$HH E3ILHI,$LHELu IM I,$u IIFIHuLMtALHL}HIHHImu I^Le E3MD$IppHFL9n<LHIHH?LNMHIAH+u HH/u HIFIHLMtAHHHILH]H/u HI/IMLmLmurL}E3IIE3HHHMHE0HuHEHHE(HMLt$`LHHE8,؅HD$X3HT$`DDHT$hHD$X L|$` HD$X HD$XL|$` H)0uH HۣAH Ll$`HD$XHMHD$XHEHEHM0 I/u IlE3 uHH|$`HEHELu$HD$`HD$XE3HD$XE3HEHEHEHELl$`H @L qM@HvH =Ll$`롻둻H|$`뎻HD$XE3HD$XE3HEHEHEHEH|$`GL|$`HD$XHD$XHEHEHEHE~!H HwLH H|$`HxjH|$`HD$XHMHD$XHEHELl$`HMHD$XHD$XHEHEH|$`HD$XHD$XLl$`H}yHD$X HD$XL|$``HD$X HD$XLl$`GHD$X HD$XLl$`.HD$X HD$XH|$`HD$X HD$XL|$`HD$X HD$XL|$`HD$X HD$XH|$`Ld$pLeI$H]XHHEE3IHH HH _LHtH+u H3MuH|$`H|$hCMEIxpH L9H ZILHHLGHIAH+Hu HCHHImu I'HFE3H3LELHuHLMtA"HM3HT$`DHT$hHL$XHoIL$I$H(HE3H zLH0L(6I,$Hu IcE3Hu$HD$XE3Ld$`E3HD$XLd$hH+ILH}u!Ht$PHM|HHFLMtA/HHu&HD$X3DHT$`HD$XDHT$hHځH;tsHGH;u 3H9_3H;&uGf.ztE3FAHH؅y)HD$XE3Ld$`E3HD$XLd$hH/u H63t I$II$y !$3ۅÅID$IH({LMtAHHu !H؀E3E3HJ؅y&HD$XE3E3Ld$`3HD$XHL$h!CH/u HwHlE3HHH9 ku$LkLD$hMtI0H zސH zLjHjHD$hLMu!HD$XE33Ld$`HD$XE3'I@IH.vLMtAHLHuE33HD$X'HD$XLd$`[HL$hH)uHFHHuLMtAHD$hLHtHH9CuLkL{IEIH);u H,3HkH3HLHD$xLH4H)uHH\$hHIBAL+MHIHHHH?L YHMtImu IE3H+u H3HL$hHL$xH)ug3I/HL$xu IQE3HuE3HIHH]gHHuHD$XE3Ld$`(HD$XHjLH yHD$XE33HD$XHL$`(H/u HHHLIHHH1HD$X3E3HL$`HD$XE3HL$h+?H HDM@H 33HL$`HL$hHMHL$XE3Ht$PL5HMHt H)u'HL$xHt H)uMtImu IMtI,$u IHL$`Ht H)uMtI/u IHtH/u HHL$hHt H)uMH AD3bLt$X3LeH]HM($ ($0L$HL$XHt H)u;HMHt H)u&MtI.u IHML$PHt H)uHMHt H)uHMpHt H)uMtI,$u IHML$`Ht H)uHMhHt H)uHM Ht H)uoHMHt H)uZHMHt H)uEHM(Ht H)u0HM0Ht H)uHtH+u HH$pHtH.u HHMH$hHt H)uHEHtH(u HHHH3Hx_]LUSATHl$HpHF]H3HEIs3I{3MkHMsIM{E3HEE3HEHEHxcHM A){H]HhH(E)CLMHLd$(E3D$ E3HHELHuL5W HIHu IHb LuH(HhHLd$(E3D$ E3HLHuL5 HIHu IaEG3E9FL}u3E… Hf.D(zu/HtL5; Hf.ȟ(zuHtL5O H 6eA\E3HAH9cuH=cHtHHVs@HMsHLCHHdHHH scH=tcHtHBdH dHeHHu%OHuH [HܑLH HuL5< HGHHpLMtAHHHuL5C H/u H(YHHuL5 HAH9CuLsHsIHH) u HjIHLuHH}H?HUHAL+E3L HLMtI.u IH/u HH.Iu HIMuL5_ I؅yL5@I/u IH=yHUE3LmIHHHiHEeH/Hu HkHuL5E3E33HH+u H6L5WH]A(LIHHu (HHu H iHS HMLEHD$pHuDl$hHL$`HD$XDl$PHL$HHt$@Dl$8HL$0H TH|$(D$ wHHu I/u IXH/u HIH.u H:IH a3HAH9N`uH=M`HtHHo@HoHLCHHIaHHH `H=`HtHBH (aHȯHHu%HuH H?LH HuL5HGHHrLMtAHLHuL5H/u HHH9CuHsLsHIH) u H3HEHUHEHHHuHL}H?AL+E3L I蜭HHtH.u HI.u IHuL5H\HHH]H0HEHHuL5sHHHu H-H _H]HAH9W^uH5V^HtHHm@HmHLCHHB_HHH ^H5^HtHBH !_HHHu%HuH H8LH HuL5HFHHfLMtAyLHHuL5H.u HH w^3E3HAH9O]uL%N]Mt I$HlAHlHLCLH)^HHH ]L%]MtI$BH ^H觬LHu%HuH HLH MuL5ƿ1ID$IHUiLMtA^HLHuL5I,$u IHE3H9CuL{LkIIEH) u H3HEHUHEHHL}HAL+H?L E3IyHMtI/u IzImu IjE3HuL5˾6HI9Fu"I^I~HHI.u I'IIܹHH]HHuH?HUHAL+E3L HѩLHtH+u HH.u HH/Iu HIMuL5I蘯؅yL5AI/u IjH=qHUE3LmIHHHbHEH/Hu HHuL5:E3E33H蠵H+u HL5R HELELHP MIH bHHD$pHnDl$hHL$`HD$XHEDl$PHL$HHD$@HEDl$8HL$0H 6HD$(D$ {pHHL5MMtE3I/MuI&E3E3HtH/u H HtH.u HMtI,$u IMtImu ILeH vHuMH}DLm3"T3Mt4"I/u ILeHHuLmH}Imu IoD($ ($0L$HL$PL$XHtH/u H2H$`HtH.u HH$hMtI,$u IHHMH3bHpA\[]LUWAUIkHHOH3HEHE3IIL;I[ AIsAMcLHMH XMsM{ELMHAHULEH9WuHWHtHL5k@L5kIMFZHHXHHH WHWHtHBH gXIHHu%HuH H~MH "HuL%&mHCHH6kLMtAHELHuL%mH+u H[ID$H;7u I$IAH H;u HAXI'H H;u HAXIHWI6HHuL%gnMHv^HHEHZHEH]QLCHMǃIADȺ7HHuL%mH+u HHMH'H9AuLyLqIIH)uXIHlAHEL+HL}HHuH?E3L IHEHHMtI/u IH.u HI.u IHuL%HmH+u HI$HILeIH)uHiIHL%iH+Leu~3H]LHHH9GQHLRH9'UMu$H5#UHtHHhHEHHHhHHEL@>HHlUHHH TH5THtHDHUH GUHHu&HuH HaLEH HuL%tHFHHhLMtAHELHuL%ƶtH.u H=ID$H;u I$IAH H;u HAXI'H H;u HAXIHTIHHu uHGH;uHHAH "H;u HAXH'H +H;u HAXHHTHHHu uH[HHEH<[HEHAWHEHuH]MHD}MHND LCH@LAHMߺ虩LHu t/H.u HH+u HHMHH9Au LqLiIIEH)u33H iAHEL+HLuHL}H?E3L IHEHUHMtI.u IVI/u IGImu I7Hu tSH+HHoLMuLIGH TL5fH;IWHuIHBHtgHHwKLvAOIAOAGHH HHoAOAGHH HUHA`IILACAOHAIO+H;uAGX IIULMt5I/u IWH]߹LeELuLHu|xHImL%4u IHtH+u HMtI.u IHtH.u H}MI/IaHbMIGIHH\HEHEHkZIG HH u`HUHUH+Lu HI.u II/u IMu+xL%>E3MH mD3/JH}LeL$L$H$H$MtI,$u IzL$HtH/u H^IHMH3ǰHA]_]MMMMMMMMMHl$ WAVAWH HAE3HXHELMtAH\$HHHt$PHuH-6wHGHH\LMtAHHHu H/u HsHMLHWHALMtAHHuH-HCHHK\LMtADHHuH-mH+u HH QHHHtXH/u HH~UHHHuH-CH+u HHHHHuJH/H-ԯu HYHtH+u HELH fD3GMu)6H/u HH.Lu H HLI.u IHt$PIH\$HHl$XH A_A^_LUSVIHHCH3HH03I{HMcIMkMMsE3M{E3HE3A)sHEHEHIHM A){H]HhH]H(E)CE3HL|$(E3E3D$ HHEHHuH= HHHu HH$I H]H(HhHL|$(E3D$ E3IHEHHuH= ?HHHu HHH H]H(HhHL|$(E3D$ E3HHEHHuH= HHHu HHMHEH]Q39PuC3;u3ɅHf.(zuHtH=- WHZf.rD(zuHtH= "I%f.=(zupHtH=Ĭ 3f/H=z`HUE3LmIHHHVHEH/Hu HHuH=Y E3E33HwH+u HH=) S3fA/H=_HUE3LmIHHH,WHESH/Hu HYHuH= E3E33HۣH+u H$H= fA.H=C_HUE3LmIHHHUHE踖H/Hu HHuH=" LE3E33H@H+u HH= H](LI LHuH=ª  (HHuH= ] A(ĺLHu  % HHS LE3HD$pMHOL$hHD$`Lt$XL$PHD$HHt$@L$8H {HD$0Ld$(D$ ]HHu  I/u II,$u IzH.u HkI.u I\Hu2H GHAH9>uH5>HtHHU@HUHLC>HHlGHHH A>H5B>HtHBH KGHHHu%ռHuH HbtLH HuH=  4 HFHHOLMtAHLHu  q H.u HGH FHAH9=uL-=Mt IEHTAHTHLC1LH_FHHH D=L-E=MtIEBܻH =FHݔLHu%ǻHuH ӺHTsLH MuH= IEIHPLMtAHEHHHuH= Imu I-HֺE3H9CuL{HsIHH) u H3HEHU@HEHHHEAHEPL+HL}@HE3H?HL 襒HMtI/u IH.u H3HuHuH= H$H9Gu LwLgII$H)u HLHHLuHH]H?HUHAL+E3L ILMtI.u IH+u HI,$u IMuH=F p I̗؅yH=' B I/u IH=YHUE3HuIHHHPHELH/Hu HRHuH= E3E33HԝH+u HH= H mCHAH9b:uL%a:Mt I$HQAHQHLCLH$CHHH :L%:MtI$BH CH袑LHu%HuH HpLH MuH= ID$IHJLMtAYHLHuH= UI,$u IH VBLHuHAH9T9uL=S9MtIHwP@HnPHLCٴLHBHHH 9L= 9MtIBH AH膐LHu%pHuH |HnLH MuH= IGIHuLLMtA>HLHuH=d &I/u I۴HAAH9Cu HsLkHIEH) u HIHEHUXHE`HHEAHEhL+HHuXHE3H?IL PHHtL)>u HRM)}Leu I?MHuH= HϵIH9Gu!HwLwHIH)u HIIHHu HH](H?HU HAL+E3L I詍LHtH.u HH+u HI.u IMuH= Iv؅yH=ѡ I/u IHH=UHUE3LeIHHHLHEH/Hu HHuH=` E3E33H~H+u HDzH=0 ZH ?HAH9,6uL5+6MtIH?M@H6MHLCLH>HHH 5L55MtIBMH >HNLHu%8HuH DHkLH iMuH=m IFIHeFLMtALHHuH=, I.u IH >E3HAH9&5u!H%5H]HtHH=%LDH=LHLGHHEH=HHH 4H4HtHF/H =H0HEHHu%HuH "HjLH GHuMHCHHLGLMtAHLHH+u H3HEH5H9Gu H_LHIH)u Hc3IHEHUpH}HHExAHH]pL+HHE3H?IL LHtH+u HI/u IE3M H.AH=Du HɯMtI/u IMtI,$u IMtI.u IMtImu IwHMHH)HI9Eu#MeIuI$HImu I/3HLe0HLu8H?HU0HAL+E3L H݈LMtI,$u IݮI.u IήH.u HMuH=# MI詎؅yH= I/u I{H=PHUE3HEIHHHFHE%H/Hu H+HuH= E3E33H譔H+u HH=_ HE3HMLEt$`LHP MIHAHD$XH|$Pt$HHD$@HL$8H T{t$0HD$(HEHD$ OHHuIH= MtI/u I\HuH YZLuLLeD333Mt,I/u I$HLeLuI,$u I D($($($L$L$L$H$MtI.u IL$HtH.u HHHH3H^[]LUSIkHXH)/H3HEIs3I{3Mc3MkMMsE3M{MLHD$`3HEHD$XE3HN5HT$8HM A)sHhH(LMH|$pLD$PHHt$(E3D$ E3IHD$@HHuLd$8 3HHHu HsH4 HhH(HH|$(E3D$ E3IHD$HHHD$PHuLd$8 HHHu H+E3H]LD$P 9{ If.q(zuHtLd$8 \I\HD$HHHuЬHtLd$8 ,HEA(LuLd$8 HDWH*A _LuLd$8 L}L;=#A HuEH@H蕙HD$0LHuLd$8 HHc@HLfIHD$PHH{H@LMuH MHAHHL$PHLHYHF3IH9Cu*HCLcHHD$XHI$H) u Hf3IHD$XHUHEHHH}HML+H?L E3IHD$8LHtL).u H3L)+HD$XHD$xu H3HD$PHD$hMM),$u IըHHEHHxL)MtL;u HIHuMtL;tIEIIuHlLE33E3HE(LD$HLH`HH ^Hͦ3HHT$8HT$0HUL H oUA .HMLL$xLD$hHT$0\HD$xHD$XyLD$hHT$0LȹHD$hHD$PLHD$0HHD$8LHE3HI]I/Hu II.u IE3HtVH茇H+Du HrEx7LL$XHLD$PHT$8T3E3HT$XLD$PHT$8 HEL=H@xHL(Ht H)u HtH.u HMI,$IڦHt$@Ld$8WHL$8Ht H)uHL$PHt H)uHL$XHt H)uH|$XHEH@xHL(Ht H)ueHtH.u HQMtI,$u I<L}AH 2E3LD$PHAH9;,H6,HD$8HHcHEH@xHL(Ht H)uѥHtH.u HMtI,$u IHQ3E3I>~I/Hu IHuLd$8 H.u HaHt$@H PHH dL= HL$0L))u(Ht$@Ld$8H?EH?HLC,HHD$8HU1HHH +H +HtHDӦH 41HHD$8Hu%HuH ȥHI^LH Ld$8Mu ID$IH:LMtAHLHu 3I,$u I,H 0HAH9Z*u"Ha*HD$8HtHH=>EH=>HLGHHD$8H;0HHH *H*HtHDH 0H~HD$8Hu%HuH H/]LH ӡLd$8Mu 3ID$IH<LMtAoHHu I,$u IHE3ILd$8H9Cu$HCLsHD$PHIH+u Hۢ3H\$PHULH]IL}H+I?HH}L E3I|HD$XHtH+u HE3H/LD$Pu HuI.u IfHL$XE3Hu 3 HHHL$pHHu4He+WHO Ld$XLAL}LHGEH8HD$0IiHD$hHHuLd$8 U H>8EI7HHHHY3IH9Cu$HCH[HHD$XHHH)u|3HD$XLIHEL+I?HELuM HE3H2{LHtH.u H3H+HD$Xu HMu,L= H/u HHt$@3Ld$8v I.u I۠E33HD$xLM~3H|$0Ll$HMO`M(IO ~HI;|H|$hLl$xIWH0.E3HyH/Hu HcHu Ld$8 3 Ht$@AH;H.BuLd$8 H:A HBuLd$8 u L;-ɠH D,HAH9i$uH=h$HtHHl:@Hc:HLCΞHH+HHH !$H="$HtHBzH +H{zHHu%eHuH qHXLH HuLd$8 HGHHd5LMtA5HLHu m H/u HٞH :+HAH9#uH=~#HtHH5b9@H5Y9HLFĝHH*HHH 7#H=8#HtHBpH *HqyHHu%[HuH gHWLH Ht$@Hu HGHHR7LMtA+LHu f H/u HҝH{AAH9Cu!H{LcHI$H) u H3H|$PHH}HLmH?HUHLuAE3L+IL IwHHtL)?u HKM)>HD$Pu I4E3M)<$u I"E3Hu 3HHHu HHIH &LHֹLALHu*L 3A H A(#Ld$8 &L;-zMtHV)I讈u H (E33HAH9Y!uL5X!MtIH6@H6HLC^LH(HHH !L5!MtIB H k(H wLHu%HuH HULH &Mu Ll$`2IFIH1LMtAŞHD$PHHu Ll$`I.u IbIEIH8LMtAuLHu Ll$`H 'HAH9, uH=+ HtHL=5@L=5IMGHH?'HHH H=HtHBH 'IuHHu%HuH H5TMH ٘Hu Ll$`HGHH3LMtAxHD$XHHu Ll$`H/u HH3AAH9Cu#HCHsHD$PHHH) u Hܙ3HD$XHUH\$PHHEAHHuL+LuHE3H?HL sLHtL)>u HM)>u IwHL$XE3L)9udL);H|$Xu HQE3LD$PMu Ll$`I$I$Hu IIIH\$`H|$pH@"E3WHO LAH %"HHt$@LL$HLHD$0LALHu*L 3AH =3 E3L;-pLl$Pt"HJ%I袄u3 E3HtH+u HSIEIH5LMtAfHD$PHHu3Ll$`E3 HH:LHu3Ll$`E3 jH+u H֗E3I,$Lt$Pu IHuEH8.E3HHD$HHHu Ll$`H-EHHD$XHLHH3IH9Cu/LsLSILT$XIIH+uH$LT$X3ILLt$xIHT$xL+I?HE3M LuE3IpLHtH/u HӖH+Lt$Pu HLt$XMu2H|$HL= H/u HHt$@ILl$`I,$u IsM2H|$0MHD$hLl$`HLl$`fI0LN`H0I@L0I8MH0 HN tE3IE E9] 1M0fIH@ILcQEuH(H0IH@(D88t"H(HP8| HB(HcB AuAHA0H;0}HHA0IH0xLa0IH@(IH(H+0ZMExYJ(IHH9|%L$AIHHH)0IyIcHD(IH(H0AIE;]HHD$hMHH|$HE3H]"HMmH/Hu HH 3E3Ht$@L=Ll$PMtImu IZHtH/u HFMtI,$u I1MtI.u IHL$XHt H)uMH 5AD3kHL$p3Lt$`H+u H֓HL$pHt$@HHHt H)uMtI.u IHtH.u HHE($L$(L$0L$8L$@H$HH$PHtH(u H=HHMH3覀HX[]@USVWATAUAVAWHl$HXHH3HEHHLHMM3HEH HMDHMIHMHhH(HMHMHMHM E3HLl$(E3E3D$ IHEHHu s HHHu HRH H]H(HhHLl$(E3D$ E3HHEHHu t HHHu HH H]H(HhHLl$(E3D$ E3IHEHHu u' HHHu HHU3HEH]R9PuC3;u3ɅRIeLHuMHt x HeHHu'Ht y IeHHuHt zm J6H;}|H=3HUE3LmIHHH+HEjH/Hu HHu } E3E33HwH+u Hg}L}IMIEՎLHuH5~=HLHuH5~HHEHHu H*IW LEH nD$pMHD$hH)H\$`D$XHD$PH*Lt$HD$@ HD$8Ld$0D$(|$ 1HHu ~`Imu IZI,$u IJI.u I;H+u H,HH}HHHu H}H c3H]HAH9BuL5AMtIH=)@H=|)HLGLHHHH L5MtIBH HiLHu%~HuH H HLH Mu IFIH"LMtASHELHu^I.H5j|u IH}LH}Mt ImH}H}uIǍH}H}MI.u IH HAH9uH=HtHH5+(@H5"(HLFHHHHH H=HtHB9H H:hHHu%$HuH 0HFLH UHHGHH%LMtAHELHEHH/u HH 3HE3HAH9uL=MtIH5%'@H5'HLFLHHHH L=MtIB3H H4gLHu%HuH *HELH OMIGIH LMtAHEHLHzI/u IHHE3H9Fu"HFLvHEHIH)u Hj3HEHUHuHHE AHEL+HE(E3HHuHIH?L eHHtH.u HI.u I3HEHHMHH9AuLqLiIIEH)uŠ3LL}HHLu0H?AL+H}8L L}@HE0E3HIjdHMtI.u IkH/u H\Imu ILH33HEE3HEH5xI,$u IHMHt H)uHtH/u HHMHt H)uډMtI/u IƉHMHKH)A6HUHDH9BuHZLbHI$H) u Hq3HH]HHuH?HUHAL+E3L I&cHHtH+u H'H.u HI,$u IHu Hh؅yH5Tw)H/u HˈH=+HUE3HEIHHH"HEubH/Hu H{Hu E3E33HoH+u HMHEH @gLED$`HHP LHH"HD$XH8"L|$PD$HHD$@HEHD$8HF"D$0 HD$(HEHD$ *HHu"H5+vHtCH/u=H2H/u HHH]HjHHuAH5uHEH 5H}LLuDL}3HE Hu3Mt IL}LuI/u IMtI.u I HtH/u HHtH.u HHHMHH3LtHXA_A^A]A\_^[]HVWHhHX3H;HhL` LLhLLpLxeH DHAH9{'uHz'HtHH-v@H-mHLEhHHHHH 3'H4'HtHBH uHaHHu%HuH H?LH 0HHu AHfH9Cu"LsHHCHIHH))uHLLt$ IH|$(H+I?HD$ L HE3HA_HMtI.u IBH+u H3Hu AXH2H\HBu"H EHNAH AI_HH؆Ht AH LHAH9%uH%HtHH-@H-HLEHHHHH %H%HtHBLH HM_HHu%7HuH CH=LH hHHu A*HH9Cu"L{HHCHIHH))uƒHLL|$0IH|$8H+I?HD$0L HE3Hy]LMtI/u IzH+u HkMu AIHT$@H E3ILt$@ILd$HI.Hu IHuABH+uHH(uI$IIEHILAHHu"AL 4qH (3. H(uL|$PLt$XLl$`L$H$H$HtH.u HhHh_^H\$Hl$Ht$ WAVAWH H 3LHAH9#uH#HtHH-R@H-IHLE$HHRHHH "H"HtHBЃH 1H\HHu%HuH ǂHH;LH HuYH;Yt0HHmuH+L5ou2HI'HCIHHLAHHu'L5oMH 0D3Hu%H(uH5ҁHH/u HH\$@HHt$XHl$HH A_A^_H\$Hl$Ht$WH H  IHHAH9!uH!HtHH=@H=HLGHH HHH D!HE!HtHB5H H6[HHu% HuH ,H9LH Q~Hu H5UnHCHHbLMtAHHt?H+u HH~HHu H5m1LHH$XHHuEH+H5mu H^HtH/u HJLH /3A3!H/u H H+u HHH\$0Hl$8Ht$@H _H\$Hl$Ht$WH H E IHHAH9uHHtHH=@H=HLG}HH HHH HHtHBuH HvYHHu%`HuH lH7LH |Hu H5lHCHHLMtA3HHt?H+u H}H|HHu H5xHGHH/Lgu H$xI$HT$8E3L|$0IHl$8IQI,$Hu IwHuH-MfHGHH LMtAzLHuH-fH/u HwH HAH94uH=3HtHH@HHLCqvHHHHH H=HtHByH ~HRHHu%yHuH xH0LH 9uHuH-=eHGHH LMtAyHHuH-d^H/u HyvHCIHMHLMtA~yyYH+H-du H/vMtI.u IvHtH/u HvLH m%D3k(I.u IuH+u HuL=vIMtI,$u IuILt$pH\$xH@A_A\_^]H\$Ht$WH HAHzHHH HLHH;s,H3HHHH;sLCH AHLHH;rH\$0HHt$8IHH _H\$Ht$WH IHHHCH H#H;wH\$0Ht$8H _H\$Hl$Ht$WAVAWH A8IDIHEwuHAH ЉAAAAAEAfA;sOA+ǙDf;s<EuHFHЉA G DfA;rH\$@Hl$HHt$PH A_A^_H\$Hl$Ht$WH HAjHًH DLD;s$3ҋD;sHCH DLD;rH\$0Hl$8Ht$@I AH _H\$Hl$Ht$WAVAWH A8IDIHAouHAH ЉA A)AEDDED:sS@͸A+ǙD:s=EuHFHЉA/ DAD@:rH\$@Hl$HHt$PfAAH A_A^_@SH)$HD)L$p(D)T$`D(D9fD/fD/"f/v1f/v+HAH 3(AXYf/fn)$D)$EWD)\$PD)d$@D)l$0E(D)t$ E(E^D^HCH HCD(H D(A(A(%m(A(A(m(XfD/rA(AXfA/vfA/D(t$ D(l$0vfA/v ^(JA(l(A(^lA^\fA/v[W=ԁ(xlnWflD(\$PD($($D(d$@($D(L$pD(T$`Hİ[($lJn\(/A(H(/X^(@SH IMWf.zWf/fH*r'Yf/(r HH [`HH [6\(Y(f/r HcH+HH [#HcH+HH [3H [HPSVWAVAWH0A9I)pD)xLD)@DG6D)HD)PD)xD)hD%~D)XD)HD)8A9QAAf.AA@EQEy EY8Aq0Eq`EipD$8AAHD$(AAPD$HAAXD$PAAh$xAAxD$ AfDnD$@D$`D|$0D\$XEE(AQD\EqAfD/D$`v $`D(E(fEnE\EQED|$0A(Ey AY(AX(Ay(~i,AYWs0f.wfQ(iY{}A(Y }\>i}D(EXfn(D\$X(D[8X }AX^(l$8X|A\k@AX(D(\$P\d$((cHAYDXl$HkP[X\(AYEX^(EY\D(EYDkp^EX(AYDY(AXDs`Y(A^^AX$xChXL$ KxT$@EYH$hfnEYDl$`Dt$hD$pDL$xD$xfDIGIIG(IYt$@fA/(pfA/gf/t$ (v<D$HA^\pg,A;Wf.zt\t$ YAYA^XL$((5g,aWf.zRA\YAYUL$pA\^t$PXt$(((Yt$P\AXAXT :2A^\fA/f,+HcHHHHȃA(fnAYA\f/(^=zEW(X=LzD*YA^X=zA^AXYA(AXD^1fA(D(\f/DXfA/DD1EWɍGEWWD*FA+EW*CE(D*A+(EYD*A(YA^E(E(EYEYeD(W(A^*X \yDYveYD$8A(YL$0DXA(Y$`^KeyWy%y-y*Y(DX(^\(^\(^\^5y(\(^=y^DX(A^\(A^\(A^\(A^\(A^^DX(A^\(A^\(A^\(A^\(A^^DX(A^\(A^\(A^\(A^\A^^DXfE/D%wDL$xD$/D$`D\$XDl$`Dt$hAFA(fnA(^d$0Y;,N;L+DGQfWB*B((^W*\^W\*Y(Y(^W\*BY(^W\*BY(^W\*BY(^W\*BY(^W\*Y(^\YA;,;Q@fn(^\Y;~+%O;+DFQWB*((^W\*B^(^W\*^^W(*B\^W\*B^^(^W\*B^(^W\*B^(^W\*^(^\^A;/;(ffffn(^\^;~f/iD$8AY\Xa,$pL$0H$hD+A(sA({E(CE(KE(SE([E(kE(pE(`fA/E(cDFAIA_A^_^[HHXHhHpWHA9I)p)xHD)HD(D)Pt)A9Qu#AAf.zuEQ AyAq0D+D)D$PE\fDnEID)\$ AiA(EQ AE `AY_C(E(WEYA(D[XAYX+f.wfQ_YBtAXD(\$ fA/wD(A,D(D$Ps0HG3H(f/(vK;~HG3H((,\fnË+fnAYAYY^f/w(t$pL$IkI[Is E(KE(S(|$`I_MHt$H|$AVH Lt$X3IHDEuMIHt$8H|$@H A^^H\$0HEuHGHAAȀA HH;|H\$0Ht$8H|$@H A^H\$ WATAUAVAWH E3EDMHEfEu*MH|$xAIfH\$hH A_A^A]A\_øfD;uUIML|$xEuHCH DAAAC+fAHI;|H\$hH A_A^A]A\_Ht$XD8T$p}IIHI HHH HHH HHf ML|$xfEuHCH DAAAAf#fA;wfAfAwHI;|MMAvHl$Pft$`EuHCH DAAAAf;s@A+ƙf;s.fEuHCH DAAAAf;rt$`HD$xfAfBu HAH ЋY|$PtfDHCH #;w;E3E3*Hu HAH |$PHt L4HH\$0Hl$8Ht$@H _H\$ UVWAVAWH L|$xIILHMu&MIHIHH\$hH A_A^_^]øLd$PH;u.3H@HCH ЋIIHH;|3@8|$ptlHHH HHH HHH HHH HHH LI L H^@HCH A#;wIIHH;|6H-E`Ll$XDd$`EfHCH ЋIA;s)3ҋAD;sHCH ЋIA;rDd$`H II HH;|Ll$X3Hu+HDHCH IIHH;||$ptbHHH HHH HHH HHH HHH LI L H~FHCH I#H;wIIHH;|$H~DHHIIHH;|Ld$PH\$hH A_A^_^]@SUAUAVAWH E3AMDHEEu"MHL$xMH A_A^A]][PWH|$X@uSIMqLt$xEuHCH DAAAC+A>HI;|H|$XH A_A^A]][Ht$PID8T$ptkHHH HHH HH@ MLt$xEuHCH DAAAA@"@:wAB6HI;|MLd$`DeEEuHCH DAAAEEE:sFA+řD:s1EuHCH DAAAAAD@:rHD$xfAEDHI;pLd$`Ht$PH|$XH A_A^A]][H(EuH(H\$ H\$X;u%HAH HL$`H\$ $H(HL$`) H\$ $H(Hl$VAVAWH EALfEuHl$PH A_A^^øH\$@f;u=H\$h;uHAH HL$pf1sHL$pA f1^|$`tBH\$hH|$HH|$p;uIFIЉG Af#f;wH|$HLL$pLD$hefH\$@Hl$PH A_A^^H\$Hl$Ht$WH AAHEuH\$hH|$HH|$p;uIFIЉ/ A"@:wH|$HLL$p@LD$hN@H\$@Hl$PH A_A^^H(Y fXH(H8)t$ (Y(t$ H8@SHP)|$@HD)D$0D(D)L$ D(Y f!A(HY rf( XXAYD(D$0AYD(L$ ^((|$@HP[H8)t$ (Y(t$ H8H8)t$ (Y(t$ H8HHf/ e)|$ (reH(|$ HH)t$0(gWWS^(Qf/fH(t$0sH,(|$ HH@SHPHQH )t$@()|$0(D)D$ DD\f/vfAYXf/w(t$@(|$0D(D$ HP[@SHPHAHH )t$@)|$0(D)D$ D(5/(\f/wfHCH (\f/v(PWeP(t$@AYD(D$ \((|$0HP[H\$WH HHHu 3H\$8H _Ht$0HHH HHH HHH HHH HHH HH H H;w'@HCH ЋH#H;wHt$0H\$8H _HCH H#H;wHt$0H\$8H _@SH@HAHH )t$0()|$ (f/c(r0c\\OY\((t$0(|$ H@[Wf/v$X(`OYX(t$0(|$ H@[((H(t$0(|$ H@[WHD)L$@D JfA.D)T$0D(zt fD.czuWD(L$@D(T$0HĈH$)t$p)|$`D)D$PD)\$ DcfE/vA(A\,3WA(*AX^(YxN(D(Y (\:bfE/X D(\$ DYYX nYX ZYX FYX 2YX YX YX Y(t$pX ^X aDXD\v|A\(MD\Hu(|$`A(D(D$PH$D(L$@D(T$0HĈ@SHPHAHH )t$@)|$0(D)D$ D(W(f/wHCH f/(va\^(M(t$@Y(|$0AXD(D$ HP[HH)t$0()|$ (7YX(t$0(|$ HHLHSH)p()xHD)@D)HD(D)PD)XDaD)`AWD)l$ yNHCD(H fA/(D%}EWHCH AY5ND(EWA(AYf/rN( L(A(K^AX(KH,H| fA.z(u&HCH fA/(r fA/rL$A(sA({E(CE(KE(SE([E(cE(kI[HHXHhHpWATAUAVAWHp)p3)xMD)@MD)HD)PLH$D_A(LhMH$EWD ^=_A^ƅtpfA.ztgfD/L͋Ifnr!Y(f/r Hc;fHc1A(Hc\Yf/r HcH+:HcH+3+A<~#A\4HI;dH$~A\(t$`L\$pI[0Ik8Is@E(CE(KE(S(|$PIA_A^A]A\_@SH0)t$ H(j  "H\^Y(t$ H0[@SHPf.H)|$0(D)D$ D(<u^(|$0D(D$ HP[Wf.z)u'DYM]A( X(|$0D(D$ HP[fD/)t$@vWD\DY ]A( HD(xW(f.w WQ(IXEXYAX(3Y=\(HfnAXY \5 X(t$@(|$0D(D$ HP[@SHP)t$@H)|$0(D)D$ (D((HYY56\( (t$@XAYD(D$ ^((|$0HP[HH)t$0()|$ (gY(t$0X(|$ HHH8)t$ (^(t$ H8H%I@SH@f/ :\Hr H@[Wf.z u3H@[W \H|$P()t$0)|$ ^GHC(H 3f/(vHSH Yf/w(|$ H|$P(t$0H@[HSVWH)p()xHD)@D)HD)PD)XD)d$`D)l$PD(D)t$@D)|$0FW$fA.w EWEQ A(FD(DY5Z ZDZ=cZDX5zZDD=ZA(A(\ZYY^\YX DZA(\HZD($$DX UZ^DT$ D\D$HFHHF(H\D((T )D(D\A(A^AXYAXXyYEfD/ KY,r fE/xYfA/v fE/tGfDnEfE.zt fD.]Yz u EWfE/vA(A\,3WA(*AX^(YEfE/((Y5D(\ XXDYYXYXYXYXYXYXYXYXt^XXXDXD\v$|DA\(GDD\Hu=XA(+D($DXEY$A^AXCW\*Y$A\A\f/sD$DT$ D(t$@L$A(sA({E(CE(KE(SE([E(cE(kD(|$0I_^[H(HAH ЋHH(H(HAH H(H(HAH HH(H8)|$ (W8XE *W#X^(|$ H8BH8)t$ (o(WXf.wWQY(t$ H8(BY(t$ H8@SH0)t$ H H( ^((t$ H0[H\$Ht$WH@HAHH )|$ HL9HWHHxH*HHHH H*XAYWZDHH;|D(D$ (|$0(t$@H\$`Hl$hHt$pHP_@SHD)L$pHD fA.D)T$`D(z u D)\$PEWfE.z uWfE/)$)$D)$D)d$@D%:QE(E(E\E^HCH H(|fD/(rA((;f/rDA(\A^;(A(AW(AYAX;Xf/rD\4OWD)l$0D)t$ A(Y Of.w WQ(N;D-NE(D5ND^HhA(D(YAXfD/sHC(H YYA(AY(AYYA(\f/w=:((:A(\XA(AYAYAYXf/lD(t$ D(l$0DYA(($D($($D(d$@D(\$PD(L$pD(T$`Hİ[@SHD)$HD MD)T$pD(E.z u *D)\$`EWE.z uW E/)$)$D)$D)d$PD)l$@D-NE(D%LE\E(E^HCH WHH*AYD/(rA((Y9/rRA(\A^19(A(AW(AYAX 9X/rD\sLWD)t$0D)|$ A(Y M.w WQ(8D51LE(D=(LD-3LD^HD(A(AYAXD/sHC(H YYA(AYWH*(AYAYYA(\/w<=8((28A(\XA(AYAYAYX/^D(|$ D(t$0DYA(D(d$P($D($($D(l$@D(\$`D($D(T$pH[HHXHhHpWAVAWH`)xL5 D)@H-(DLHD)HID uL)pHGHHHWHHI#xH*HHHH H*XAY<tAWH; rIHHGЅtm( \YX(AYY6f/^((t$PL\$`I[ Ik(Is0E(CE(K(|$@IA_A^_AW8=K(HGHYAWb8((AWYXf/wGfHGHAW58HG(HYAW8((AWYXf/vX53JHsAW(,HHXHhHpWHp)xH-'D)@HDJD)HD ID)PDJ)pHCH Ћ WH*Y sAW; rdH HC WW\ZH*AYYXZ(AYY4f/g((t$`L\$pI[IkIs E(CE(KE(S(|$PI_WH*AYAW6=I(HCH YWH*AYAW6((AWYX/wiDHCH WH*AYAWM6HC(H YWH*AYAW#6((AWYX/vX5HsAW(HHUVAVAWHxHXMHxLL`HLhL%$)pL-)x3D)@DHD)HD HD)PDHHFHHHHHWHH#xH*HHHH H*XAY|tAWI; HHFЅt@A A\Y(AYXY2f/PAW4HF(HAYAW4AW(XYf/wCHFHAWu4HF(HAYAWZ4AW(XYf/vX=uFHsAWAt(A\\AYYWf.w WQ(!-\((t$`(|$PD(D$@D(L$0D(T$ HxH(HAH ЋH(HHHAH )t$0()|$ (Y(t$0X(|$ HH@SHf.H)$(D)l$0D(<u LAgf/v"HAH X\TYD@?f/)$D)$D)L$pvD E(D^DX@f/ D (Y ?WYAXf.w WQ (+(AXW(Xf.wQv+\(X^D(DYXEXD^HCH D)T$`D)\$PD)d$@D)t$ D%B?AY*HCD(H AXEYA(EXD^A\YD5>EWA((\Y\fA/~(^*AX\fA/s[HCH AYW*HCD(H AXEYA(EXD^A\YA((\Y\fA/rA()HC(H  =f/D(t$ D(T$`vW53? S>AX(T<AX)3A\D(d$@fD/D(\$PtiY_ (^Wf.wfQ()Y>AXf/vX5=f/5=vX5>(D($($D(L$p($D(l$0Hİ[@SHP)t$@H)|$0(D)D$ D(K(A(Y 4=YY(YYXWf.w WQ((HC\H EX(A^YX((X^f/r( Y^((t$@(|$0D(D$ HP[H8W)t$ f.(zt \^#((t$ H8LSHf/ <HE)CD(rE(CHİ[;)$)$=D)L$pD\D)T$`D)\$PD)d$@D)l$0A(D)t$ 'D-q<A(W <D(A(i'D%E(E^D\D(HCH HC(H D((\(AYXA('&fA/(wf/w(A(^X&((\AYA^YA^f/{D(t$ D(l$0D(d$@D(\$PD(T$`D(L$p($D($,($Hİ[HSH)pH5f/D)PD(D)XD(&f/)xD)@D(D)HD(D)`D^D)l$ D^DH HAH H D(HAH D(A(A(%(A(A(%Xf/rD(L$`Wf/D(D$pv8^(D(d$0($D(l$ ($D(T$PD(\$@HĠ[A(W%(A(A^F%A^(_\\%(( %X%\($w A(H( X^(gH(Y 8 XH(H8H )t$ (HAH \($W9Y(t$ H8@SHP)|$@HD)D$0D(D)L$ D(Y &8 A(HY 8( XXAYD(D$0AYD(L$ ^((|$@HP[H8)t$ ( Y(t$ H8@SHpyHtA3AHAHp[)t$`)|$PD)D$@D)L$0EWD)T$ D@H HAH H (XHAH A\D((EXYE\A(AYXfA/sfA.zt(&#D(T$ (Y 8WD(L$0^(t$`f.w WQ(#(CY(|$PAYD(D$@KHp[HH)t$0()|$ (YX(t$0(|$ HHh"HSH)pH)xD)@D)HD(D)PD)XDyA(D)`\+"HCD(H fA/(EWHCH AY!E(D\A(AYf/rP(!(A(!^AX(!,| fA.zuHCH fA/(rHfA/r(|$pL$A(sE(CE(KE(SE([E(cI[@SH0)t$ H(H (\\^YK(t$ HH0[@SHP)|$0Wf.HD)D$ (D(z'u%Y=4(zX(|$0D(D$ HP[f/)t$@vT\Y=f4(>H(W(fA.w WAQ A(o XXYX(JH A(Y 4jHfnXY 3fE.X<u5(t$@(|$0D(D$ HP[@SH)t$pWf.H)|$`(D)D$PD(D)L$@D)\$ D(z!uD i3AY?(XfD/D =3vW\D)T$0AYHD(kW(f.w WQ(:XEXYAXD(T$0FH (AY4HfnAXAYf.(X<u53A(HAYm(|$`XD(L$@AYD(\$ AYD(D$P^((t$pHĀ[HH)t$0()|$ (Y(t$0X(|$ HHH8H )|$ (HAH (]\WM3^\<(|$ H8HHH )t$0)|$ (HAH 5 ((\W2W2(^\(((t$0(|$ HH@SH R1If/2fnr'Yf/(r 蕼HH [HH [h\(Y(f/r_+HcH [迱+HcH [HHf/ 0)|$ (rEH(|$ HHHAH )t$0W1(\^(n(t$0(|$ ,HHHH\$UVAVHPIILHA ~ L L|$xA;)t$@AL)|$0D)D$ WDfDnEfD/A(vMH|$pB<HFHfn(^X,fn\Atf/wH|$p(|$0D\(t$@L|$xA,D(D$ +D;NHcH$HPA^^]3H$HHPA^^]H(HH(HSHE/H)p)x=ZD)@D(D)HD\D)PD)XD)`D)l$ A(D%8D(D-/D\D(E^DHCH HC(H \D(A((fA/(wf/w(A(^X~((\AYA^YA^f/rL$A({E(CE(KE(SE([E(cE(k,A(sHI[H8HAH )t$ (W4/(WYX/f.wWQY(t$ H8(Y(t$ H8@SH0)t$ HH(^((t$ H0[H(H HAH  \(PW.H(@SHD)L$pHD efA.D)\$PD(z(u&H HAH D\A(W1.D)T$`EWfE.z uWfE/)$)$D)$D)d$@D%-E(E(E\E^H HAH H (HAH A(\(efD/(AWrA((Wf/rHA(\A^.(A(AW(AYAXXf/sD\+WD)l$0D)t$ A(Y V,f.w WQ(D-^+E(D5+D^@HA(D(YAXfD/sH (YHAH YA(AY(AYYA(\f/w=E((:A(\XA(AYAYAYXf/iD(t$ D(l$0DYA(($D($($D(d$@D(T$`D(L$pD(\$PHİ[@SHP)t$@H)|$0(D)D$ Y=g*WD(f.w WQ (((AYHWf.wQ^(|$0D(D$ ^((t$@HP[@SHf.HD)$D(D)t$ D(<u *fA/v"HAH X\Y)N4)fA/)$)$D)L$pD D)T$`D)\$PD)d$@D)l$0vA(A^AXxA(WY )AYAXf.w WQ (J(AXW(Xf.wQ'\A(AX^(YXAX^HCH D% )AYHCD(H XDY(EXD^A\AYD-(EWA((\Y\fA/s|fD(^hAX\fA/sZHCH AY'HCD(H XDY(EXD^A\AYA((\Y\fA/rA(HC(H  'f/D(l$0D(T$`D(L$p($vW5( (AX(TAX3A\D(d$@fD/D(\$P($tYD($D(t$ Hİ[@SHP)t$@H)|$0(D)D$ D(K(A(Y d'YY(YYXWf.w WQ(H \EX(HAH A^YX((X^f/r( Y^((t$@(|$0D(D$ HP[HHW)|$ f.(zt@H )t$0HAH 5((\5Wj'^(.(t$0(|$ HHHHX D@UVWATAUAVAWH)pB)xA;D)@ED)HDLD)T$pAD)\$`ODA+fAnD;fnD)d$PD)l$@D)t$0D)|$ ED%O%DMɋEA+ƋHfEn^fnЍCfAnEYA(\DYYEXYfn^WAXf.wfDQ ( D(WAF*AGWW*ȍCE(DY5,&*DX5&Y^,AW+*s(CW*aXA+W*J(EWXA+*/DY%XEX$A(#E;EWAALD*EXfA/wD(D=W$EWDHGHHGD(HA\AYA^AXfD/wfA/s,AGW*y(EW*g(AWXC*N(A+XWA*1X$\A(A\AY\Z#f/s/A(\AYfA/A(Xf/;$L$A(sA({NE(C+E(KE;E(SE([MI[XE(cE(kD(t$0D(|$ IA_A^A]A\_^]H(MA8HI H(@SEHALALtA@McPLHcL#IcJHCHHCDtDH#LL3I[ffH; IuHfuHH(t9t(tt H(vGH(IH(MH(H\$Ht$H|$ AVH HL3m؈D$@@=!tO HH D u)Ut HjH [ @2@u?HH8t$HtLƺIHL Aݳ3H\$0Ht$8H|$HH A^ùH\$WH0@ 3H\$@H0_ȉS؈D$ = u6g3@ j룹4HHX L@PHHVWAVH@ILu93Bw@H0!HuX؉\$0LƋI؉D$0LƋI ؉D$0u6u2L3IHH HtL3Iatuj Tj rj j j j j j j k k a` multiply the output of `random_sample` by `(b-a)` and add `a`:: (b - a) * random_sample() + a .. note:: New code should use the `~numpy.random.Generator.random` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray of floats Array of random floats of shape `size` (unless ``size=None``, in which case a single float is returned). See Also -------- random.Generator.random: which should be used for new code. Examples -------- >>> np.random.random_sample() 0.47108547995356098 # random >>> type(np.random.random_sample()) >>> np.random.random_sample((5,)) array([ 0.30220482, 0.86820401, 0.1654503 , 0.11659149, 0.54323428]) # random Three-by-two array of random numbers from [-5, 0): >>> 5 * np.random.random_sample((3, 2)) - 5 array([[-3.99149989, -0.52338984], # random [-2.99091858, -0.79479508], [-1.23204345, -1.75224494]]) random_sample random(size=None) Return random floats in the half-open interval [0.0, 1.0). Alias for `random_sample` to ease forward-porting to the new random API. random binomial(n, p, size=None) Draw samples from a binomial distribution. Samples are drawn from a binomial distribution with specified parameters, n trials and p probability of success where n an integer >= 0 and p is in the interval [0,1]. (n may be input as a float, but it is truncated to an integer in use) .. note:: New code should use the `~numpy.random.Generator.binomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- n : int or array_like of ints Parameter of the distribution, >= 0. Floats are also accepted, but they will be truncated to integers. p : float or array_like of floats Parameter of the distribution, >= 0 and <=1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``n`` and ``p`` are both scalars. Otherwise, ``np.broadcast(n, p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized binomial distribution, where each sample is equal to the number of successes over the n trials. See Also -------- scipy.stats.binom : probability density function, distribution or cumulative density function, etc. random.Generator.binomial: which should be used for new code. Notes ----- The probability mass function (PMF) for the binomial distribution is .. math:: P(N) = \binom{n}{N}p^N(1-p)^{n-N}, where :math:`n` is the number of trials, :math:`p` is the probability of success, and :math:`N` is the number of successes. When estimating the standard error of a proportion in a population by using a random sample, the normal distribution works well unless the product p*n <=5, where p = population proportion estimate, and n = number of samples, in which case the binomial distribution is used instead. For example, a sample of 15 people shows 4 who are left handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4, so the binomial distribution should be used in this case. References ---------- .. [1] Dalgaard, Peter, "Introductory Statistics with R", Springer-Verlag, 2002. .. [2] Glantz, Stanton A. "Primer of Biostatistics.", McGraw-Hill, Fifth Edition, 2002. .. [3] Lentner, Marvin, "Elementary Applied Statistics", Bogden and Quigley, 1972. .. [4] Weisstein, Eric W. "Binomial Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/BinomialDistribution.html .. [5] Wikipedia, "Binomial distribution", https://en.wikipedia.org/wiki/Binomial_distribution Examples -------- Draw samples from the distribution: >>> n, p = 10, .5 # number of trials, probability of each trial >>> s = np.random.binomial(n, p, 1000) # result of flipping a coin 10 times, tested 1000 times. A real world example. A company drills 9 wild-cat oil exploration wells, each with an estimated probability of success of 0.1. All nine wells fail. What is the probability of that happening? Let's do 20,000 trials of the model, and count the number that generate zero positive results. >>> sum(np.random.binomial(9, 0.1, 20000) == 0)/20000. # answer = 0.38885, or 38%. binomial beta(a, b, size=None) Draw samples from a Beta distribution. The Beta distribution is a special case of the Dirichlet distribution, and is related to the Gamma distribution. It has the probability distribution function .. math:: f(x; a,b) = \frac{1}{B(\alpha, \beta)} x^{\alpha - 1} (1 - x)^{\beta - 1}, where the normalization, B, is the beta function, .. math:: B(\alpha, \beta) = \int_0^1 t^{\alpha - 1} (1 - t)^{\beta - 1} dt. It is often seen in Bayesian inference and order statistics. .. note:: New code should use the `~numpy.random.Generator.beta` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Alpha, positive (>0). b : float or array_like of floats Beta, positive (>0). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` and ``b`` are both scalars. Otherwise, ``np.broadcast(a, b).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized beta distribution. See Also -------- random.Generator.beta: which should be used for new code. beta exponential(scale=1.0, size=None) Draw samples from an exponential distribution. Its probability density function is .. math:: f(x; \frac{1}{\beta}) = \frac{1}{\beta} \exp(-\frac{x}{\beta}), for ``x > 0`` and 0 elsewhere. :math:`\beta` is the scale parameter, which is the inverse of the rate parameter :math:`\lambda = 1/\beta`. The rate parameter is an alternative, widely used parameterization of the exponential distribution [3]_. The exponential distribution is a continuous analogue of the geometric distribution. It describes many common situations, such as the size of raindrops measured over many rainstorms [1]_, or the time between page requests to Wikipedia [2]_. .. note:: New code should use the `~numpy.random.Generator.exponential` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- scale : float or array_like of floats The scale parameter, :math:`\beta = 1/\lambda`. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``scale`` is a scalar. Otherwise, ``np.array(scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized exponential distribution. Examples -------- A real world example: Assume a company has 10000 customer support agents and the average time between customer calls is 4 minutes. >>> n = 10000 >>> time_between_calls = np.random.default_rng().exponential(scale=4, size=n) What is the probability that a customer will call in the next 4 to 5 minutes? >>> x = ((time_between_calls < 5).sum())/n >>> y = ((time_between_calls < 4).sum())/n >>> x-y 0.08 # may vary See Also -------- random.Generator.exponential: which should be used for new code. References ---------- .. [1] Peyton Z. Peebles Jr., "Probability, Random Variables and Random Signal Principles", 4th ed, 2001, p. 57. .. [2] Wikipedia, "Poisson process", https://en.wikipedia.org/wiki/Poisson_process .. [3] Wikipedia, "Exponential distribution", https://en.wikipedia.org/wiki/Exponential_distribution exponential standard_exponential(size=None) Draw samples from the standard exponential distribution. `standard_exponential` is identical to the exponential distribution with a scale parameter of 1. .. note:: New code should use the `~numpy.random.Generator.standard_exponential` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray Drawn samples. See Also -------- random.Generator.standard_exponential: which should be used for new code. Examples -------- Output a 3x8000 array: >>> n = np.random.standard_exponential((3, 8000)) standard_exponential tomaxint(size=None) Return a sample of uniformly distributed random integers in the interval [0, ``np.iinfo("long").max``]. .. warning:: This function uses the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : ndarray Drawn samples, with shape `size`. See Also -------- randint : Uniform sampling over a given half-open interval of integers. random_integers : Uniform sampling over a given closed interval of integers. Examples -------- >>> rs = np.random.RandomState() # need a RandomState object >>> rs.tomaxint((2,2,2)) array([[[1170048599, 1600360186], # random [ 739731006, 1947757578]], [[1871712945, 752307660], [1601631370, 1479324245]]]) >>> rs.tomaxint((2,2,2)) < np.iinfo(np.int_).max array([[[ True, True], [ True, True]], [[ True, True], [ True, True]]]) tomaxint randint(low, high=None, size=None, dtype=int) Return random integers from `low` (inclusive) to `high` (exclusive). Return random integers from the "discrete uniform" distribution of the specified dtype in the "half-open" interval [`low`, `high`). If `high` is None (the default), then results are from [0, `low`). .. note:: New code should use the `~numpy.random.Generator.integers` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- low : int or array-like of ints Lowest (signed) integers to be drawn from the distribution (unless ``high=None``, in which case this parameter is one above the *highest* such integer). high : int or array-like of ints, optional If provided, one above the largest (signed) integer to be drawn from the distribution (see above for behavior if ``high=None``). If array-like, must contain integer values size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. dtype : dtype, optional Desired dtype of the result. Byteorder must be native. The default value is long. .. warning:: This function defaults to the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Which corresponds to `np.intp`. (`dtype=int` is not the same as in most NumPy functions.) Returns ------- out : int or ndarray of ints `size`-shaped array of random integers from the appropriate distribution, or a single such random int if `size` not provided. See Also -------- random_integers : similar to `randint`, only for the closed interval [`low`, `high`], and 1 is the lowest value if `high` is omitted. random.Generator.integers: which should be used for new code. Examples -------- >>> np.random.randint(2, size=10) array([1, 0, 0, 0, 1, 1, 0, 0, 1, 0]) # random >>> np.random.randint(1, size=10) array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0]) Generate a 2 x 4 array of ints between 0 and 4, inclusive: >>> np.random.randint(5, size=(2, 4)) array([[4, 0, 2, 1], # random [3, 2, 2, 0]]) Generate a 1 x 3 array with 3 different upper bounds >>> np.random.randint(1, [3, 5, 10]) array([2, 2, 9]) # random Generate a 1 by 3 array with 3 different lower bounds >>> np.random.randint([1, 5, 7], 10) array([9, 8, 7]) # random Generate a 2 by 4 array using broadcasting with dtype of uint8 >>> np.random.randint([1, 3, 5, 7], [[10], [20]], dtype=np.uint8) array([[ 8, 6, 9, 7], # random [ 1, 16, 9, 12]], dtype=uint8) randint negative_binomial(n, p, size=None) Draw samples from a negative binomial distribution. Samples are drawn from a negative binomial distribution with specified parameters, `n` successes and `p` probability of success where `n` is > 0 and `p` is in the interval [0, 1]. .. note:: New code should use the `~numpy.random.Generator.negative_binomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- n : float or array_like of floats Parameter of the distribution, > 0. p : float or array_like of floats Parameter of the distribution, >= 0 and <=1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``n`` and ``p`` are both scalars. Otherwise, ``np.broadcast(n, p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized negative binomial distribution, where each sample is equal to N, the number of failures that occurred before a total of n successes was reached. .. warning:: This function returns the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. See Also -------- random.Generator.negative_binomial: which should be used for new code. Notes ----- The probability mass function of the negative binomial distribution is .. math:: P(N;n,p) = \frac{\Gamma(N+n)}{N!\Gamma(n)}p^{n}(1-p)^{N}, where :math:`n` is the number of successes, :math:`p` is the probability of success, :math:`N+n` is the number of trials, and :math:`\Gamma` is the gamma function. When :math:`n` is an integer, :math:`\frac{\Gamma(N+n)}{N!\Gamma(n)} = \binom{N+n-1}{N}`, which is the more common form of this term in the pmf. The negative binomial distribution gives the probability of N failures given n successes, with a success on the last trial. If one throws a die repeatedly until the third time a "1" appears, then the probability distribution of the number of non-"1"s that appear before the third "1" is a negative binomial distribution. References ---------- .. [1] Weisstein, Eric W. "Negative Binomial Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/NegativeBinomialDistribution.html .. [2] Wikipedia, "Negative binomial distribution", https://en.wikipedia.org/wiki/Negative_binomial_distribution Examples -------- Draw samples from the distribution: A real world example. A company drills wild-cat oil exploration wells, each with an estimated probability of success of 0.1. What is the probability of having one success for each successive well, that is what is the probability of a single success after drilling 5 wells, after 6 wells, etc.? >>> s = np.random.negative_binomial(1, 0.1, 100000) >>> for i in range(1, 11): # doctest: +SKIP ... probability = sum(s>> import numpy as np >>> s = np.random.poisson(5, 10000) Display histogram of the sample: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 14, density=True) >>> plt.show() Draw each 100 values for lambda 100 and 500: >>> s = np.random.poisson(lam=(100., 500.), size=(100, 2)) poisson zipf(a, size=None) Draw samples from a Zipf distribution. Samples are drawn from a Zipf distribution with specified parameter `a` > 1. The Zipf distribution (also known as the zeta distribution) is a discrete probability distribution that satisfies Zipf's law: the frequency of an item is inversely proportional to its rank in a frequency table. .. note:: New code should use the `~numpy.random.Generator.zipf` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Distribution parameter. Must be greater than 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Zipf distribution. See Also -------- scipy.stats.zipf : probability density function, distribution, or cumulative density function, etc. random.Generator.zipf: which should be used for new code. Notes ----- The probability mass function (PMF) for the Zipf distribution is .. math:: p(k) = \frac{k^{-a}}{\zeta(a)}, for integers :math:`k \geq 1`, where :math:`\zeta` is the Riemann Zeta function. It is named for the American linguist George Kingsley Zipf, who noted that the frequency of any word in a sample of a language is inversely proportional to its rank in the frequency table. References ---------- .. [1] Zipf, G. K., "Selected Studies of the Principle of Relative Frequency in Language," Cambridge, MA: Harvard Univ. Press, 1932. Examples -------- Draw samples from the distribution: >>> a = 4.0 >>> n = 20000 >>> s = np.random.zipf(a, n) Display the histogram of the samples, along with the expected histogram based on the probability density function: >>> import matplotlib.pyplot as plt >>> from scipy.special import zeta # doctest: +SKIP `bincount` provides a fast histogram for small integers. >>> count = np.bincount(s) >>> k = np.arange(1, s.max() + 1) >>> plt.bar(k, count[1:], alpha=0.5, label='sample count') >>> plt.plot(k, n*(k**-a)/zeta(a), 'k.-', alpha=0.5, ... label='expected count') # doctest: +SKIP >>> plt.semilogy() >>> plt.grid(alpha=0.4) >>> plt.legend() >>> plt.title(f'Zipf sample, a={a}, size={n}') >>> plt.show() zipf geometric(p, size=None) Draw samples from the geometric distribution. Bernoulli trials are experiments with one of two outcomes: success or failure (an example of such an experiment is flipping a coin). The geometric distribution models the number of trials that must be run in order to achieve success. It is therefore supported on the positive integers, ``k = 1, 2, ...``. The probability mass function of the geometric distribution is .. math:: f(k) = (1 - p)^{k - 1} p where `p` is the probability of success of an individual trial. .. note:: New code should use the `~numpy.random.Generator.geometric` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- p : float or array_like of floats The probability of success of an individual trial. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``p`` is a scalar. Otherwise, ``np.array(p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized geometric distribution. See Also -------- random.Generator.geometric: which should be used for new code. Examples -------- Draw ten thousand values from the geometric distribution, with the probability of an individual success equal to 0.35: >>> z = np.random.geometric(p=0.35, size=10000) How many trials succeeded after a single run? >>> (z == 1).sum() / 10000. 0.34889999999999999 #random geometric hypergeometric(ngood, nbad, nsample, size=None) Draw samples from a Hypergeometric distribution. Samples are drawn from a hypergeometric distribution with specified parameters, `ngood` (ways to make a good selection), `nbad` (ways to make a bad selection), and `nsample` (number of items sampled, which is less than or equal to the sum ``ngood + nbad``). .. note:: New code should use the `~numpy.random.Generator.hypergeometric` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- ngood : int or array_like of ints Number of ways to make a good selection. Must be nonnegative. nbad : int or array_like of ints Number of ways to make a bad selection. Must be nonnegative. nsample : int or array_like of ints Number of items sampled. Must be at least 1 and at most ``ngood + nbad``. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if `ngood`, `nbad`, and `nsample` are all scalars. Otherwise, ``np.broadcast(ngood, nbad, nsample).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized hypergeometric distribution. Each sample is the number of good items within a randomly selected subset of size `nsample` taken from a set of `ngood` good items and `nbad` bad items. See Also -------- scipy.stats.hypergeom : probability density function, distribution or cumulative density function, etc. random.Generator.hypergeometric: which should be used for new code. Notes ----- The probability mass function (PMF) for the Hypergeometric distribution is .. math:: P(x) = \frac{\binom{g}{x}\binom{b}{n-x}}{\binom{g+b}{n}}, where :math:`0 \le x \le n` and :math:`n-b \le x \le g` for P(x) the probability of ``x`` good results in the drawn sample, g = `ngood`, b = `nbad`, and n = `nsample`. Consider an urn with black and white marbles in it, `ngood` of them are black and `nbad` are white. If you draw `nsample` balls without replacement, then the hypergeometric distribution describes the distribution of black balls in the drawn sample. Note that this distribution is very similar to the binomial distribution, except that in this case, samples are drawn without replacement, whereas in the Binomial case samples are drawn with replacement (or the sample space is infinite). As the sample space becomes large, this distribution approaches the binomial. References ---------- .. [1] Lentner, Marvin, "Elementary Applied Statistics", Bogden and Quigley, 1972. .. [2] Weisstein, Eric W. "Hypergeometric Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/HypergeometricDistribution.html .. [3] Wikipedia, "Hypergeometric distribution", https://en.wikipedia.org/wiki/Hypergeometric_distribution Examples -------- Draw samples from the distribution: >>> ngood, nbad, nsamp = 100, 2, 10 # number of good, number of bad, and number of samples >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000) >>> from matplotlib.pyplot import hist >>> hist(s) # note that it is very unlikely to grab both bad items Suppose you have an urn with 15 white and 15 black marbles. If you pull 15 marbles at random, how likely is it that 12 or more of them are one color? >>> s = np.random.hypergeometric(15, 15, 15, 100000) >>> sum(s>=12)/100000. + sum(s<=3)/100000. # answer = 0.003 ... pretty unlikely! hypergeometric bytes(length) Return random bytes. .. note:: New code should use the `~numpy.random.Generator.bytes` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- length : int Number of random bytes. Returns ------- out : bytes String of length `length`. See Also -------- random.Generator.bytes: which should be used for new code. Examples -------- >>> np.random.bytes(10) b' eh\x85\x022SZ\xbf\xa4' #random bytes 0 8Q A e`H& S `@R `p> > `HQ`$ T` 0S `l`i`l`Hr`W`i`v! X`Xo j``" e`p j`" m`# T`m Y`p k`" k`e`q xo`' xh`hr i`v! 0k`" Hk`(" m`أ$ P `Ȯ, r`* !`- n`H% e`q e`q m`p$ h`r @d`p W`o h`r Xd`q f`(q`( n`% i`w! X`n `k`h`r Hq`( !`(, o`p' hq`H( m`p# i`r 0m`# h`s j`0w! X`o h`(s 0X`o p U`pO `> 0`5 X2 PQ `! @S `PR `0`P0 T0 V0 X0 \0 `0`0`q 0 m 8& 5 0`0`;`0`;`O `d1`;`N `Q ``R `0`T=`l1`0`;`f N`\=`f`; t1`0`|1`N`O `0 U`V`d=`1`0`N`;`@ `( N`N`0`PS `4 pQ `6 l=`  ; `U`f`0`V`1`(U`0`% 3 t=` 0`d0`;`;`;`Q ``wb N 0`|=`h0`;`N`;`pR `= O `< `0` <`1 ``S `/ $0`9 pS `<`<`R `$<`N`,<`4<`=`1 `<<`G D<`l0 Q `S{ 1`d`Q ` u O `N`=`P[! Q `1`=`U`n{ (0`p0`1`=`t0`L<`T`T<`\<`d<`1`1`=` T`Q `Q O `l<`O `O `R `x0`1`O `,0`00`t<`1`=`40` 1`N`Ѓ 1`Q R =`T U`=`1`R `|<`=`P `1`1`R `P `q Q ` R `P 1`1`<`1 `=`X`2`# 8p =`S `<` 2`2`R = `1`N `|0`=`S `P{ i`ж 80`<` P `=` 2 `,2`42`d` 1`0T `<` g 0`<`<2`j`p @T ` >`  0`O`<` ( H# PT ``T i`Xn >` >`<`<`<` >`0P `@P `(>` 1`<0`<`pn <`0>` D2`S ` 8>`$1`O`p& V`(1`PP `,1`<` p# 8o Hs L2`01`<`(r R``P `<`T2`@>`q O`" <`0( H>`V`/ 8U`` O`pT `X`\2`<`=`pP `: P>`P `(O`41`0O`S `81`T `P ` =`d2`@0`X>`=`T `l2` w :`d`P `R `D@ P `=`8O` ;`$=` R ` ;`;` ; `,;`0`R `V`PB j`6 U` V`z S ` E ,=`hs V! `>`@O`P `<1`V V @1`0`4;`P;`X; `HO``1`P `W PO`S `] d;`D0`0 h>`p>`x>`4=`XO`Ph+ W`>`V`>`H0``O`Q `<=`Q `v l;`k t;`(Q `hO` 0R `L0`/ D=`xi L=`|;`p choice(a, size=None, replace=True, p=None) Generates a random sample from a given 1-D array .. note:: New code should use the `~numpy.random.Generator.choice` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. .. warning:: This function uses the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- a : 1-D array-like or int If an ndarray, a random sample is generated from its elements. If an int, the random sample is generated as if it were ``np.arange(a)`` size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. replace : boolean, optional Whether the sample is with or without replacement. Default is True, meaning that a value of ``a`` can be selected multiple times. p : 1-D array-like, optional The probabilities associated with each entry in a. If not given, the sample assumes a uniform distribution over all entries in ``a``. Returns ------- samples : single item or ndarray The generated random samples Raises ------ ValueError If a is an int and less than zero, if a or p are not 1-dimensional, if a is an array-like of size 0, if p is not a vector of probabilities, if a and p have different lengths, or if replace=False and the sample size is greater than the population size See Also -------- randint, shuffle, permutation random.Generator.choice: which should be used in new code Notes ----- Setting user-specified probabilities through ``p`` uses a more general but less efficient sampler than the default. The general sampler produces a different sample than the optimized sampler even if each element of ``p`` is 1 / len(a). Sampling random rows from a 2-D array is not possible with this function, but is possible with `Generator.choice` through its ``axis`` keyword. Examples -------- Generate a uniform random sample from np.arange(5) of size 3: >>> np.random.choice(5, 3) array([0, 3, 4]) # random >>> #This is equivalent to np.random.randint(0,5,3) Generate a non-uniform random sample from np.arange(5) of size 3: >>> np.random.choice(5, 3, p=[0.1, 0, 0.3, 0.6, 0]) array([3, 3, 0]) # random Generate a uniform random sample from np.arange(5) of size 3 without replacement: >>> np.random.choice(5, 3, replace=False) array([3,1,0]) # random >>> #This is equivalent to np.random.permutation(np.arange(5))[:3] Generate a non-uniform random sample from np.arange(5) of size 3 without replacement: >>> np.random.choice(5, 3, replace=False, p=[0.1, 0, 0.3, 0.6, 0]) array([2, 3, 0]) # random Any of the above can be repeated with an arbitrary array-like instead of just integers. For instance: >>> aa_milne_arr = ['pooh', 'rabbit', 'piglet', 'Christopher'] >>> np.random.choice(aa_milne_arr, 5, p=[0.5, 0.1, 0.1, 0.3]) array(['pooh', 'pooh', 'pooh', 'Christopher', 'piglet'], # random dtype='>> a = .6 >>> s = np.random.logseries(a, 10000) >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s) # plot against distribution >>> def logseries(k, p): ... return -p**k/(k*np.log(1-p)) >>> plt.plot(bins, logseries(bins, a)*count.max()/ ... logseries(bins, a).max(), 'r') >>> plt.show() logseries multivariate_normal(mean, cov, size=None, check_valid='warn', tol=1e-8) Draw random samples from a multivariate normal distribution. The multivariate normal, multinormal or Gaussian distribution is a generalization of the one-dimensional normal distribution to higher dimensions. Such a distribution is specified by its mean and covariance matrix. These parameters are analogous to the mean (average or "center") and variance (standard deviation, or "width," squared) of the one-dimensional normal distribution. .. note:: New code should use the `~numpy.random.Generator.multivariate_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : 1-D array_like, of length N Mean of the N-dimensional distribution. cov : 2-D array_like, of shape (N, N) Covariance matrix of the distribution. It must be symmetric and positive-semidefinite for proper sampling. size : int or tuple of ints, optional Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are generated, and packed in an `m`-by-`n`-by-`k` arrangement. Because each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``. If no shape is specified, a single (`N`-D) sample is returned. check_valid : { 'warn', 'raise', 'ignore' }, optional Behavior when the covariance matrix is not positive semidefinite. tol : float, optional Tolerance when checking the singular values in covariance matrix. cov is cast to double before the check. Returns ------- out : ndarray The drawn samples, of shape *size*, if that was provided. If not, the shape is ``(N,)``. In other words, each entry ``out[i,j,...,:]`` is an N-dimensional value drawn from the distribution. See Also -------- random.Generator.multivariate_normal: which should be used for new code. Notes ----- The mean is a coordinate in N-dimensional space, which represents the location where samples are most likely to be generated. This is analogous to the peak of the bell curve for the one-dimensional or univariate normal distribution. Covariance indicates the level to which two variables vary together. From the multivariate normal distribution, we draw N-dimensional samples, :math:`X = [x_1, x_2, ... x_N]`. The covariance matrix element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`. The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its "spread"). Instead of specifying the full covariance matrix, popular approximations include: - Spherical covariance (`cov` is a multiple of the identity matrix) - Diagonal covariance (`cov` has non-negative elements, and only on the diagonal) This geometrical property can be seen in two dimensions by plotting generated data-points: >>> mean = [0, 0] >>> cov = [[1, 0], [0, 100]] # diagonal covariance Diagonal covariance means that points are oriented along x or y-axis: >>> import matplotlib.pyplot as plt >>> x, y = np.random.multivariate_normal(mean, cov, 5000).T >>> plt.plot(x, y, 'x') >>> plt.axis('equal') >>> plt.show() Note that the covariance matrix must be positive semidefinite (a.k.a. nonnegative-definite). Otherwise, the behavior of this method is undefined and backwards compatibility is not guaranteed. References ---------- .. [1] Papoulis, A., "Probability, Random Variables, and Stochastic Processes," 3rd ed., New York: McGraw-Hill, 1991. .. [2] Duda, R. O., Hart, P. E., and Stork, D. G., "Pattern Classification," 2nd ed., New York: Wiley, 2001. Examples -------- >>> mean = (1, 2) >>> cov = [[1, 0], [0, 1]] >>> x = np.random.multivariate_normal(mean, cov, (3, 3)) >>> x.shape (3, 3, 2) Here we generate 800 samples from the bivariate normal distribution with mean [0, 0] and covariance matrix [[6, -3], [-3, 3.5]]. The expected variances of the first and second components of the sample are 6 and 3.5, respectively, and the expected correlation coefficient is -3/sqrt(6*3.5) ≈ -0.65465. >>> cov = np.array([[6, -3], [-3, 3.5]]) >>> pts = np.random.multivariate_normal([0, 0], cov, size=800) Check that the mean, covariance, and correlation coefficient of the sample are close to the expected values: >>> pts.mean(axis=0) array([ 0.0326911 , -0.01280782]) # may vary >>> np.cov(pts.T) array([[ 5.96202397, -2.85602287], [-2.85602287, 3.47613949]]) # may vary >>> np.corrcoef(pts.T)[0, 1] -0.6273591314603949 # may vary We can visualize this data with a scatter plot. The orientation of the point cloud illustrates the negative correlation of the components of this sample. >>> import matplotlib.pyplot as plt >>> plt.plot(pts[:, 0], pts[:, 1], '.', alpha=0.5) >>> plt.axis('equal') >>> plt.grid() >>> plt.show() multivariate_normal multinomial(n, pvals, size=None) Draw samples from a multinomial distribution. The multinomial distribution is a multivariate generalization of the binomial distribution. Take an experiment with one of ``p`` possible outcomes. An example of such an experiment is throwing a dice, where the outcome can be 1 through 6. Each sample drawn from the distribution represents `n` such experiments. Its values, ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome was ``i``. .. note:: New code should use the `~numpy.random.Generator.multinomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. .. warning:: This function defaults to the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- n : int Number of experiments. pvals : sequence of floats, length p Probabilities of each of the ``p`` different outcomes. These must sum to 1 (however, the last element is always assumed to account for the remaining probability, as long as ``sum(pvals[:-1]) <= 1)``. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : ndarray The drawn samples, of shape *size*, if that was provided. If not, the shape is ``(N,)``. In other words, each entry ``out[i,j,...,:]`` is an N-dimensional value drawn from the distribution. See Also -------- random.Generator.multinomial: which should be used for new code. Examples -------- Throw a dice 20 times: >>> np.random.multinomial(20, [1/6.]*6, size=1) array([[4, 1, 7, 5, 2, 1]]) # random It landed 4 times on 1, once on 2, etc. Now, throw the dice 20 times, and 20 times again: >>> np.random.multinomial(20, [1/6.]*6, size=2) array([[3, 4, 3, 3, 4, 3], # random [2, 4, 3, 4, 0, 7]]) For the first run, we threw 3 times 1, 4 times 2, etc. For the second, we threw 2 times 1, 4 times 2, etc. A loaded die is more likely to land on number 6: >>> np.random.multinomial(100, [1/7.]*5 + [2/7.]) array([11, 16, 14, 17, 16, 26]) # random The probability inputs should be normalized. As an implementation detail, the value of the last entry is ignored and assumed to take up any leftover probability mass, but this should not be relied on. A biased coin which has twice as much weight on one side as on the other should be sampled like so: >>> np.random.multinomial(100, [1.0 / 3, 2.0 / 3]) # RIGHT array([38, 62]) # random not like: >>> np.random.multinomial(100, [1.0, 2.0]) # WRONG Traceback (most recent call last): ValueError: pvals < 0, pvals > 1 or pvals contains NaNs multinomial dirichlet(alpha, size=None) Draw samples from the Dirichlet distribution. Draw `size` samples of dimension k from a Dirichlet distribution. A Dirichlet-distributed random variable can be seen as a multivariate generalization of a Beta distribution. The Dirichlet distribution is a conjugate prior of a multinomial distribution in Bayesian inference. .. note:: New code should use the `~numpy.random.Generator.dirichlet` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- alpha : sequence of floats, length k Parameter of the distribution (length ``k`` for sample of length ``k``). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n)``, then ``m * n * k`` samples are drawn. Default is None, in which case a vector of length ``k`` is returned. Returns ------- samples : ndarray, The drawn samples, of shape ``(size, k)``. Raises ------ ValueError If any value in ``alpha`` is less than or equal to zero See Also -------- random.Generator.dirichlet: which should be used for new code. Notes ----- The Dirichlet distribution is a distribution over vectors :math:`x` that fulfil the conditions :math:`x_i>0` and :math:`\sum_{i=1}^k x_i = 1`. The probability density function :math:`p` of a Dirichlet-distributed random vector :math:`X` is proportional to .. math:: p(x) \propto \prod_{i=1}^{k}{x^{\alpha_i-1}_i}, where :math:`\alpha` is a vector containing the positive concentration parameters. The method uses the following property for computation: let :math:`Y` be a random vector which has components that follow a standard gamma distribution, then :math:`X = \frac{1}{\sum_{i=1}^k{Y_i}} Y` is Dirichlet-distributed References ---------- .. [1] David McKay, "Information Theory, Inference and Learning Algorithms," chapter 23, https://www.inference.org.uk/mackay/itila/ .. [2] Wikipedia, "Dirichlet distribution", https://en.wikipedia.org/wiki/Dirichlet_distribution Examples -------- Taking an example cited in Wikipedia, this distribution can be used if one wanted to cut strings (each of initial length 1.0) into K pieces with different lengths, where each piece had, on average, a designated average length, but allowing some variation in the relative sizes of the pieces. >>> s = np.random.dirichlet((10, 5, 3), 20).transpose() >>> import matplotlib.pyplot as plt >>> plt.barh(range(20), s[0]) >>> plt.barh(range(20), s[1], left=s[0], color='g') >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r') >>> plt.title("Lengths of Strings") dirichlet0H`xnumpy/random/mtrand.pyxcpython/contextvars.pxdnumpy/__init__.cython-30.pxdcpython/type.pxdcpython/bool.pxdcpython/complex.pxdnumpy/random/bit_generator.pxd uniform(low=0.0, high=1.0, size=None) Draw samples from a uniform distribution. Samples are uniformly distributed over the half-open interval ``[low, high)`` (includes low, but excludes high). In other words, any value within the given interval is equally likely to be drawn by `uniform`. .. note:: New code should use the `~numpy.random.Generator.uniform` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- low : float or array_like of floats, optional Lower boundary of the output interval. All values generated will be greater than or equal to low. The default value is 0. high : float or array_like of floats Upper boundary of the output interval. All values generated will be less than or equal to high. The high limit may be included in the returned array of floats due to floating-point rounding in the equation ``low + (high-low) * random_sample()``. The default value is 1.0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``low`` and ``high`` are both scalars. Otherwise, ``np.broadcast(low, high).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized uniform distribution. See Also -------- randint : Discrete uniform distribution, yielding integers. random_integers : Discrete uniform distribution over the closed interval ``[low, high]``. random_sample : Floats uniformly distributed over ``[0, 1)``. random : Alias for `random_sample`. rand : Convenience function that accepts dimensions as input, e.g., ``rand(2,2)`` would generate a 2-by-2 array of floats, uniformly distributed over ``[0, 1)``. random.Generator.uniform: which should be used for new code. Notes ----- The probability density function of the uniform distribution is .. math:: p(x) = \frac{1}{b - a} anywhere within the interval ``[a, b)``, and zero elsewhere. When ``high`` == ``low``, values of ``low`` will be returned. If ``high`` < ``low``, the results are officially undefined and may eventually raise an error, i.e. do not rely on this function to behave when passed arguments satisfying that inequality condition. The ``high`` limit may be included in the returned array of floats due to floating-point rounding in the equation ``low + (high-low) * random_sample()``. For example: >>> x = np.float32(5*0.99999999) >>> x np.float32(5.0) Examples -------- Draw samples from the distribution: >>> s = np.random.uniform(-1,0,1000) All values are within the given interval: >>> np.all(s >= -1) True >>> np.all(s < 0) True Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 15, density=True) >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r') >>> plt.show() uniform__module___cython_3_1_2._common_types_metatype shuffle(x) Modify a sequence in-place by shuffling its contents. This function only shuffles the array along the first axis of a multi-dimensional array. The order of sub-arrays is changed but their contents remains the same. .. note:: New code should use the `~numpy.random.Generator.shuffle` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- x : ndarray or MutableSequence The array, list or mutable sequence to be shuffled. Returns ------- None See Also -------- random.Generator.shuffle: which should be used for new code. Examples -------- >>> arr = np.arange(10) >>> np.random.shuffle(arr) >>> arr [1 7 5 2 9 4 3 6 0 8] # random Multi-dimensional arrays are only shuffled along the first axis: >>> arr = np.arange(9).reshape((3, 3)) >>> np.random.shuffle(arr) >>> arr array([[3, 4, 5], # random [6, 7, 8], [0, 1, 2]]) shuffle rand(d0, d1, ..., dn) Random values in a given shape. .. note:: This is a convenience function for users porting code from Matlab, and wraps `random_sample`. That function takes a tuple to specify the size of the output, which is consistent with other NumPy functions like `numpy.zeros` and `numpy.ones`. Create an array of the given shape and populate it with random samples from a uniform distribution over ``[0, 1)``. Parameters ---------- d0, d1, ..., dn : int, optional The dimensions of the returned array, must be non-negative. If no argument is given a single Python float is returned. Returns ------- out : ndarray, shape ``(d0, d1, ..., dn)`` Random values. See Also -------- random Examples -------- >>> np.random.rand(3,2) array([[ 0.14022471, 0.96360618], #random [ 0.37601032, 0.25528411], #random [ 0.49313049, 0.94909878]]) #random rand randn(d0, d1, ..., dn) Return a sample (or samples) from the "standard normal" distribution. .. note:: This is a convenience function for users porting code from Matlab, and wraps `standard_normal`. That function takes a tuple to specify the size of the output, which is consistent with other NumPy functions like `numpy.zeros` and `numpy.ones`. .. note:: New code should use the `~numpy.random.Generator.standard_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. If positive int_like arguments are provided, `randn` generates an array of shape ``(d0, d1, ..., dn)``, filled with random floats sampled from a univariate "normal" (Gaussian) distribution of mean 0 and variance 1. A single float randomly sampled from the distribution is returned if no argument is provided. Parameters ---------- d0, d1, ..., dn : int, optional The dimensions of the returned array, must be non-negative. If no argument is given a single Python float is returned. Returns ------- Z : ndarray or float A ``(d0, d1, ..., dn)``-shaped array of floating-point samples from the standard normal distribution, or a single such float if no parameters were supplied. See Also -------- standard_normal : Similar, but takes a tuple as its argument. normal : Also accepts mu and sigma arguments. random.Generator.standard_normal: which should be used for new code. Notes ----- For random samples from the normal distribution with mean ``mu`` and standard deviation ``sigma``, use:: sigma * np.random.randn(...) + mu Examples -------- >>> np.random.randn() 2.1923875335537315 # random Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> 3 + 2.5 * np.random.randn(2, 4) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random randnfunc_doc__doc__func_name__name____qualname__func_dict__dict__func_globals__globals__func_closure__closure__func_code__code__func_defaults__defaults____kwdefaults____annotations___is_coroutine random_integers(low, high=None, size=None) Random integers of type `numpy.int_` between `low` and `high`, inclusive. Return random integers of type `numpy.int_` from the "discrete uniform" distribution in the closed interval [`low`, `high`]. If `high` is None (the default), then results are from [1, `low`]. The `numpy.int_` type translates to the C long integer type and its precision is platform dependent. This function has been deprecated. Use randint instead. .. deprecated:: 1.11.0 Parameters ---------- low : int Lowest (signed) integer to be drawn from the distribution (unless ``high=None``, in which case this parameter is the *highest* such integer). high : int, optional If provided, the largest (signed) integer to be drawn from the distribution (see above for behavior if ``high=None``). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : int or ndarray of ints `size`-shaped array of random integers from the appropriate distribution, or a single such random int if `size` not provided. See Also -------- randint : Similar to `random_integers`, only for the half-open interval [`low`, `high`), and 0 is the lowest value if `high` is omitted. Notes ----- To sample from N evenly spaced floating-point numbers between a and b, use:: a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.) Examples -------- >>> np.random.random_integers(5) 4 # random >>> type(np.random.random_integers(5)) >>> np.random.random_integers(5, size=(3,2)) array([[5, 4], # random [3, 3], [4, 5]]) Choose five random numbers from the set of five evenly-spaced numbers between 0 and 2.5, inclusive (*i.e.*, from the set :math:`{0, 5/8, 10/8, 15/8, 20/8}`): >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4. array([ 0.625, 1.25 , 0.625, 0.625, 2.5 ]) # random Roll two six sided dice 1000 times and sum the results: >>> d1 = np.random.random_integers(1, 6, 1000) >>> d2 = np.random.random_integers(1, 6, 1000) >>> dsums = d1 + d2 Display results as a histogram: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(dsums, 11, density=True) >>> plt.show() __dictoffset__random_integers__vectorcalloffset____weaklistoffset__(TXabdfijklnpsuvx)., ?_df_dpgcid_initmuninpstsz%qtolalowargsatolbetaboolcopy__exit____func__highint8intpitemleftlesslocklong__main__meanmnixmode__name__nbadndimnoncparrprodrandranfrtolseed standard_normal(size=None) Draw samples from a standard Normal distribution (mean=0, stdev=1). .. note:: New code should use the `~numpy.random.Generator.standard_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray A floating-point array of shape ``size`` of drawn samples, or a single sample if ``size`` was not specified. See Also -------- normal : Equivalent function with additional ``loc`` and ``scale`` arguments for setting the mean and standard deviation. random.Generator.standard_normal: which should be used for new code. Notes ----- For random samples from the normal distribution with mean ``mu`` and standard deviation ``sigma``, use one of:: mu + sigma * np.random.standard_normal(size=...) np.random.normal(mu, sigma, size=...) Examples -------- >>> np.random.standard_normal() 2.1923875335537315 #random >>> s = np.random.standard_normal(8000) >>> s array([ 0.6888893 , 0.78096262, -0.89086505, ..., 0.49876311, # random -0.38672696, -0.4685006 ]) # random >>> s.shape (8000,) >>> s = np.random.standard_normal(size=(3, 4, 2)) >>> s.shape (3, 4, 2) Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> 3 + 2.5 * np.random.standard_normal(size=(2, 4)) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random selfsidesizesort__spec__sqrttakestandard_normaltemp__test__typewaldwarnzipfA Jaqahigh__all__alphaarray at 0xbytes__class__dfdendfnumdiricdtypeempty__enter__equalfinfoflagsfleftfmodefoundgammagaussindexint16int32int64isnankappalnbad_lowmnarrn_arrngoodniternumpyoleftomodeonbadp_arrp_sumpowerpvalsraiserandnrangeravelrightscaleshapesigmastateuint8valuex_ptrzerosarangeastypebitgenchoicecumsumdoubleenablefrightgumbel_highignoreinvacckwargslegacylengthlngood_maskedmethod__module__multinn_uniqnormalobject_offsetongoodorightpareto_pickle_randrandomreducesamplestrideuint16uint32uint64uniqueunsafeAt:QgQ_MT19937 normal(loc=0.0, scale=1.0, size=None) Draw random samples from a normal (Gaussian) distribution. The probability density function of the normal distribution, first derived by De Moivre and 200 years later by both Gauss and Laplace independently [2]_, is often called the bell curve because of its characteristic shape (see the example below). The normal distributions occurs often in nature. For example, it describes the commonly occurring distribution of samples influenced by a large number of tiny, random disturbances, each with its own unique distribution [2]_. .. note:: New code should use the `~numpy.random.Generator.normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats Mean ("centre") of the distribution. scale : float or array_like of floats Standard deviation (spread or "width") of the distribution. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized normal distribution. See Also -------- scipy.stats.norm : probability density function, distribution or cumulative density function, etc. random.Generator.normal: which should be used for new code. Notes ----- The probability density for the Gaussian distribution is .. math:: p(x) = \frac{1}{\sqrt{ 2 \pi \sigma^2 }} e^{ - \frac{ (x - \mu)^2 } {2 \sigma^2} }, where :math:`\mu` is the mean and :math:`\sigma` the standard deviation. The square of the standard deviation, :math:`\sigma^2`, is called the variance. The function has its peak at the mean, and its "spread" increases with the standard deviation (the function reaches 0.607 times its maximum at :math:`x + \sigma` and :math:`x - \sigma` [2]_). This implies that normal is more likely to return samples lying close to the mean, rather than those far away. References ---------- .. [1] Wikipedia, "Normal distribution", https://en.wikipedia.org/wiki/Normal_distribution .. [2] P. R. Peebles Jr., "Central Limit Theorem" in "Probability, Random Variables and Random Signal Principles", 4th ed., 2001, pp. 51, 51, 125. Examples -------- Draw samples from the distribution: >>> mu, sigma = 0, 0.1 # mean and standard deviation >>> s = np.random.normal(mu, sigma, 1000) Verify the mean and the standard deviation: >>> abs(mu - np.mean(s)) 0.0 # may vary >>> abs(sigma - np.std(s, ddof=1)) 0.0 # may vary Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) * ... np.exp( - (bins - mu)**2 / (2 * sigma**2) ), ... linewidth=2, color='r') >>> plt.show() Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> np.random.normal(3, 2.5, size=(2, 4)) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random alpha <= 0asarraybg_typebuf_ptrcapsulecastingdisablenormal_dtypefloat64greaterlaplace_mt19937nsamplepoissonrandintrandomsreplacereshapeshufflestridestobytestotsizeuniformval_arrweibullSequenceallclosebinomial_endpoint__getstate__isfiniteisnativeisscalaritemsizelnsamplelogisticn_uint32onsampleoperatorpop_size__qualname__rayleigh__reduce__reversed__set_name____setstate__subtracttomaxint_cython_3_1_2.cython_function_or_methodval_datavonmiseswarnings + 1) insteadMT19937TypeErroralpha_arrchisquaredirichletgeometricget_statehas_gaussis_scalarisenabledleft > modelognormallogseriesset_statesingletonwriteableIndexErrorValueErroralpha_dataempty_likeflat_foundissubdtypeleft == rightless_equallogical_ormode > right__pyx_vtable__stacklevelstandard_ttriangularImportErrorRandomStateUserWarningcheck_validexponentialfinal_shape!h {!1:!4ztSZZ[method_namemultinomialpermutationresult_typeAoQ $D( 1.0uniform_samplesA~ 3avS 4'q 4'qQ!N t1A'qMt7!aQ[%QP t1A]!4}F$gQ(!9A[RandomState.betaRandomState.randRandomState.seedRandomState.waldRandomState.zipf"!Z t1A(]&G1at1t1may_share_memory__randomstate_ctorAN 3avS 4~Q 4~Qe1RandomState.bytesRandomState.gamma standard_gamma(shape, size=None) Draw samples from a standard Gamma distribution. Samples are drawn from a Gamma distribution with specified parameters, shape (sometimes designated "k") and scale=1. .. note:: New code should use the `~numpy.random.Generator.standard_gamma` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- shape : float or array_like of floats Parameter, must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``shape`` is a scalar. Otherwise, ``np.array(shape).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized standard gamma distribution. See Also -------- scipy.stats.gamma : probability density function, distribution or cumulative density function, etc. random.Generator.standard_gamma: which should be used for new code. Notes ----- The probability density for the Gamma distribution is .. math:: p(x) = x^{k-1}\frac{e^{-x/\theta}}{\theta^k\Gamma(k)}, where :math:`k` is the shape and :math:`\theta` the scale, and :math:`\Gamma` is the Gamma function. The Gamma distribution is often used to model the times to failure of electronic components, and arises naturally in processes for which the waiting times between Poisson distributed events are relevant. References ---------- .. [1] Weisstein, Eric W. "Gamma Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/GammaDistribution.html .. [2] Wikipedia, "Gamma distribution", https://en.wikipedia.org/wiki/Gamma_distribution Examples -------- Draw samples from the distribution: >>> shape, scale = 2., 1. # mean and width >>> s = np.random.standard_gamma(shape, 1000000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> import scipy.special as sps # doctest: +SKIP >>> count, bins, ignored = plt.hist(s, 50, density=True) >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ # doctest: +SKIP ... (sps.gamma(shape) * scale**shape)) >>> plt.plot(bins, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() RandomState.powerRandomState.randnaF t1A%Qd-vTuAQ[get_bit_generatornegative_binomialset_bit_generatorstandard_gammaDeprecationWarningJan t1A_AT6WA9A[ !N t1A)$m6WA9AQQRandomState.choiceRandomState.gumbelRandomState.normalRandomState.paretoRandomState.random+Zqr t1A%Qd*F$gQ9A[a^ t1A]!4}F$gQuAuA[array is read-onlyasyncio.coroutinescline_in_traceback;jb t1A&at:V4wa9A[ngood + nbad < nsample!1^ t1A^1D V4wa9A9A[#1r t1AZqMt7!9A9A[2!b t1A2!4}F$gUVa(![AP t1A^1D V4wauAQ[QD t1A_ATfDquAQ[RandomState.laplaceRandomState.poissonRandomState.randintRandomState.shuffleRandomState.uniformRandomState.weibullmultivariate_normalnumpy.random.mtrand'qh t1A&at:V4waa9A[you are shuffling a 'RandomState.__reduce__RandomState.binomialRandomState.logisticRandomState.rayleighRandomState.tomaxintRandomState.vonmises)X t1A,ATfDq9AQQZqP t1A&at:V4wa9AQ[noncentral_chisquarestandard_exponentialRandomState.chisquareRandomState.dirichletRandomState.geometricRandomState.get_stateRandomState.lognormalRandomState.logseriesRandomState.set_state#1P t1A-Qd-vTPQ[TE[PQ1r d!1(ZvT1e1AA}AQ4A^ t1A*!4}F$gQ9A9A("5Q*AX d!1+1D &G3aqAA}AQJaN t1A_ATfDqQ9AQRandomState.__getstate__RandomState.__setstate__RandomState.standard_tRandomState.triangularA< &WBcBb t81C|5""HG1F("BaRandomState.exponentialRandomState.f (line 1747)RandomState.multinomialRandomState.permutationa must be 1-dimensional b d!1&at:V4wce1AA}AQnumpy/random/mtrand.pyx'p' must be 1-dimensionalRandomState.noncentral_fRandomState.random_sampleRandomState.seed (line 232) n d!1-Qd*F$gSPQe1AA}AQprobabilities contain NaNRandomState.bytes (line 819)RandomState.hypergeometricRandomState.rand (line 1196)RandomState.standard_gammaRandomState.wald (line 3179)RandomState.zipf (line 3694)Range exceeds valid boundsmean must be 1 dimensional+1n d!1.at=d'QTTUe1e1A}AQRandomState.choice (line 855)RandomState.gamma (line 1663)RandomState.power (line 2573)RandomState.randn (line 1240)RandomState.random_integersRandomState.standard_cauchyRandomState.standard_normal'a' and 'p' must have same sizeRandomState.gumbel (line 2776)RandomState.normal (line 1472)RandomState.pareto (line 2366)RandomState.randint (line 688)pvals must be a 1-d sequence_RandomState__randomstate_ctorRandomState.laplace (line 2682)RandomState.negative_binomialRandomState.poisson (line 3611)RandomState.shuffle (line 4564)RandomState.tomaxint (line 625)RandomState.uniform (line 1069)RandomState.weibull (line 2469)probabilities do not sum to 1state dictionary is not valid.( 1` 51 E1,A 1 ! E1)t81EAV2SV6+1` CqN"2! ]#Q!Q 2T"Kq 1 *AQ[ Qa 51 Q qqQfCq&!,a;m1A AM!  "BaaEaqAQau$9!4q:DAQ$b2Q!1EaqARrhar3bBbq5QD ~-Q ]#U*=T ]!6A )1Am1AkaqavYa 2T!5#SBa  !9A+Qe:QWCqTBaq a *AQ 51 Q qqQfCq6'r Ay Qa}AQ QU!3e7!  U!1)!4zQd!9EQTTUUYYZ!qRandomState.binomial (line 3365)RandomState.logistic (line 2900)RandomState.rayleigh (line 3102)RandomState.vonmises (line 2277) dirichlet(alpha, size=None) Draw samples from the Dirichlet distribution. Draw `size` samples of dimension k from a Dirichlet distribution. A Dirichlet-distributed random variable can be seen as a multivariate generalization of a Beta distribution. The Dirichlet distribution is a conjugate prior of a multinomial distribution in Bayesian inference. .. note:: New code should use the `~numpy.random.Generator.dirichlet` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- alpha : sequence of floats, length k Parameter of the distribution (length ``k`` for sample of length ``k``). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n)``, then ``m * n * k`` samples are drawn. Default is None, in which case a vector of length ``k`` is returned. Returns ------- samples : ndarray, The drawn samples, of shape ``(size, k)``. Raises ------ ValueError If any value in ``alpha`` is less than or equal to zero See Also -------- random.Generator.dirichlet: which should be used for new code. Notes ----- The Dirichlet distribution is a distribution over vectors :math:`x` that fulfil the conditions :math:`x_i>0` and :math:`\sum_{i=1}^k x_i = 1`. The probability density function :math:`p` of a Dirichlet-distributed random vector :math:`X` is proportional to .. math:: p(x) \propto \prod_{i=1}^{k}{x^{\alpha_i-1}_i}, where :math:`\alpha` is a vector containing the positive concentration parameters. The method uses the following property for computation: let :math:`Y` be a random vector which has components that follow a standard gamma distribution, then :math:`X = \frac{1}{\sum_{i=1}^k{Y_i}} Y` is Dirichlet-distributed References ---------- .. [1] David McKay, "Information Theory, Inference and Learning Algorithms," chapter 23, https://www.inference.org.uk/mackay/itila/ .. [2] Wikipedia, "Dirichlet distribution", https://en.wikipedia.org/wiki/Dirichlet_distribution Examples -------- Taking an example cited in Wikipedia, this distribution can be used if one wanted to cut strings (each of initial length 1.0) into K pieces with different lengths, where each piece had, on average, a designated average length, but allowing some variation in the relative sizes of the pieces. >>> s = np.random.dirichlet((10, 5, 3), 20).transpose() >>> import matplotlib.pyplot as plt >>> plt.barh(range(20), s[0]) >>> plt.barh(range(20), s[1], left=s[0], color='g') >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r') >>> plt.title("Lengths of Strings") laplace(loc=0.0, scale=1.0, size=None) Draw samples from the Laplace or double exponential distribution with specified location (or mean) and scale (decay). The Laplace distribution is similar to the Gaussian/normal distribution, but is sharper at the peak and has fatter tails. It represents the difference between two independent, identically distributed exponential random variables. .. note:: New code should use the `~numpy.random.Generator.laplace` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional The position, :math:`\mu`, of the distribution peak. Default is 0. scale : float or array_like of floats, optional :math:`\lambda`, the exponential decay. Default is 1. Must be non- negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Laplace distribution. See Also -------- random.Generator.laplace: which should be used for new code. Notes ----- It has the probability density function .. math:: f(x; \mu, \lambda) = \frac{1}{2\lambda} \exp\left(-\frac{|x - \mu|}{\lambda}\right). The first law of Laplace, from 1774, states that the frequency of an error can be expressed as an exponential function of the absolute magnitude of the error, which leads to the Laplace distribution. For many problems in economics and health sciences, this distribution seems to model the data better than the standard Gaussian distribution. References ---------- .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). "Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, 9th printing," New York: Dover, 1972. .. [2] Kotz, Samuel, et. al. "The Laplace Distribution and Generalizations, " Birkhauser, 2001. .. [3] Weisstein, Eric W. "Laplace Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/LaplaceDistribution.html .. [4] Wikipedia, "Laplace distribution", https://en.wikipedia.org/wiki/Laplace_distribution Examples -------- Draw samples from the distribution >>> loc, scale = 0., 1. >>> s = np.random.laplace(loc, scale, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> x = np.arange(-8., 8., .01) >>> pdf = np.exp(-abs(x-loc)/scale)/(2.*scale) >>> plt.plot(x, pdf) Plot Gaussian for comparison: >>> g = (1/(scale * np.sqrt(2 * np.pi)) * ... np.exp(-(x - loc)**2 / (2 * scale**2))) >>> plt.plot(x,g) ' object which is not a subclass of 'Sequence'; `shuffle` is not guaranteed to behave correctly. E.g., non-numpy array/tensor objects with view semantics may contain duplicates after shuffling. permutation(x) Randomly permute a sequence, or return a permuted range. If `x` is a multi-dimensional array, it is only shuffled along its first index. .. note:: New code should use the `~numpy.random.Generator.permutation` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- x : int or array_like If `x` is an integer, randomly permute ``np.arange(x)``. If `x` is an array, make a copy and shuffle the elements randomly. Returns ------- out : ndarray Permuted sequence or array range. See Also -------- random.Generator.permutation: which should be used for new code. Examples -------- >>> np.random.permutation(10) array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6]) # random >>> np.random.permutation([1, 4, 9, 12, 15]) array([15, 1, 9, 4, 12]) # random >>> arr = np.arange(9).reshape((3, 3)) >>> np.random.permutation(arr) array([[6, 7, 8], # random [0, 1, 2], [3, 4, 5]]) poisson(lam=1.0, size=None) Draw samples from a Poisson distribution. The Poisson distribution is the limit of the binomial distribution for large N. .. note:: New code should use the `~numpy.random.Generator.poisson` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- lam : float or array_like of floats Expected number of events occurring in a fixed-time interval, must be >= 0. A sequence must be broadcastable over the requested size. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``lam`` is a scalar. Otherwise, ``np.array(lam).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Poisson distribution. See Also -------- random.Generator.poisson: which should be used for new code. Notes ----- The probability mass function (PMF) of Poisson distribution is .. math:: f(k; \lambda)=\frac{\lambda^k e^{-\lambda}}{k!} For events with an expected separation :math:`\lambda` the Poisson distribution :math:`f(k; \lambda)` describes the probability of :math:`k` events occurring within the observed interval :math:`\lambda`. Because the output is limited to the range of the C int64 type, a ValueError is raised when `lam` is within 10 sigma of the maximum representable value. References ---------- .. [1] Weisstein, Eric W. "Poisson Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/PoissonDistribution.html .. [2] Wikipedia, "Poisson distribution", https://en.wikipedia.org/wiki/Poisson_distribution Examples -------- Draw samples from the distribution: >>> import numpy as np >>> s = np.random.poisson(5, 10000) Display histogram of the sample: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 14, density=True) >>> plt.show() Draw each 100 values for lambda 100 and 500: >>> s = np.random.poisson(lam=(100., 500.), size=(100, 2))  >> np.random.rand(3,2) array([[ 0.14022471, 0.96360618], #random [ 0.37601032, 0.25528411], #random [ 0.49313049, 0.94909878]]) #random randn(d0, d1, ..., dn) Return a sample (or samples) from the "standard normal" distribution. .. note:: This is a convenience function for users porting code from Matlab, and wraps `standard_normal`. That function takes a tuple to specify the size of the output, which is consistent with other NumPy functions like `numpy.zeros` and `numpy.ones`. .. note:: New code should use the `~numpy.random.Generator.standard_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. If positive int_like arguments are provided, `randn` generates an array of shape ``(d0, d1, ..., dn)``, filled with random floats sampled from a univariate "normal" (Gaussian) distribution of mean 0 and variance 1. A single float randomly sampled from the distribution is returned if no argument is provided. Parameters ---------- d0, d1, ..., dn : int, optional The dimensions of the returned array, must be non-negative. If no argument is given a single Python float is returned. Returns ------- Z : ndarray or float A ``(d0, d1, ..., dn)``-shaped array of floating-point samples from the standard normal distribution, or a single such float if no parameters were supplied. See Also -------- standard_normal : Similar, but takes a tuple as its argument. normal : Also accepts mu and sigma arguments. random.Generator.standard_normal: which should be used for new code. Notes ----- For random samples from the normal distribution with mean ``mu`` and standard deviation ``sigma``, use:: sigma * np.random.randn(...) + mu Examples -------- >>> np.random.randn() 2.1923875335537315 # random Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> 3 + 2.5 * np.random.randn(2, 4) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random random_sample(size=None) Return random floats in the half-open interval [0.0, 1.0). Results are from the "continuous uniform" distribution over the stated interval. To sample :math:`Unif[a, b), b > a` multiply the output of `random_sample` by `(b-a)` and add `a`:: (b - a) * random_sample() + a .. note:: New code should use the `~numpy.random.Generator.random` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray of floats Array of random floats of shape `size` (unless ``size=None``, in which case a single float is returned). See Also -------- random.Generator.random: which should be used for new code. Examples -------- >>> np.random.random_sample() 0.47108547995356098 # random >>> type(np.random.random_sample()) >>> np.random.random_sample((5,)) array([ 0.30220482, 0.86820401, 0.1654503 , 0.11659149, 0.54323428]) # random Three-by-two array of random numbers from [-5, 0): >>> 5 * np.random.random_sample((3, 2)) - 5 array([[-3.99149989, -0.52338984], # random [-2.99091858, -0.79479508], [-1.23204345, -1.75224494]]) seed(seed=None) Reseed a legacy MT19937 BitGenerator Notes ----- This is a convenience, legacy function. The best practice is to **not** reseed a BitGenerator, rather to recreate a new one. This method is here for legacy reasons. This example demonstrates best practice. >>> from numpy.random import MT19937 >>> from numpy.random import RandomState, SeedSequence >>> rs = RandomState(MT19937(SeedSequence(123456789))) # Later, you want to restart the stream >>> rs = RandomState(MT19937(SeedSequence(987654321))) shuffle(x) Modify a sequence in-place by shuffling its contents. This function only shuffles the array along the first axis of a multi-dimensional array. The order of sub-arrays is changed but their contents remains the same. .. note:: New code should use the `~numpy.random.Generator.shuffle` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- x : ndarray or MutableSequence The array, list or mutable sequence to be shuffled. Returns ------- None See Also -------- random.Generator.shuffle: which should be used for new code. Examples -------- >>> arr = np.arange(10) >>> np.random.shuffle(arr) >>> arr [1 7 5 2 9 4 3 6 0 8] # random Multi-dimensional arrays are only shuffled along the first axis: >>> arr = np.arange(9).reshape((3, 3)) >>> np.random.shuffle(arr) >>> arr array([[3, 4, 5], # random [6, 7, 8], [0, 1, 2]]) standard_cauchy(size=None) Draw samples from a standard Cauchy distribution with mode = 0. Also known as the Lorentz distribution. .. note:: New code should use the `~numpy.random.Generator.standard_cauchy` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- samples : ndarray or scalar The drawn samples. See Also -------- random.Generator.standard_cauchy: which should be used for new code. Notes ----- The probability density function for the full Cauchy distribution is .. math:: P(x; x_0, \gamma) = \frac{1}{\pi \gamma \bigl[ 1+ (\frac{x-x_0}{\gamma})^2 \bigr] } and the Standard Cauchy distribution just sets :math:`x_0=0` and :math:`\gamma=1` The Cauchy distribution arises in the solution to the driven harmonic oscillator problem, and also describes spectral line broadening. It also describes the distribution of values at which a line tilted at a random angle will cut the x axis. When studying hypothesis tests that assume normality, seeing how the tests perform on data from a Cauchy distribution is a good indicator of their sensitivity to a heavy-tailed distribution, since the Cauchy looks very much like a Gaussian distribution, but with heavier tails. References ---------- .. [1] NIST/SEMATECH e-Handbook of Statistical Methods, "Cauchy Distribution", https://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm .. [2] Weisstein, Eric W. "Cauchy Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/CauchyDistribution.html .. [3] Wikipedia, "Cauchy distribution" https://en.wikipedia.org/wiki/Cauchy_distribution Examples -------- Draw samples and plot the distribution: >>> import matplotlib.pyplot as plt >>> s = np.random.standard_cauchy(1000000) >>> s = s[(s>-25) & (s<25)] # truncate distribution so it plots well >>> plt.hist(s, bins=100) >>> plt.show() gamma(shape, scale=1.0, size=None) Draw samples from a Gamma distribution. Samples are drawn from a Gamma distribution with specified parameters, `shape` (sometimes designated "k") and `scale` (sometimes designated "theta"), where both parameters are > 0. .. note:: New code should use the `~numpy.random.Generator.gamma` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- shape : float or array_like of floats The shape of the gamma distribution. Must be non-negative. scale : float or array_like of floats, optional The scale of the gamma distribution. Must be non-negative. Default is equal to 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``shape`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(shape, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized gamma distribution. See Also -------- scipy.stats.gamma : probability density function, distribution or cumulative density function, etc. random.Generator.gamma: which should be used for new code. Notes ----- The probability density for the Gamma distribution is .. math:: p(x) = x^{k-1}\frac{e^{-x/\theta}}{\theta^k\Gamma(k)}, where :math:`k` is the shape and :math:`\theta` the scale, and :math:`\Gamma` is the Gamma function. The Gamma distribution is often used to model the times to failure of electronic components, and arises naturally in processes for which the waiting times between Poisson distributed events are relevant. References ---------- .. [1] Weisstein, Eric W. "Gamma Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/GammaDistribution.html .. [2] Wikipedia, "Gamma distribution", https://en.wikipedia.org/wiki/Gamma_distribution Examples -------- Draw samples from the distribution: >>> shape, scale = 2., 2. # mean=4, std=2*sqrt(2) >>> s = np.random.gamma(shape, scale, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> import scipy.special as sps # doctest: +SKIP >>> count, bins, ignored = plt.hist(s, 50, density=True) >>> y = bins**(shape-1)*(np.exp(-bins/scale) / # doctest: +SKIP ... (sps.gamma(shape)*scale**shape)) >>> plt.plot(bins, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() standard_exponential(size=None) Draw samples from the standard exponential distribution. `standard_exponential` is identical to the exponential distribution with a scale parameter of 1. .. note:: New code should use the `~numpy.random.Generator.standard_exponential` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray Drawn samples. See Also -------- random.Generator.standard_exponential: which should be used for new code. Examples -------- Output a 3x8000 array: >>> n = np.random.standard_exponential((3, 8000)) standard_gamma(shape, size=None) Draw samples from a standard Gamma distribution. Samples are drawn from a Gamma distribution with specified parameters, shape (sometimes designated "k") and scale=1. .. note:: New code should use the `~numpy.random.Generator.standard_gamma` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- shape : float or array_like of floats Parameter, must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``shape`` is a scalar. Otherwise, ``np.array(shape).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized standard gamma distribution. See Also -------- scipy.stats.gamma : probability density function, distribution or cumulative density function, etc. random.Generator.standard_gamma: which should be used for new code. Notes ----- The probability density for the Gamma distribution is .. math:: p(x) = x^{k-1}\frac{e^{-x/\theta}}{\theta^k\Gamma(k)}, where :math:`k` is the shape and :math:`\theta` the scale, and :math:`\Gamma` is the Gamma function. The Gamma distribution is often used to model the times to failure of electronic components, and arises naturally in processes for which the waiting times between Poisson distributed events are relevant. References ---------- .. [1] Weisstein, Eric W. "Gamma Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/GammaDistribution.html .. [2] Wikipedia, "Gamma distribution", https://en.wikipedia.org/wiki/Gamma_distribution Examples -------- Draw samples from the distribution: >>> shape, scale = 2., 1. # mean and width >>> s = np.random.standard_gamma(shape, 1000000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> import scipy.special as sps # doctest: +SKIP >>> count, bins, ignored = plt.hist(s, 50, density=True) >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ # doctest: +SKIP ... (sps.gamma(shape) * scale**shape)) >>> plt.plot(bins, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() standard_normal(size=None) Draw samples from a standard Normal distribution (mean=0, stdev=1). .. note:: New code should use the `~numpy.random.Generator.standard_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : float or ndarray A floating-point array of shape ``size`` of drawn samples, or a single sample if ``size`` was not specified. See Also -------- normal : Equivalent function with additional ``loc`` and ``scale`` arguments for setting the mean and standard deviation. random.Generator.standard_normal: which should be used for new code. Notes ----- For random samples from the normal distribution with mean ``mu`` and standard deviation ``sigma``, use one of:: mu + sigma * np.random.standard_normal(size=...) np.random.normal(mu, sigma, size=...) Examples -------- >>> np.random.standard_normal() 2.1923875335537315 #random >>> s = np.random.standard_normal(8000) >>> s array([ 0.6888893 , 0.78096262, -0.89086505, ..., 0.49876311, # random -0.38672696, -0.4685006 ]) # random >>> s.shape (8000,) >>> s = np.random.standard_normal(size=(3, 4, 2)) >>> s.shape (3, 4, 2) Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> 3 + 2.5 * np.random.standard_normal(size=(2, 4)) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random wald(mean, scale, size=None) Draw samples from a Wald, or inverse Gaussian, distribution. As the scale approaches infinity, the distribution becomes more like a Gaussian. Some references claim that the Wald is an inverse Gaussian with mean equal to 1, but this is by no means universal. The inverse Gaussian distribution was first studied in relationship to Brownian motion. In 1956 M.C.K. Tweedie used the name inverse Gaussian because there is an inverse relationship between the time to cover a unit distance and distance covered in unit time. .. note:: New code should use the `~numpy.random.Generator.wald` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : float or array_like of floats Distribution mean, must be > 0. scale : float or array_like of floats Scale parameter, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mean`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(mean, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Wald distribution. See Also -------- random.Generator.wald: which should be used for new code. Notes ----- The probability density function for the Wald distribution is .. math:: P(x;mean,scale) = \sqrt{\frac{scale}{2\pi x^3}}e^ \frac{-scale(x-mean)^2}{2\cdotp mean^2x} As noted above the inverse Gaussian distribution first arise from attempts to model Brownian motion. It is also a competitor to the Weibull for use in reliability modeling and modeling stock returns and interest rate processes. References ---------- .. [1] Brighton Webs Ltd., Wald Distribution, https://web.archive.org/web/20090423014010/http://www.brighton-webs.co.uk:80/distributions/wald.asp .. [2] Chhikara, Raj S., and Folks, J. Leroy, "The Inverse Gaussian Distribution: Theory : Methodology, and Applications", CRC Press, 1988. .. [3] Wikipedia, "Inverse Gaussian distribution" https://en.wikipedia.org/wiki/Inverse_Gaussian_distribution Examples -------- Draw values from the distribution and plot the histogram: >>> import matplotlib.pyplot as plt >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, density=True) >>> plt.show() 1P a /qq/qq0/ ]!7%}A\a{Z[ $AQ %Qa $AQ vRqj vRqj vSj 4q,AT6WAt1t1 Kq 2T"HAWA *AQ 2T"HAWA *AQ 2T"F!7! *AQq!4AT6Q#4q#4q$D22DA !L  ! rqbaq 51 A qwa AQ A 3at83a *AQ Cq83cSas#Saq *AQ 4vQcCvQa *AQ d!57!4vQa D XRs$fAQ" c1 CuCq !2!;nA ]!85 Ql-qP]]^ Ya IQ y wbbj $aq6at:V4wVYYZyha!A = 2T"E"D *AQ$AQ&C1D RXX\\]%-Ya%,HA%/{!}AQA+81H Bhaq 1F#Q 86!5j y2T5s!j fCq *AQ qaq y2T5s!j 2WA 1A 2U!2V1Biq z%q2[*AavRuARvQawa /q_$7t1 +]!1 qcj qcj IQe1 rqj rG1Bbj s!64rj EA 4q A 2U!7&! A 1 1 rawas"A"$nAQca'8QbWARwhad(!3jQ uBajbj r2^1Bb2Q*AQAU!&bU&gRqE%rwb*AR{!"G1A3b#]!3e1('%}A"%q#U!1awgRs)1cad,ay!9A :T1G1 #U!1 1F#Q 1 4zSc "F!4vQa qaq 1qA^ :Qd' "G1CvR|1Cr  1b 3fBa *AQ 3fCq r"!5baq  1 bq6$fBa HAQs!1Ad 51Q*!1D,bavV2Q|=m1A EaqWAAU"5Qat1q*Jaz a ~.awoQ/qq ]!6m1Ja #1A $AQ F"A t2Yaqm1A 4q)$jd'd!t1Tr!7!! 31A 4rQb ! -qt1A%Qd*F$gQ$aDQRandomState.chisquare (line 1925)RandomState.dirichlet (line 4415)RandomState.geometric (line 3790)RandomState.hypergeometric (line 3852)RandomState.lognormal (line 2986)RandomState.logseries (line 3983)RandomState.multivariate_normalRandomState.standard_gamma (line 1581)Unsupported dtype %r for randint'a' cannot be empty unless no samples are takena must be greater than 0 unless no samples are taken binomial(n, p, size=None) Draw samples from a binomial distribution. Samples are drawn from a binomial distribution with specified parameters, n trials and p probability of success where n an integer >= 0 and p is in the interval [0,1]. (n may be input as a float, but it is truncated to an integer in use) .. note:: New code should use the `~numpy.random.Generator.binomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- n : int or array_like of ints Parameter of the distribution, >= 0. Floats are also accepted, but they will be truncated to integers. p : float or array_like of floats Parameter of the distribution, >= 0 and <=1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``n`` and ``p`` are both scalars. Otherwise, ``np.broadcast(n, p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized binomial distribution, where each sample is equal to the number of successes over the n trials. See Also -------- scipy.stats.binom : probability density function, distribution or cumulative density function, etc. random.Generator.binomial: which should be used for new code. Notes ----- The probability mass function (PMF) for the binomial distribution is .. math:: P(N) = \binom{n}{N}p^N(1-p)^{n-N}, where :math:`n` is the number of trials, :math:`p` is the probability of success, and :math:`N` is the number of successes. When estimating the standard error of a proportion in a population by using a random sample, the normal distribution works well unless the product p*n <=5, where p = population proportion estimate, and n = number of samples, in which case the binomial distribution is used instead. For example, a sample of 15 people shows 4 who are left handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4, so the binomial distribution should be used in this case. References ---------- .. [1] Dalgaard, Peter, "Introductory Statistics with R", Springer-Verlag, 2002. .. [2] Glantz, Stanton A. "Primer of Biostatistics.", McGraw-Hill, Fifth Edition, 2002. .. [3] Lentner, Marvin, "Elementary Applied Statistics", Bogden and Quigley, 1972. .. [4] Weisstein, Eric W. "Binomial Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/BinomialDistribution.html .. [5] Wikipedia, "Binomial distribution", https://en.wikipedia.org/wiki/Binomial_distribution Examples -------- Draw samples from the distribution: >>> n, p = 10, .5 # number of trials, probability of each trial >>> s = np.random.binomial(n, p, 1000) # result of flipping a coin 10 times, tested 1000 times. A real world example. A company drills 9 wild-cat oil exploration wells, each with an estimated probability of success of 0.1. All nine wells fail. What is the probability of that happening? Let's do 20,000 trials of the model, and count the number that generate zero positive results. >>> sum(np.random.binomial(9, 0.1, 20000) == 0)/20000. # answer = 0.38885, or 38%. bytes(length) Return random bytes. .. note:: New code should use the `~numpy.random.Generator.bytes` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- length : int Number of random bytes. Returns ------- out : bytes String of length `length`. See Also -------- random.Generator.bytes: which should be used for new code. Examples -------- >>> np.random.bytes(10) b' eh\x85\x022SZ\xbf\xa4' #random chisquare(df, size=None) Draw samples from a chi-square distribution. When `df` independent random variables, each with standard normal distributions (mean 0, variance 1), are squared and summed, the resulting distribution is chi-square (see Notes). This distribution is often used in hypothesis testing. .. note:: New code should use the `~numpy.random.Generator.chisquare` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- df : float or array_like of floats Number of degrees of freedom, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``df`` is a scalar. Otherwise, ``np.array(df).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized chi-square distribution. Raises ------ ValueError When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``) is given. See Also -------- random.Generator.chisquare: which should be used for new code. Notes ----- The variable obtained by summing the squares of `df` independent, standard normally distributed random variables: .. math:: Q = \sum_{i=1}^{\mathtt{df}} X^2_i is chi-square distributed, denoted .. math:: Q \sim \chi^2_k. The probability density function of the chi-squared distribution is .. math:: p(x) = \frac{(1/2)^{k/2}}{\Gamma(k/2)} x^{k/2 - 1} e^{-x/2}, where :math:`\Gamma` is the gamma function, .. math:: \Gamma(x) = \int_0^{-\infty} t^{x - 1} e^{-t} dt. References ---------- .. [1] NIST "Engineering Statistics Handbook" https://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm Examples -------- >>> np.random.chisquare(2,4) array([ 1.89920014, 9.00867716, 3.13710533, 5.62318272]) # random choice(a, size=None, replace=True, p=None) Generates a random sample from a given 1-D array .. note:: New code should use the `~numpy.random.Generator.choice` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. .. warning:: This function uses the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- a : 1-D array-like or int If an ndarray, a random sample is generated from its elements. If an int, the random sample is generated as if it were ``np.arange(a)`` size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. replace : boolean, optional Whether the sample is with or without replacement. Default is True, meaning that a value of ``a`` can be selected multiple times. p : 1-D array-like, optional The probabilities associated with each entry in a. If not given, the sample assumes a uniform distribution over all entries in ``a``. Returns ------- samples : single item or ndarray The generated random samples Raises ------ ValueError If a is an int and less than zero, if a or p are not 1-dimensional, if a is an array-like of size 0, if p is not a vector of probabilities, if a and p have different lengths, or if replace=False and the sample size is greater than the population size See Also -------- randint, shuffle, permutation random.Generator.choice: which should be used in new code Notes ----- Setting user-specified probabilities through ``p`` uses a more general but less efficient sampler than the default. The general sampler produces a different sample than the optimized sampler even if each element of ``p`` is 1 / len(a). Sampling random rows from a 2-D array is not possible with this function, but is possible with `Generator.choice` through its ``axis`` keyword. Examples -------- Generate a uniform random sample from np.arange(5) of size 3: >>> np.random.choice(5, 3) array([0, 3, 4]) # random >>> #This is equivalent to np.random.randint(0,5,3) Generate a non-uniform random sample from np.arange(5) of size 3: >>> np.random.choice(5, 3, p=[0.1, 0, 0.3, 0.6, 0]) array([3, 3, 0]) # random Generate a uniform random sample from np.arange(5) of size 3 without replacement: >>> np.random.choice(5, 3, replace=False) array([3,1,0]) # random >>> #This is equivalent to np.random.permutation(np.arange(5))[:3] Generate a non-uniform random sample from np.arange(5) of size 3 without replacement: >>> np.random.choice(5, 3, replace=False, p=[0.1, 0, 0.3, 0.6, 0]) array([2, 3, 0]) # random Any of the above can be repeated with an arbitrary array-like instead of just integers. For instance: >>> aa_milne_arr = ['pooh', 'rabbit', 'piglet', 'Christopher'] >>> np.random.choice(aa_milne_arr, 5, p=[0.5, 0.1, 0.1, 0.3]) array(['pooh', 'pooh', 'pooh', 'Christopher', 'piglet'], # random dtype=' 0`` and 0 elsewhere. :math:`\beta` is the scale parameter, which is the inverse of the rate parameter :math:`\lambda = 1/\beta`. The rate parameter is an alternative, widely used parameterization of the exponential distribution [3]_. The exponential distribution is a continuous analogue of the geometric distribution. It describes many common situations, such as the size of raindrops measured over many rainstorms [1]_, or the time between page requests to Wikipedia [2]_. .. note:: New code should use the `~numpy.random.Generator.exponential` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- scale : float or array_like of floats The scale parameter, :math:`\beta = 1/\lambda`. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``scale`` is a scalar. Otherwise, ``np.array(scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized exponential distribution. Examples -------- A real world example: Assume a company has 10000 customer support agents and the average time between customer calls is 4 minutes. >>> n = 10000 >>> time_between_calls = np.random.default_rng().exponential(scale=4, size=n) What is the probability that a customer will call in the next 4 to 5 minutes? >>> x = ((time_between_calls < 5).sum())/n >>> y = ((time_between_calls < 4).sum())/n >>> x-y 0.08 # may vary See Also -------- random.Generator.exponential: which should be used for new code. References ---------- .. [1] Peyton Z. Peebles Jr., "Probability, Random Variables and Random Signal Principles", 4th ed, 2001, p. 57. .. [2] Wikipedia, "Poisson process", https://en.wikipedia.org/wiki/Poisson_process .. [3] Wikipedia, "Exponential distribution", https://en.wikipedia.org/wiki/Exponential_distribution f(dfnum, dfden, size=None) Draw samples from an F distribution. Samples are drawn from an F distribution with specified parameters, `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom in denominator), where both parameters must be greater than zero. The random variate of the F distribution (also known as the Fisher distribution) is a continuous probability distribution that arises in ANOVA tests, and is the ratio of two chi-square variates. .. note:: New code should use the `~numpy.random.Generator.f` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- dfnum : float or array_like of floats Degrees of freedom in numerator, must be > 0. dfden : float or array_like of float Degrees of freedom in denominator, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``dfnum`` and ``dfden`` are both scalars. Otherwise, ``np.broadcast(dfnum, dfden).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Fisher distribution. See Also -------- scipy.stats.f : probability density function, distribution or cumulative density function, etc. random.Generator.f: which should be used for new code. Notes ----- The F statistic is used to compare in-group variances to between-group variances. Calculating the distribution depends on the sampling, and so it is a function of the respective degrees of freedom in the problem. The variable `dfnum` is the number of samples minus one, the between-groups degrees of freedom, while `dfden` is the within-groups degrees of freedom, the sum of the number of samples in each group minus the number of groups. References ---------- .. [1] Glantz, Stanton A. "Primer of Biostatistics.", McGraw-Hill, Fifth Edition, 2002. .. [2] Wikipedia, "F-distribution", https://en.wikipedia.org/wiki/F-distribution Examples -------- An example from Glantz[1], pp 47-40: Two groups, children of diabetics (25 people) and children from people without diabetes (25 controls). Fasting blood glucose was measured, case group had a mean value of 86.1, controls had a mean value of 82.2. Standard deviations were 2.09 and 2.49 respectively. Are these data consistent with the null hypothesis that the parents diabetic status does not affect their children's blood glucose levels? Calculating the F statistic from the data gives a value of 36.01. Draw samples from the distribution: >>> dfnum = 1. # between group degrees of freedom >>> dfden = 48. # within groups degrees of freedom >>> s = np.random.f(dfnum, dfden, 1000) The lower bound for the top 1% of the samples is : >>> np.sort(s)[-10] 7.61988120985 # random So there is about a 1% chance that the F statistic will exceed 7.62, the measured value is 36, so the null hypothesis is rejected at the 1% level. gamma(shape, scale=1.0, size=None) Draw samples from a Gamma distribution. Samples are drawn from a Gamma distribution with specified parameters, `shape` (sometimes designated "k") and `scale` (sometimes designated "theta"), where both parameters are > 0. .. note:: New code should use the `~numpy.random.Generator.gamma` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- shape : float or array_like of floats The shape of the gamma distribution. Must be non-negative. scale : float or array_like of floats, optional The scale of the gamma distribution. Must be non-negative. Default is equal to 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``shape`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(shape, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized gamma distribution. See Also -------- scipy.stats.gamma : probability density function, distribution or cumulative density function, etc. random.Generator.gamma: which should be used for new code. Notes ----- The probability density for the Gamma distribution is .. math:: p(x) = x^{k-1}\frac{e^{-x/\theta}}{\theta^k\Gamma(k)}, where :math:`k` is the shape and :math:`\theta` the scale, and :math:`\Gamma` is the Gamma function. The Gamma distribution is often used to model the times to failure of electronic components, and arises naturally in processes for which the waiting times between Poisson distributed events are relevant. References ---------- .. [1] Weisstein, Eric W. "Gamma Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/GammaDistribution.html .. [2] Wikipedia, "Gamma distribution", https://en.wikipedia.org/wiki/Gamma_distribution Examples -------- Draw samples from the distribution: >>> shape, scale = 2., 2. # mean=4, std=2*sqrt(2) >>> s = np.random.gamma(shape, scale, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> import scipy.special as sps # doctest: +SKIP >>> count, bins, ignored = plt.hist(s, 50, density=True) >>> y = bins**(shape-1)*(np.exp(-bins/scale) / # doctest: +SKIP ... (sps.gamma(shape)*scale**shape)) >>> plt.plot(bins, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() geometric(p, size=None) Draw samples from the geometric distribution. Bernoulli trials are experiments with one of two outcomes: success or failure (an example of such an experiment is flipping a coin). The geometric distribution models the number of trials that must be run in order to achieve success. It is therefore supported on the positive integers, ``k = 1, 2, ...``. The probability mass function of the geometric distribution is .. math:: f(k) = (1 - p)^{k - 1} p where `p` is the probability of success of an individual trial. .. note:: New code should use the `~numpy.random.Generator.geometric` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- p : float or array_like of floats The probability of success of an individual trial. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``p`` is a scalar. Otherwise, ``np.array(p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized geometric distribution. See Also -------- random.Generator.geometric: which should be used for new code. Examples -------- Draw ten thousand values from the geometric distribution, with the probability of an individual success equal to 0.35: >>> z = np.random.geometric(p=0.35, size=10000) How many trials succeeded after a single run? >>> (z == 1).sum() / 10000. 0.34889999999999999 #random gumbel(loc=0.0, scale=1.0, size=None) Draw samples from a Gumbel distribution. Draw samples from a Gumbel distribution with specified location and scale. For more information on the Gumbel distribution, see Notes and References below. .. note:: New code should use the `~numpy.random.Generator.gumbel` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional The location of the mode of the distribution. Default is 0. scale : float or array_like of floats, optional The scale parameter of the distribution. Default is 1. Must be non- negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Gumbel distribution. See Also -------- scipy.stats.gumbel_l scipy.stats.gumbel_r scipy.stats.genextreme weibull random.Generator.gumbel: which should be used for new code. Notes ----- The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value Type I) distribution is one of a class of Generalized Extreme Value (GEV) distributions used in modeling extreme value problems. The Gumbel is a special case of the Extreme Value Type I distribution for maximums from distributions with "exponential-like" tails. The probability density for the Gumbel distribution is .. math:: p(x) = \frac{e^{-(x - \mu)/ \beta}}{\beta} e^{ -e^{-(x - \mu)/ \beta}}, where :math:`\mu` is the mode, a location parameter, and :math:`\beta` is the scale parameter. The Gumbel (named for German mathematician Emil Julius Gumbel) was used very early in the hydrology literature, for modeling the occurrence of flood events. It is also used for modeling maximum wind speed and rainfall rates. It is a "fat-tailed" distribution - the probability of an event in the tail of the distribution is larger than if one used a Gaussian, hence the surprisingly frequent occurrence of 100-year floods. Floods were initially modeled as a Gaussian process, which underestimated the frequency of extreme events. It is one of a class of extreme value distributions, the Generalized Extreme Value (GEV) distributions, which also includes the Weibull and Frechet. The function has a mean of :math:`\mu + 0.57721\beta` and a variance of :math:`\frac{\pi^2}{6}\beta^2`. References ---------- .. [1] Gumbel, E. J., "Statistics of Extremes," New York: Columbia University Press, 1958. .. [2] Reiss, R.-D. and Thomas, M., "Statistical Analysis of Extreme Values from Insurance, Finance, Hydrology and Other Fields," Basel: Birkhauser Verlag, 2001. Examples -------- Draw samples from the distribution: >>> mu, beta = 0, 0.1 # location and scale >>> s = np.random.gumbel(mu, beta, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta) ... * np.exp( -np.exp( -(bins - mu) /beta) ), ... linewidth=2, color='r') >>> plt.show() Show how an extreme value distribution can arise from a Gaussian process and compare to a Gaussian: >>> means = [] >>> maxima = [] >>> for i in range(0,1000) : ... a = np.random.normal(mu, beta, 1000) ... means.append(a.mean()) ... maxima.append(a.max()) >>> count, bins, ignored = plt.hist(maxima, 30, density=True) >>> beta = np.std(maxima) * np.sqrt(6) / np.pi >>> mu = np.mean(maxima) - 0.57721*beta >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta) ... * np.exp(-np.exp(-(bins - mu)/beta)), ... linewidth=2, color='r') >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi)) ... * np.exp(-(bins - mu)**2 / (2 * beta**2)), ... linewidth=2, color='g') >>> plt.show() hypergeometric(ngood, nbad, nsample, size=None) Draw samples from a Hypergeometric distribution. Samples are drawn from a hypergeometric distribution with specified parameters, `ngood` (ways to make a good selection), `nbad` (ways to make a bad selection), and `nsample` (number of items sampled, which is less than or equal to the sum ``ngood + nbad``). .. note:: New code should use the `~numpy.random.Generator.hypergeometric` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- ngood : int or array_like of ints Number of ways to make a good selection. Must be nonnegative. nbad : int or array_like of ints Number of ways to make a bad selection. Must be nonnegative. nsample : int or array_like of ints Number of items sampled. Must be at least 1 and at most ``ngood + nbad``. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if `ngood`, `nbad`, and `nsample` are all scalars. Otherwise, ``np.broadcast(ngood, nbad, nsample).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized hypergeometric distribution. Each sample is the number of good items within a randomly selected subset of size `nsample` taken from a set of `ngood` good items and `nbad` bad items. See Also -------- scipy.stats.hypergeom : probability density function, distribution or cumulative density function, etc. random.Generator.hypergeometric: which should be used for new code. Notes ----- The probability mass function (PMF) for the Hypergeometric distribution is .. math:: P(x) = \frac{\binom{g}{x}\binom{b}{n-x}}{\binom{g+b}{n}}, where :math:`0 \le x \le n` and :math:`n-b \le x \le g` for P(x) the probability of ``x`` good results in the drawn sample, g = `ngood`, b = `nbad`, and n = `nsample`. Consider an urn with black and white marbles in it, `ngood` of them are black and `nbad` are white. If you draw `nsample` balls without replacement, then the hypergeometric distribution describes the distribution of black balls in the drawn sample. Note that this distribution is very similar to the binomial distribution, except that in this case, samples are drawn without replacement, whereas in the Binomial case samples are drawn with replacement (or the sample space is infinite). As the sample space becomes large, this distribution approaches the binomial. References ---------- .. [1] Lentner, Marvin, "Elementary Applied Statistics", Bogden and Quigley, 1972. .. [2] Weisstein, Eric W. "Hypergeometric Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/HypergeometricDistribution.html .. [3] Wikipedia, "Hypergeometric distribution", https://en.wikipedia.org/wiki/Hypergeometric_distribution Examples -------- Draw samples from the distribution: >>> ngood, nbad, nsamp = 100, 2, 10 # number of good, number of bad, and number of samples >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000) >>> from matplotlib.pyplot import hist >>> hist(s) # note that it is very unlikely to grab both bad items Suppose you have an urn with 15 white and 15 black marbles. If you pull 15 marbles at random, how likely is it that 12 or more of them are one color? >>> s = np.random.hypergeometric(15, 15, 15, 100000) >>> sum(s>=12)/100000. + sum(s<=3)/100000. # answer = 0.003 ... pretty unlikely! logistic(loc=0.0, scale=1.0, size=None) Draw samples from a logistic distribution. Samples are drawn from a logistic distribution with specified parameters, loc (location or mean, also median), and scale (>0). .. note:: New code should use the `~numpy.random.Generator.logistic` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional Parameter of the distribution. Default is 0. scale : float or array_like of floats, optional Parameter of the distribution. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized logistic distribution. See Also -------- scipy.stats.logistic : probability density function, distribution or cumulative density function, etc. random.Generator.logistic: which should be used for new code. Notes ----- The probability density for the Logistic distribution is .. math:: P(x) = P(x) = \frac{e^{-(x-\mu)/s}}{s(1+e^{-(x-\mu)/s})^2}, where :math:`\mu` = location and :math:`s` = scale. The Logistic distribution is used in Extreme Value problems where it can act as a mixture of Gumbel distributions, in Epidemiology, and by the World Chess Federation (FIDE) where it is used in the Elo ranking system, assuming the performance of each player is a logistically distributed random variable. References ---------- .. [1] Reiss, R.-D. and Thomas M. (2001), "Statistical Analysis of Extreme Values, from Insurance, Finance, Hydrology and Other Fields," Birkhauser Verlag, Basel, pp 132-133. .. [2] Weisstein, Eric W. "Logistic Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/LogisticDistribution.html .. [3] Wikipedia, "Logistic-distribution", https://en.wikipedia.org/wiki/Logistic_distribution Examples -------- Draw samples from the distribution: >>> loc, scale = 10, 1 >>> s = np.random.logistic(loc, scale, 10000) >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, bins=50) # plot against distribution >>> def logist(x, loc, scale): ... return np.exp((loc-x)/scale)/(scale*(1+np.exp((loc-x)/scale))**2) >>> lgst_val = logist(bins, loc, scale) >>> plt.plot(bins, lgst_val * count.max() / lgst_val.max()) >>> plt.show() lognormal(mean=0.0, sigma=1.0, size=None) Draw samples from a log-normal distribution. Draw samples from a log-normal distribution with specified mean, standard deviation, and array shape. Note that the mean and standard deviation are not the values for the distribution itself, but of the underlying normal distribution it is derived from. .. note:: New code should use the `~numpy.random.Generator.lognormal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : float or array_like of floats, optional Mean value of the underlying normal distribution. Default is 0. sigma : float or array_like of floats, optional Standard deviation of the underlying normal distribution. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mean`` and ``sigma`` are both scalars. Otherwise, ``np.broadcast(mean, sigma).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized log-normal distribution. See Also -------- scipy.stats.lognorm : probability density function, distribution, cumulative density function, etc. random.Generator.lognormal: which should be used for new code. Notes ----- A variable `x` has a log-normal distribution if `log(x)` is normally distributed. The probability density function for the log-normal distribution is: .. math:: p(x) = \frac{1}{\sigma x \sqrt{2\pi}} e^{(-\frac{(ln(x)-\mu)^2}{2\sigma^2})} where :math:`\mu` is the mean and :math:`\sigma` is the standard deviation of the normally distributed logarithm of the variable. A log-normal distribution results if a random variable is the *product* of a large number of independent, identically-distributed variables in the same way that a normal distribution results if the variable is the *sum* of a large number of independent, identically-distributed variables. References ---------- .. [1] Limpert, E., Stahel, W. A., and Abbt, M., "Log-normal Distributions across the Sciences: Keys and Clues," BioScience, Vol. 51, No. 5, May, 2001. https://stat.ethz.ch/~stahel/lognormal/bioscience.pdf .. [2] Reiss, R.D. and Thomas, M., "Statistical Analysis of Extreme Values," Basel: Birkhauser Verlag, 2001, pp. 31-32. Examples -------- Draw samples from the distribution: >>> mu, sigma = 3., 1. # mean and standard deviation >>> s = np.random.lognormal(mu, sigma, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 100, density=True, align='mid') >>> x = np.linspace(min(bins), max(bins), 10000) >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2)) ... / (x * sigma * np.sqrt(2 * np.pi))) >>> plt.plot(x, pdf, linewidth=2, color='r') >>> plt.axis('tight') >>> plt.show() Demonstrate that taking the products of random samples from a uniform distribution can be fit well by a log-normal probability density function. >>> # Generate a thousand samples: each is the product of 100 random >>> # values, drawn from a normal distribution. >>> b = [] >>> for i in range(1000): ... a = 10. + np.random.standard_normal(100) ... b.append(np.prod(a)) >>> b = np.array(b) / np.min(b) # scale values to be positive >>> count, bins, ignored = plt.hist(b, 100, density=True, align='mid') >>> sigma = np.std(np.log(b)) >>> mu = np.mean(np.log(b)) >>> x = np.linspace(min(bins), max(bins), 10000) >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2)) ... / (x * sigma * np.sqrt(2 * np.pi))) >>> plt.plot(x, pdf, color='r', linewidth=2) >>> plt.show() logseries(p, size=None) Draw samples from a logarithmic series distribution. Samples are drawn from a log series distribution with specified shape parameter, 0 <= ``p`` < 1. .. note:: New code should use the `~numpy.random.Generator.logseries` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- p : float or array_like of floats Shape parameter for the distribution. Must be in the range [0, 1). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``p`` is a scalar. Otherwise, ``np.array(p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized logarithmic series distribution. See Also -------- scipy.stats.logser : probability density function, distribution or cumulative density function, etc. random.Generator.logseries: which should be used for new code. Notes ----- The probability density for the Log Series distribution is .. math:: P(k) = \frac{-p^k}{k \ln(1-p)}, where p = probability. The log series distribution is frequently used to represent species richness and occurrence, first proposed by Fisher, Corbet, and Williams in 1943 [2]. It may also be used to model the numbers of occupants seen in cars [3]. References ---------- .. [1] Buzas, Martin A.; Culver, Stephen J., Understanding regional species diversity through the log series distribution of occurrences: BIODIVERSITY RESEARCH Diversity & Distributions, Volume 5, Number 5, September 1999 , pp. 187-195(9). .. [2] Fisher, R.A,, A.S. Corbet, and C.B. Williams. 1943. The relation between the number of species and the number of individuals in a random sample of an animal population. Journal of Animal Ecology, 12:42-58. .. [3] D. J. Hand, F. Daly, D. Lunn, E. Ostrowski, A Handbook of Small Data Sets, CRC Press, 1994. .. [4] Wikipedia, "Logarithmic distribution", https://en.wikipedia.org/wiki/Logarithmic_distribution Examples -------- Draw samples from the distribution: >>> a = .6 >>> s = np.random.logseries(a, 10000) >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s) # plot against distribution >>> def logseries(k, p): ... return -p**k/(k*np.log(1-p)) >>> plt.plot(bins, logseries(bins, a)*count.max()/ ... logseries(bins, a).max(), 'r') >>> plt.show() multinomial(n, pvals, size=None) Draw samples from a multinomial distribution. The multinomial distribution is a multivariate generalization of the binomial distribution. Take an experiment with one of ``p`` possible outcomes. An example of such an experiment is throwing a dice, where the outcome can be 1 through 6. Each sample drawn from the distribution represents `n` such experiments. Its values, ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome was ``i``. .. note:: New code should use the `~numpy.random.Generator.multinomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. .. warning:: This function defaults to the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- n : int Number of experiments. pvals : sequence of floats, length p Probabilities of each of the ``p`` different outcomes. These must sum to 1 (however, the last element is always assumed to account for the remaining probability, as long as ``sum(pvals[:-1]) <= 1)``. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : ndarray The drawn samples, of shape *size*, if that was provided. If not, the shape is ``(N,)``. In other words, each entry ``out[i,j,...,:]`` is an N-dimensional value drawn from the distribution. See Also -------- random.Generator.multinomial: which should be used for new code. Examples -------- Throw a dice 20 times: >>> np.random.multinomial(20, [1/6.]*6, size=1) array([[4, 1, 7, 5, 2, 1]]) # random It landed 4 times on 1, once on 2, etc. Now, throw the dice 20 times, and 20 times again: >>> np.random.multinomial(20, [1/6.]*6, size=2) array([[3, 4, 3, 3, 4, 3], # random [2, 4, 3, 4, 0, 7]]) For the first run, we threw 3 times 1, 4 times 2, etc. For the second, we threw 2 times 1, 4 times 2, etc. A loaded die is more likely to land on number 6: >>> np.random.multinomial(100, [1/7.]*5 + [2/7.]) array([11, 16, 14, 17, 16, 26]) # random The probability inputs should be normalized. As an implementation detail, the value of the last entry is ignored and assumed to take up any leftover probability mass, but this should not be relied on. A biased coin which has twice as much weight on one side as on the other should be sampled like so: >>> np.random.multinomial(100, [1.0 / 3, 2.0 / 3]) # RIGHT array([38, 62]) # random not like: >>> np.random.multinomial(100, [1.0, 2.0]) # WRONG Traceback (most recent call last): ValueError: pvals < 0, pvals > 1 or pvals contains NaNs multivariate_normal(mean, cov, size=None, check_valid='warn', tol=1e-8) Draw random samples from a multivariate normal distribution. The multivariate normal, multinormal or Gaussian distribution is a generalization of the one-dimensional normal distribution to higher dimensions. Such a distribution is specified by its mean and covariance matrix. These parameters are analogous to the mean (average or "center") and variance (standard deviation, or "width," squared) of the one-dimensional normal distribution. .. note:: New code should use the `~numpy.random.Generator.multivariate_normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : 1-D array_like, of length N Mean of the N-dimensional distribution. cov : 2-D array_like, of shape (N, N) Covariance matrix of the distribution. It must be symmetric and positive-semidefinite for proper sampling. size : int or tuple of ints, optional Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are generated, and packed in an `m`-by-`n`-by-`k` arrangement. Because each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``. If no shape is specified, a single (`N`-D) sample is returned. check_valid : { 'warn', 'raise', 'ignore' }, optional Behavior when the covariance matrix is not positive semidefinite. tol : float, optional Tolerance when checking the singular values in covariance matrix. cov is cast to double before the check. Returns ------- out : ndarray The drawn samples, of shape *size*, if that was provided. If not, the shape is ``(N,)``. In other words, each entry ``out[i,j,...,:]`` is an N-dimensional value drawn from the distribution. See Also -------- random.Generator.multivariate_normal: which should be used for new code. Notes ----- The mean is a coordinate in N-dimensional space, which represents the location where samples are most likely to be generated. This is analogous to the peak of the bell curve for the one-dimensional or univariate normal distribution. Covariance indicates the level to which two variables vary together. From the multivariate normal distribution, we draw N-dimensional samples, :math:`X = [x_1, x_2, ... x_N]`. The covariance matrix element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`. The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its "spread"). Instead of specifying the full covariance matrix, popular approximations include: - Spherical covariance (`cov` is a multiple of the identity matrix) - Diagonal covariance (`cov` has non-negative elements, and only on the diagonal) This geometrical property can be seen in two dimensions by plotting generated data-points: >>> mean = [0, 0] >>> cov = [[1, 0], [0, 100]] # diagonal covariance Diagonal covariance means that points are oriented along x or y-axis: >>> import matplotlib.pyplot as plt >>> x, y = np.random.multivariate_normal(mean, cov, 5000).T >>> plt.plot(x, y, 'x') >>> plt.axis('equal') >>> plt.show() Note that the covariance matrix must be positive semidefinite (a.k.a. nonnegative-definite). Otherwise, the behavior of this method is undefined and backwards compatibility is not guaranteed. References ---------- .. [1] Papoulis, A., "Probability, Random Variables, and Stochastic Processes," 3rd ed., New York: McGraw-Hill, 1991. .. [2] Duda, R. O., Hart, P. E., and Stork, D. G., "Pattern Classification," 2nd ed., New York: Wiley, 2001. Examples -------- >>> mean = (1, 2) >>> cov = [[1, 0], [0, 1]] >>> x = np.random.multivariate_normal(mean, cov, (3, 3)) >>> x.shape (3, 3, 2) Here we generate 800 samples from the bivariate normal distribution with mean [0, 0] and covariance matrix [[6, -3], [-3, 3.5]]. The expected variances of the first and second components of the sample are 6 and 3.5, respectively, and the expected correlation coefficient is -3/sqrt(6*3.5) ≈ -0.65465. >>> cov = np.array([[6, -3], [-3, 3.5]]) >>> pts = np.random.multivariate_normal([0, 0], cov, size=800) Check that the mean, covariance, and correlation coefficient of the sample are close to the expected values: >>> pts.mean(axis=0) array([ 0.0326911 , -0.01280782]) # may vary >>> np.cov(pts.T) array([[ 5.96202397, -2.85602287], [-2.85602287, 3.47613949]]) # may vary >>> np.corrcoef(pts.T)[0, 1] -0.6273591314603949 # may vary We can visualize this data with a scatter plot. The orientation of the point cloud illustrates the negative correlation of the components of this sample. >>> import matplotlib.pyplot as plt >>> plt.plot(pts[:, 0], pts[:, 1], '.', alpha=0.5) >>> plt.axis('equal') >>> plt.grid() >>> plt.show() negative_binomial(n, p, size=None) Draw samples from a negative binomial distribution. Samples are drawn from a negative binomial distribution with specified parameters, `n` successes and `p` probability of success where `n` is > 0 and `p` is in the interval [0, 1]. .. note:: New code should use the `~numpy.random.Generator.negative_binomial` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- n : float or array_like of floats Parameter of the distribution, > 0. p : float or array_like of floats Parameter of the distribution, >= 0 and <=1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``n`` and ``p`` are both scalars. Otherwise, ``np.broadcast(n, p).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized negative binomial distribution, where each sample is equal to N, the number of failures that occurred before a total of n successes was reached. .. warning:: This function returns the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. See Also -------- random.Generator.negative_binomial: which should be used for new code. Notes ----- The probability mass function of the negative binomial distribution is .. math:: P(N;n,p) = \frac{\Gamma(N+n)}{N!\Gamma(n)}p^{n}(1-p)^{N}, where :math:`n` is the number of successes, :math:`p` is the probability of success, :math:`N+n` is the number of trials, and :math:`\Gamma` is the gamma function. When :math:`n` is an integer, :math:`\frac{\Gamma(N+n)}{N!\Gamma(n)} = \binom{N+n-1}{N}`, which is the more common form of this term in the pmf. The negative binomial distribution gives the probability of N failures given n successes, with a success on the last trial. If one throws a die repeatedly until the third time a "1" appears, then the probability distribution of the number of non-"1"s that appear before the third "1" is a negative binomial distribution. References ---------- .. [1] Weisstein, Eric W. "Negative Binomial Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/NegativeBinomialDistribution.html .. [2] Wikipedia, "Negative binomial distribution", https://en.wikipedia.org/wiki/Negative_binomial_distribution Examples -------- Draw samples from the distribution: A real world example. A company drills wild-cat oil exploration wells, each with an estimated probability of success of 0.1. What is the probability of having one success for each successive well, that is what is the probability of a single success after drilling 5 wells, after 6 wells, etc.? >>> s = np.random.negative_binomial(1, 0.1, 100000) >>> for i in range(1, 11): # doctest: +SKIP ... probability = sum(s>> import matplotlib.pyplot as plt >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000), ... bins=200, density=True) >>> plt.show() Draw values from a noncentral chisquare with very small noncentrality, and compare to a chisquare. >>> plt.figure() >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000), ... bins=np.arange(0., 25, .1), density=True) >>> values2 = plt.hist(np.random.chisquare(3, 100000), ... bins=np.arange(0., 25, .1), density=True) >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob') >>> plt.show() Demonstrate how large values of non-centrality lead to a more symmetric distribution. >>> plt.figure() >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000), ... bins=200, density=True) >>> plt.show() noncentral_f(dfnum, dfden, nonc, size=None) Draw samples from the noncentral F distribution. Samples are drawn from an F distribution with specified parameters, `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom in denominator), where both parameters > 1. `nonc` is the non-centrality parameter. .. note:: New code should use the `~numpy.random.Generator.noncentral_f` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- dfnum : float or array_like of floats Numerator degrees of freedom, must be > 0. dfden : float or array_like of floats Denominator degrees of freedom, must be > 0. nonc : float or array_like of floats Non-centrality parameter, the sum of the squares of the numerator means, must be >= 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``dfnum``, ``dfden``, and ``nonc`` are all scalars. Otherwise, ``np.broadcast(dfnum, dfden, nonc).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized noncentral Fisher distribution. See Also -------- random.Generator.noncentral_f: which should be used for new code. Notes ----- When calculating the power of an experiment (power = probability of rejecting the null hypothesis when a specific alternative is true) the non-central F statistic becomes important. When the null hypothesis is true, the F statistic follows a central F distribution. When the null hypothesis is not true, then it follows a non-central F statistic. References ---------- .. [1] Weisstein, Eric W. "Noncentral F-Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/NoncentralF-Distribution.html .. [2] Wikipedia, "Noncentral F-distribution", https://en.wikipedia.org/wiki/Noncentral_F-distribution Examples -------- In a study, testing for a specific alternative to the null hypothesis requires use of the Noncentral F distribution. We need to calculate the area in the tail of the distribution that exceeds the value of the F distribution for the null hypothesis. We'll plot the two probability distributions for comparison. >>> dfnum = 3 # between group deg of freedom >>> dfden = 20 # within groups degrees of freedom >>> nonc = 3.0 >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000) >>> NF = np.histogram(nc_vals, bins=50, density=True) >>> c_vals = np.random.f(dfnum, dfden, 1000000) >>> F = np.histogram(c_vals, bins=50, density=True) >>> import matplotlib.pyplot as plt >>> plt.plot(F[1][1:], F[0]) >>> plt.plot(NF[1][1:], NF[0]) >>> plt.show() normal(loc=0.0, scale=1.0, size=None) Draw random samples from a normal (Gaussian) distribution. The probability density function of the normal distribution, first derived by De Moivre and 200 years later by both Gauss and Laplace independently [2]_, is often called the bell curve because of its characteristic shape (see the example below). The normal distributions occurs often in nature. For example, it describes the commonly occurring distribution of samples influenced by a large number of tiny, random disturbances, each with its own unique distribution [2]_. .. note:: New code should use the `~numpy.random.Generator.normal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats Mean ("centre") of the distribution. scale : float or array_like of floats Standard deviation (spread or "width") of the distribution. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized normal distribution. See Also -------- scipy.stats.norm : probability density function, distribution or cumulative density function, etc. random.Generator.normal: which should be used for new code. Notes ----- The probability density for the Gaussian distribution is .. math:: p(x) = \frac{1}{\sqrt{ 2 \pi \sigma^2 }} e^{ - \frac{ (x - \mu)^2 } {2 \sigma^2} }, where :math:`\mu` is the mean and :math:`\sigma` the standard deviation. The square of the standard deviation, :math:`\sigma^2`, is called the variance. The function has its peak at the mean, and its "spread" increases with the standard deviation (the function reaches 0.607 times its maximum at :math:`x + \sigma` and :math:`x - \sigma` [2]_). This implies that normal is more likely to return samples lying close to the mean, rather than those far away. References ---------- .. [1] Wikipedia, "Normal distribution", https://en.wikipedia.org/wiki/Normal_distribution .. [2] P. R. Peebles Jr., "Central Limit Theorem" in "Probability, Random Variables and Random Signal Principles", 4th ed., 2001, pp. 51, 51, 125. Examples -------- Draw samples from the distribution: >>> mu, sigma = 0, 0.1 # mean and standard deviation >>> s = np.random.normal(mu, sigma, 1000) Verify the mean and the standard deviation: >>> abs(mu - np.mean(s)) 0.0 # may vary >>> abs(sigma - np.std(s, ddof=1)) 0.0 # may vary Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) * ... np.exp( - (bins - mu)**2 / (2 * sigma**2) ), ... linewidth=2, color='r') >>> plt.show() Two-by-four array of samples from the normal distribution with mean 3 and standard deviation 2.5: >>> np.random.normal(3, 2.5, size=(2, 4)) array([[-4.49401501, 4.00950034, -1.81814867, 7.29718677], # random [ 0.39924804, 4.68456316, 4.99394529, 4.84057254]]) # random pareto(a, size=None) Draw samples from a Pareto II or Lomax distribution with specified shape. The Lomax or Pareto II distribution is a shifted Pareto distribution. The classical Pareto distribution can be obtained from the Lomax distribution by adding 1 and multiplying by the scale parameter ``m`` (see Notes). The smallest value of the Lomax distribution is zero while for the classical Pareto distribution it is ``mu``, where the standard Pareto distribution has location ``mu = 1``. Lomax can also be considered as a simplified version of the Generalized Pareto distribution (available in SciPy), with the scale set to one and the location set to zero. The Pareto distribution must be greater than zero, and is unbounded above. It is also known as the "80-20 rule". In this distribution, 80 percent of the weights are in the lowest 20 percent of the range, while the other 20 percent fill the remaining 80 percent of the range. .. note:: New code should use the `~numpy.random.Generator.pareto` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Shape of the distribution. Must be positive. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Pareto distribution. See Also -------- scipy.stats.lomax : probability density function, distribution or cumulative density function, etc. scipy.stats.genpareto : probability density function, distribution or cumulative density function, etc. random.Generator.pareto: which should be used for new code. Notes ----- The probability density for the Pareto distribution is .. math:: p(x) = \frac{am^a}{x^{a+1}} where :math:`a` is the shape and :math:`m` the scale. The Pareto distribution, named after the Italian economist Vilfredo Pareto, is a power law probability distribution useful in many real world problems. Outside the field of economics it is generally referred to as the Bradford distribution. Pareto developed the distribution to describe the distribution of wealth in an economy. It has also found use in insurance, web page access statistics, oil field sizes, and many other problems, including the download frequency for projects in Sourceforge [1]_. It is one of the so-called "fat-tailed" distributions. References ---------- .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of Sourceforge projects. .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne. .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme Values, Birkhauser Verlag, Basel, pp 23-30. .. [4] Wikipedia, "Pareto distribution", https://en.wikipedia.org/wiki/Pareto_distribution Examples -------- Draw samples from the distribution: >>> a, m = 3., 2. # shape and mode >>> s = (np.random.pareto(a, 1000) + 1) * m Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, _ = plt.hist(s, 100, density=True) >>> fit = a*m**a / bins**(a+1) >>> plt.plot(bins, max(count)*fit/max(fit), linewidth=2, color='r') >>> plt.show() power(a, size=None) Draws samples in [0, 1] from a power distribution with positive exponent a - 1. Also known as the power function distribution. .. note:: New code should use the `~numpy.random.Generator.power` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Parameter of the distribution. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized power distribution. Raises ------ ValueError If a <= 0. See Also -------- random.Generator.power: which should be used for new code. Notes ----- The probability density function is .. math:: P(x; a) = ax^{a-1}, 0 \le x \le 1, a>0. The power function distribution is just the inverse of the Pareto distribution. It may also be seen as a special case of the Beta distribution. It is used, for example, in modeling the over-reporting of insurance claims. References ---------- .. [1] Christian Kleiber, Samuel Kotz, "Statistical size distributions in economics and actuarial sciences", Wiley, 2003. .. [2] Heckert, N. A. and Filliben, James J. "NIST Handbook 148: Dataplot Reference Manual, Volume 2: Let Subcommands and Library Functions", National Institute of Standards and Technology Handbook Series, June 2003. https://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf Examples -------- Draw samples from the distribution: >>> a = 5. # shape >>> samples = 1000 >>> s = np.random.power(a, samples) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, bins=30) >>> x = np.linspace(0, 1, 100) >>> y = a*x**(a-1.) >>> normed_y = samples*np.diff(bins)[0]*y >>> plt.plot(x, normed_y) >>> plt.show() Compare the power function distribution to the inverse of the Pareto. >>> from scipy import stats # doctest: +SKIP >>> rvs = np.random.power(5, 1000000) >>> rvsp = np.random.pareto(5, 1000000) >>> xx = np.linspace(0,1,100) >>> powpdf = stats.powerlaw.pdf(xx,5) # doctest: +SKIP >>> plt.figure() >>> plt.hist(rvs, bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('np.random.power(5)') >>> plt.figure() >>> plt.hist(1./(1.+rvsp), bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('inverse of 1 + np.random.pareto(5)') >>> plt.figure() >>> plt.hist(1./(1.+rvsp), bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('inverse of stats.pareto(5)') qP T 2Qs*D E!45 Q !?$k !;d+Q 7$d*AT):! *A 1 Ba)1HAXRqa~Rqq randint(low, high=None, size=None, dtype=int) Return random integers from `low` (inclusive) to `high` (exclusive). Return random integers from the "discrete uniform" distribution of the specified dtype in the "half-open" interval [`low`, `high`). If `high` is None (the default), then results are from [0, `low`). .. note:: New code should use the `~numpy.random.Generator.integers` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- low : int or array-like of ints Lowest (signed) integers to be drawn from the distribution (unless ``high=None``, in which case this parameter is one above the *highest* such integer). high : int or array-like of ints, optional If provided, one above the largest (signed) integer to be drawn from the distribution (see above for behavior if ``high=None``). If array-like, must contain integer values size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. dtype : dtype, optional Desired dtype of the result. Byteorder must be native. The default value is long. .. warning:: This function defaults to the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Which corresponds to `np.intp`. (`dtype=int` is not the same as in most NumPy functions.) Returns ------- out : int or ndarray of ints `size`-shaped array of random integers from the appropriate distribution, or a single such random int if `size` not provided. See Also -------- random_integers : similar to `randint`, only for the closed interval [`low`, `high`], and 1 is the lowest value if `high` is omitted. random.Generator.integers: which should be used for new code. Examples -------- >>> np.random.randint(2, size=10) array([1, 0, 0, 0, 1, 1, 0, 0, 1, 0]) # random >>> np.random.randint(1, size=10) array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0]) Generate a 2 x 4 array of ints between 0 and 4, inclusive: >>> np.random.randint(5, size=(2, 4)) array([[4, 0, 2, 1], # random [3, 2, 2, 0]]) Generate a 1 x 3 array with 3 different upper bounds >>> np.random.randint(1, [3, 5, 10]) array([2, 2, 9]) # random Generate a 1 by 3 array with 3 different lower bounds >>> np.random.randint([1, 5, 7], 10) array([9, 8, 7]) # random Generate a 2 by 4 array using broadcasting with dtype of uint8 >>> np.random.randint([1, 3, 5, 7], [[10], [20]], dtype=np.uint8) array([[ 8, 6, 9, 7], # random [ 1, 16, 9, 12]], dtype=uint8) random_integers(low, high=None, size=None) Random integers of type `numpy.int_` between `low` and `high`, inclusive. Return random integers of type `numpy.int_` from the "discrete uniform" distribution in the closed interval [`low`, `high`]. If `high` is None (the default), then results are from [1, `low`]. The `numpy.int_` type translates to the C long integer type and its precision is platform dependent. This function has been deprecated. Use randint instead. .. deprecated:: 1.11.0 Parameters ---------- low : int Lowest (signed) integer to be drawn from the distribution (unless ``high=None``, in which case this parameter is the *highest* such integer). high : int, optional If provided, the largest (signed) integer to be drawn from the distribution (see above for behavior if ``high=None``). size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : int or ndarray of ints `size`-shaped array of random integers from the appropriate distribution, or a single such random int if `size` not provided. See Also -------- randint : Similar to `random_integers`, only for the half-open interval [`low`, `high`), and 0 is the lowest value if `high` is omitted. Notes ----- To sample from N evenly spaced floating-point numbers between a and b, use:: a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.) Examples -------- >>> np.random.random_integers(5) 4 # random >>> type(np.random.random_integers(5)) >>> np.random.random_integers(5, size=(3,2)) array([[5, 4], # random [3, 3], [4, 5]]) Choose five random numbers from the set of five evenly-spaced numbers between 0 and 2.5, inclusive (*i.e.*, from the set :math:`{0, 5/8, 10/8, 15/8, 20/8}`): >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4. array([ 0.625, 1.25 , 0.625, 0.625, 2.5 ]) # random Roll two six sided dice 1000 times and sum the results: >>> d1 = np.random.random_integers(1, 6, 1000) >>> d2 = np.random.random_integers(1, 6, 1000) >>> dsums = d1 + d2 Display results as a histogram: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(dsums, 11, density=True) >>> plt.show() rayleigh(scale=1.0, size=None) Draw samples from a Rayleigh distribution. The :math:`\chi` and Weibull distributions are generalizations of the Rayleigh. .. note:: New code should use the `~numpy.random.Generator.rayleigh` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- scale : float or array_like of floats, optional Scale, also equals the mode. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``scale`` is a scalar. Otherwise, ``np.array(scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Rayleigh distribution. See Also -------- random.Generator.rayleigh: which should be used for new code. Notes ----- The probability density function for the Rayleigh distribution is .. math:: P(x;scale) = \frac{x}{scale^2}e^{\frac{-x^2}{2 \cdotp scale^2}} The Rayleigh distribution would arise, for example, if the East and North components of the wind velocity had identical zero-mean Gaussian distributions. Then the wind speed would have a Rayleigh distribution. References ---------- .. [1] Brighton Webs Ltd., "Rayleigh Distribution," https://web.archive.org/web/20090514091424/http://brighton-webs.co.uk:80/distributions/rayleigh.asp .. [2] Wikipedia, "Rayleigh distribution" https://en.wikipedia.org/wiki/Rayleigh_distribution Examples -------- Draw values from the distribution and plot the histogram >>> from matplotlib.pyplot import hist >>> values = hist(np.random.rayleigh(3, 100000), bins=200, density=True) Wave heights tend to follow a Rayleigh distribution. If the mean wave height is 1 meter, what fraction of waves are likely to be larger than 3 meters? >>> meanvalue = 1 >>> modevalue = np.sqrt(2 / np.pi) * meanvalue >>> s = np.random.rayleigh(modevalue, 1000000) The percentage of waves larger than 3 meters is: >>> 100.*sum(s>3)/1000000. 0.087300000000000003 # random set_state can only be used with legacy MT19937 state instances. standard_t(df, size=None) Draw samples from a standard Student's t distribution with `df` degrees of freedom. A special case of the hyperbolic distribution. As `df` gets large, the result resembles that of the standard normal distribution (`standard_normal`). .. note:: New code should use the `~numpy.random.Generator.standard_t` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- df : float or array_like of floats Degrees of freedom, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``df`` is a scalar. Otherwise, ``np.array(df).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized standard Student's t distribution. See Also -------- random.Generator.standard_t: which should be used for new code. Notes ----- The probability density function for the t distribution is .. math:: P(x, df) = \frac{\Gamma(\frac{df+1}{2})}{\sqrt{\pi df} \Gamma(\frac{df}{2})}\Bigl( 1+\frac{x^2}{df} \Bigr)^{-(df+1)/2} The t test is based on an assumption that the data come from a Normal distribution. The t test provides a way to test whether the sample mean (that is the mean calculated from the data) is a good estimate of the true mean. The derivation of the t-distribution was first published in 1908 by William Gosset while working for the Guinness Brewery in Dublin. Due to proprietary issues, he had to publish under a pseudonym, and so he used the name Student. References ---------- .. [1] Dalgaard, Peter, "Introductory Statistics With R", Springer, 2002. .. [2] Wikipedia, "Student's t-distribution" https://en.wikipedia.org/wiki/Student's_t-distribution Examples -------- From Dalgaard page 83 [1]_, suppose the daily energy intake for 11 women in kilojoules (kJ) is: >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \ ... 7515, 8230, 8770]) Does their energy intake deviate systematically from the recommended value of 7725 kJ? Our null hypothesis will be the absence of deviation, and the alternate hypothesis will be the presence of an effect that could be either positive or negative, hence making our test 2-tailed. Because we are estimating the mean and we have N=11 values in our sample, we have N-1=10 degrees of freedom. We set our significance level to 95% and compute the t statistic using the empirical mean and empirical standard deviation of our intake. We use a ddof of 1 to base the computation of our empirical standard deviation on an unbiased estimate of the variance (note: the final estimate is not unbiased due to the concave nature of the square root). >>> np.mean(intake) 6753.636363636364 >>> intake.std(ddof=1) 1142.1232221373727 >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake))) >>> t -2.8207540608310198 We draw 1000000 samples from Student's t distribution with the adequate degrees of freedom. >>> import matplotlib.pyplot as plt >>> s = np.random.standard_t(10, size=1000000) >>> h = plt.hist(s, bins=100, density=True) Does our t statistic land in one of the two critical regions found at both tails of the distribution? >>> np.sum(np.abs(t) < np.abs(s)) / float(len(s)) 0.018318 #random < 0.05, statistic is in critical region The probability value for this 2-tailed test is about 1.83%, which is lower than the 5% pre-determined significance threshold. Therefore, the probability of observing values as extreme as our intake conditionally on the null hypothesis being true is too low, and we reject the null hypothesis of no deviation. state must be a dict or a tuple.sum(pvals[:-1].astype(np.float64)) > 1.0. The pvals array is cast to 64-bit floating point prior to checking the sum. Precision changes when casting may cause problems even if the sum of the original pvals is valid. tomaxint(size=None) Return a sample of uniformly distributed random integers in the interval [0, ``np.iinfo("long").max``]. .. warning:: This function uses the C-long dtype, which is 32bit on windows and otherwise 64bit on 64bit platforms (and 32bit on 32bit ones). Since NumPy 2.0, NumPy's default integer is 32bit on 32bit platforms and 64bit on 64bit platforms. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- out : ndarray Drawn samples, with shape `size`. See Also -------- randint : Uniform sampling over a given half-open interval of integers. random_integers : Uniform sampling over a given closed interval of integers. Examples -------- >>> rs = np.random.RandomState() # need a RandomState object >>> rs.tomaxint((2,2,2)) array([[[1170048599, 1600360186], # random [ 739731006, 1947757578]], [[1871712945, 752307660], [1601631370, 1479324245]]]) >>> rs.tomaxint((2,2,2)) < np.iinfo(np.int_).max array([[[ True, True], [ True, True]], [[ True, True], [ True, True]]]) triangular(left, mode, right, size=None) Draw samples from the triangular distribution over the interval ``[left, right]``. The triangular distribution is a continuous probability distribution with lower limit left, peak at mode, and upper limit right. Unlike the other distributions, these parameters directly define the shape of the pdf. .. note:: New code should use the `~numpy.random.Generator.triangular` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- left : float or array_like of floats Lower limit. mode : float or array_like of floats The value where the peak of the distribution occurs. The value must fulfill the condition ``left <= mode <= right``. right : float or array_like of floats Upper limit, must be larger than `left`. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``left``, ``mode``, and ``right`` are all scalars. Otherwise, ``np.broadcast(left, mode, right).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized triangular distribution. See Also -------- random.Generator.triangular: which should be used for new code. Notes ----- The probability density function for the triangular distribution is .. math:: P(x;l, m, r) = \begin{cases} \frac{2(x-l)}{(r-l)(m-l)}& \text{for $l \leq x \leq m$},\\ \frac{2(r-x)}{(r-l)(r-m)}& \text{for $m \leq x \leq r$},\\ 0& \text{otherwise}. \end{cases} The triangular distribution is often used in ill-defined problems where the underlying distribution is not known, but some knowledge of the limits and mode exists. Often it is used in simulations. References ---------- .. [1] Wikipedia, "Triangular distribution" https://en.wikipedia.org/wiki/Triangular_distribution Examples -------- Draw values from the distribution and plot the histogram: >>> import matplotlib.pyplot as plt >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=200, ... density=True) >>> plt.show() uniform(low=0.0, high=1.0, size=None) Draw samples from a uniform distribution. Samples are uniformly distributed over the half-open interval ``[low, high)`` (includes low, but excludes high). In other words, any value within the given interval is equally likely to be drawn by `uniform`. .. note:: New code should use the `~numpy.random.Generator.uniform` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- low : float or array_like of floats, optional Lower boundary of the output interval. All values generated will be greater than or equal to low. The default value is 0. high : float or array_like of floats Upper boundary of the output interval. All values generated will be less than or equal to high. The high limit may be included in the returned array of floats due to floating-point rounding in the equation ``low + (high-low) * random_sample()``. The default value is 1.0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``low`` and ``high`` are both scalars. Otherwise, ``np.broadcast(low, high).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized uniform distribution. See Also -------- randint : Discrete uniform distribution, yielding integers. random_integers : Discrete uniform distribution over the closed interval ``[low, high]``. random_sample : Floats uniformly distributed over ``[0, 1)``. random : Alias for `random_sample`. rand : Convenience function that accepts dimensions as input, e.g., ``rand(2,2)`` would generate a 2-by-2 array of floats, uniformly distributed over ``[0, 1)``. random.Generator.uniform: which should be used for new code. Notes ----- The probability density function of the uniform distribution is .. math:: p(x) = \frac{1}{b - a} anywhere within the interval ``[a, b)``, and zero elsewhere. When ``high`` == ``low``, values of ``low`` will be returned. If ``high`` < ``low``, the results are officially undefined and may eventually raise an error, i.e. do not rely on this function to behave when passed arguments satisfying that inequality condition. The ``high`` limit may be included in the returned array of floats due to floating-point rounding in the equation ``low + (high-low) * random_sample()``. For example: >>> x = np.float32(5*0.99999999) >>> x np.float32(5.0) Examples -------- Draw samples from the distribution: >>> s = np.random.uniform(-1,0,1000) All values are within the given interval: >>> np.all(s >= -1) True >>> np.all(s < 0) True Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 15, density=True) >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r') >>> plt.show() vonmises(mu, kappa, size=None) Draw samples from a von Mises distribution. Samples are drawn from a von Mises distribution with specified mode (mu) and concentration (kappa), on the interval [-pi, pi]. The von Mises distribution (also known as the circular normal distribution) is a continuous probability distribution on the unit circle. It may be thought of as the circular analogue of the normal distribution. .. note:: New code should use the `~numpy.random.Generator.vonmises` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mu : float or array_like of floats Mode ("center") of the distribution. kappa : float or array_like of floats Concentration of the distribution, has to be >=0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mu`` and ``kappa`` are both scalars. Otherwise, ``np.broadcast(mu, kappa).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized von Mises distribution. See Also -------- scipy.stats.vonmises : probability density function, distribution, or cumulative density function, etc. random.Generator.vonmises: which should be used for new code. Notes ----- The probability density for the von Mises distribution is .. math:: p(x) = \frac{e^{\kappa cos(x-\mu)}}{2\pi I_0(\kappa)}, where :math:`\mu` is the mode and :math:`\kappa` the concentration, and :math:`I_0(\kappa)` is the modified Bessel function of order 0. The von Mises is named for Richard Edler von Mises, who was born in Austria-Hungary, in what is now the Ukraine. He fled to the United States in 1939 and became a professor at Harvard. He worked in probability theory, aerodynamics, fluid mechanics, and philosophy of science. References ---------- .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). "Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, 9th printing," New York: Dover, 1972. .. [2] von Mises, R., "Mathematical Theory of Probability and Statistics", New York: Academic Press, 1964. Examples -------- Draw samples from the distribution: >>> mu, kappa = 0.0, 4.0 # mean and concentration >>> s = np.random.vonmises(mu, kappa, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> from scipy.special import i0 # doctest: +SKIP >>> plt.hist(s, 50, density=True) >>> x = np.linspace(-np.pi, np.pi, num=51) >>> y = np.exp(kappa*np.cos(x-mu))/(2*np.pi*i0(kappa)) # doctest: +SKIP >>> plt.plot(x, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() weibull(a, size=None) Draw samples from a Weibull distribution. Draw samples from a 1-parameter Weibull distribution with the given shape parameter `a`. .. math:: X = (-ln(U))^{1/a} Here, U is drawn from the uniform distribution over (0,1]. The more common 2-parameter Weibull, including a scale parameter :math:`\lambda` is just :math:`X = \lambda(-ln(U))^{1/a}`. .. note:: New code should use the `~numpy.random.Generator.weibull` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Shape parameter of the distribution. Must be nonnegative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Weibull distribution. See Also -------- scipy.stats.weibull_max scipy.stats.weibull_min scipy.stats.genextreme gumbel random.Generator.weibull: which should be used for new code. Notes ----- The Weibull (or Type III asymptotic extreme value distribution for smallest values, SEV Type III, or Rosin-Rammler distribution) is one of a class of Generalized Extreme Value (GEV) distributions used in modeling extreme value problems. This class includes the Gumbel and Frechet distributions. The probability density for the Weibull distribution is .. math:: p(x) = \frac{a} {\lambda}(\frac{x}{\lambda})^{a-1}e^{-(x/\lambda)^a}, where :math:`a` is the shape and :math:`\lambda` the scale. The function has its peak (the mode) at :math:`\lambda(\frac{a-1}{a})^{1/a}`. When ``a = 1``, the Weibull distribution reduces to the exponential distribution. References ---------- .. [1] Waloddi Weibull, Royal Technical University, Stockholm, 1939 "A Statistical Theory Of The Strength Of Materials", Ingeniorsvetenskapsakademiens Handlingar Nr 151, 1939, Generalstabens Litografiska Anstalts Forlag, Stockholm. .. [2] Waloddi Weibull, "A Statistical Distribution Function of Wide Applicability", Journal Of Applied Mechanics ASME Paper 1951. .. [3] Wikipedia, "Weibull distribution", https://en.wikipedia.org/wiki/Weibull_distribution Examples -------- Draw samples from the distribution: >>> a = 5. # shape >>> s = np.random.weibull(a, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> x = np.arange(1,100.)/50. >>> def weib(x,n,a): ... return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a) >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000)) >>> x = np.arange(1,100.)/50. >>> scale = count.max()/weib(x, 1., 5.).max() >>> plt.plot(x, weib(x, 1., 5.)*scale) >>> plt.show() "!z !qa /q_AJfM'A/q]!JfM'A 4q "!7%q "!7%q uG1,bavRq-QgQ,bar1 M! ))7! !"HAWAEaq;&=Qd#Qa++B!4s!15Qd#Qa.aqJe1/0A-Qa 1aqae1e1 51Qv31AT545T,bavRq Qazaq  U!1AU"8$jQ9:$aq zipf(a, size=None) Draw samples from a Zipf distribution. Samples are drawn from a Zipf distribution with specified parameter `a` > 1. The Zipf distribution (also known as the zeta distribution) is a discrete probability distribution that satisfies Zipf's law: the frequency of an item is inversely proportional to its rank in a frequency table. .. note:: New code should use the `~numpy.random.Generator.zipf` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Distribution parameter. Must be greater than 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Zipf distribution. See Also -------- scipy.stats.zipf : probability density function, distribution, or cumulative density function, etc. random.Generator.zipf: which should be used for new code. Notes ----- The probability mass function (PMF) for the Zipf distribution is .. math:: p(k) = \frac{k^{-a}}{\zeta(a)}, for integers :math:`k \geq 1`, where :math:`\zeta` is the Riemann Zeta function. It is named for the American linguist George Kingsley Zipf, who noted that the frequency of any word in a sample of a language is inversely proportional to its rank in the frequency table. References ---------- .. [1] Zipf, G. K., "Selected Studies of the Principle of Relative Frequency in Language," Cambridge, MA: Harvard Univ. Press, 1932. Examples -------- Draw samples from the distribution: >>> a = 4.0 >>> n = 20000 >>> s = np.random.zipf(a, n) Display the histogram of the samples, along with the expected histogram based on the probability density function: >>> import matplotlib.pyplot as plt >>> from scipy.special import zeta # doctest: +SKIP `bincount` provides a fast histogram for small integers. >>> count = np.bincount(s) >>> k = np.arange(1, s.max() + 1) >>> plt.bar(k, count[1:], alpha=0.5, label='sample count') >>> plt.plot(k, n*(k**-a)/zeta(a), 'k.-', alpha=0.5, ... label='expected count') # doctest: +SKIP >>> plt.semilogy() >>> plt.grid(alpha=0.4) >>> plt.legend() >>> plt.title(f'Zipf sample, a={a}, size={n}') >>> plt.show() A\ ##Qa :Qe:TQfA *AQ 4q5 QfCrQa nAQ Qhaq qa "F!:V2Q aqQ9Cq Qc)>hgQ Qc87! qQ qc qc4qfCrQ&[ "Kq#QQE%q1qZqrAqqQe1AQQe1 t:QcQ/t1Ba "AQE%q1qZqQd!1E!4qAh :QgQ wfCxwaj  t:QhgQiq5Cq*AQ'uAQ uATQa3awbae1Aa{%qA Ky$ay K}Bd!= O9ACannot take a larger sample than population when 'replace=False'Fewer non-zero entries in p than sizeInvalid bit generator. The bit generator must be instantized.Negative dimensions are not allowedProviding a dtype with a non-native byteorder is not supported. If you require platform-independent byteorder, call byteswap when required. In future version, providing byteorder will raise a ValueErrorRandomState.exponential (line 504)RandomState.multinomial (line 4271)RandomState.negative_binomial (line 3517) permutation(x) Randomly permute a sequence, or return a permuted range. If `x` is a multi-dimensional array, it is only shuffled along its first index. .. note:: New code should use the `~numpy.random.Generator.permutation` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- x : int or array_like If `x` is an integer, randomly permute ``np.arange(x)``. If `x` is an array, make a copy and shuffle the elements randomly. Returns ------- out : ndarray Permuted sequence or array range. See Also -------- random.Generator.permutation: which should be used for new code. Examples -------- >>> np.random.permutation(10) array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6]) # random >>> np.random.permutation([1, 4, 9, 12, 15]) array([15, 1, 9, 4, 12]) # random >>> arr = np.arange(9).reshape((3, 3)) >>> np.random.permutation(arr) array([[6, 7, 8], # random [0, 1, 2], [3, 4, 5]]) RandomState.noncentral_chisquareRandomState.noncentral_f (line 1841)RandomState.permutation (line 4689)RandomState.random_integers (line 1308)RandomState.random_sample (line 389)RandomState.standard_cauchy (line 2087)RandomState.standard_exponentialpermutationRandomState.standard_normal (line 1403)RandomState.standard_t (line 2162)RandomState.triangular (line 3256)Shuffling a one dimensional array subclass containing objects gives incorrect results for most array subclasses. Please use the new random number API instead: https://numpy.org/doc/stable/reference/random/index.html The new API fixes this issue. This version will not be fixed due to stability guarantees of the API.This function is deprecated. Please call randint(1, a must be 1-dimensional or an integercan only re-seed a MT19937 BitGeneratorcheck_valid must equal 'warn', 'raise', or 'ignore'cov must be 2 dimensional and squarecovariance is not symmetric positive-semidefinite.get_state and legacy can only be used with the MT19937 BitGenerator. To silence this warning, set `legacy` to False.legacy can only be True when the underlying bitgenerator is an instance of MT19937.mean and cov must have same lengthnumpy._core.multiarray failed to importnumpy._core.umath failed to importprobabilities are not non-negativex must be an integer or at least 1-dimensionalRandomState.multivariate_normal (line 4072)RandomState.noncentral_chisquare (line 2001)RandomState.standard_exponential (line 581)This function is deprecated. Please call randint( f(dfnum, dfden, size=None) Draw samples from an F distribution. Samples are drawn from an F distribution with specified parameters, `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom in denominator), where both parameters must be greater than zero. The random variate of the F distribution (also known as the Fisher distribution) is a continuous probability distribution that arises in ANOVA tests, and is the ratio of two chi-square variates. .. note:: New code should use the `~numpy.random.Generator.f` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- dfnum : float or array_like of floats Degrees of freedom in numerator, must be > 0. dfden : float or array_like of float Degrees of freedom in denominator, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``dfnum`` and ``dfden`` are both scalars. Otherwise, ``np.broadcast(dfnum, dfden).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Fisher distribution. See Also -------- scipy.stats.f : probability density function, distribution or cumulative density function, etc. random.Generator.f: which should be used for new code. Notes ----- The F statistic is used to compare in-group variances to between-group variances. Calculating the distribution depends on the sampling, and so it is a function of the respective degrees of freedom in the problem. The variable `dfnum` is the number of samples minus one, the between-groups degrees of freedom, while `dfden` is the within-groups degrees of freedom, the sum of the number of samples in each group minus the number of groups. References ---------- .. [1] Glantz, Stanton A. "Primer of Biostatistics.", McGraw-Hill, Fifth Edition, 2002. .. [2] Wikipedia, "F-distribution", https://en.wikipedia.org/wiki/F-distribution Examples -------- An example from Glantz[1], pp 47-40: Two groups, children of diabetics (25 people) and children from people without diabetes (25 controls). Fasting blood glucose was measured, case group had a mean value of 86.1, controls had a mean value of 82.2. Standard deviations were 2.09 and 2.49 respectively. Are these data consistent with the null hypothesis that the parents diabetic status does not affect their children's blood glucose levels? Calculating the F statistic from the data gives a value of 36.01. Draw samples from the distribution: >>> dfnum = 1. # between group degrees of freedom >>> dfden = 48. # within groups degrees of freedom >>> s = np.random.f(dfnum, dfden, 1000) The lower bound for the top 1% of the samples is : >>> np.sort(s)[-10] 7.61988120985 # random So there is about a 1% chance that the F statistic will exceed 7.62, the measured value is 36, so the null hypothesis is rejected at the 1% level. f noncentral_f(dfnum, dfden, nonc, size=None) Draw samples from the noncentral F distribution. Samples are drawn from an F distribution with specified parameters, `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom in denominator), where both parameters > 1. `nonc` is the non-centrality parameter. .. note:: New code should use the `~numpy.random.Generator.noncentral_f` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- dfnum : float or array_like of floats Numerator degrees of freedom, must be > 0. dfden : float or array_like of floats Denominator degrees of freedom, must be > 0. nonc : float or array_like of floats Non-centrality parameter, the sum of the squares of the numerator means, must be >= 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``dfnum``, ``dfden``, and ``nonc`` are all scalars. Otherwise, ``np.broadcast(dfnum, dfden, nonc).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized noncentral Fisher distribution. See Also -------- random.Generator.noncentral_f: which should be used for new code. Notes ----- When calculating the power of an experiment (power = probability of rejecting the null hypothesis when a specific alternative is true) the non-central F statistic becomes important. When the null hypothesis is true, the F statistic follows a central F distribution. When the null hypothesis is not true, then it follows a non-central F statistic. References ---------- .. [1] Weisstein, Eric W. "Noncentral F-Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/NoncentralF-Distribution.html .. [2] Wikipedia, "Noncentral F-distribution", https://en.wikipedia.org/wiki/Noncentral_F-distribution Examples -------- In a study, testing for a specific alternative to the null hypothesis requires use of the Noncentral F distribution. We need to calculate the area in the tail of the distribution that exceeds the value of the F distribution for the null hypothesis. We'll plot the two probability distributions for comparison. >>> dfnum = 3 # between group deg of freedom >>> dfden = 20 # within groups degrees of freedom >>> nonc = 3.0 >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000) >>> NF = np.histogram(nc_vals, bins=50, density=True) >>> c_vals = np.random.f(dfnum, dfden, 1000000) >>> F = np.histogram(c_vals, bins=50, density=True) >>> import matplotlib.pyplot as plt >>> plt.plot(F[1][1:], F[0]) >>> plt.plot(NF[1][1:], NF[0]) >>> plt.show() noncentral_f chisquare(df, size=None) Draw samples from a chi-square distribution. When `df` independent random variables, each with standard normal distributions (mean 0, variance 1), are squared and summed, the resulting distribution is chi-square (see Notes). This distribution is often used in hypothesis testing. .. note:: New code should use the `~numpy.random.Generator.chisquare` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- df : float or array_like of floats Number of degrees of freedom, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``df`` is a scalar. Otherwise, ``np.array(df).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized chi-square distribution. Raises ------ ValueError When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``) is given. See Also -------- random.Generator.chisquare: which should be used for new code. Notes ----- The variable obtained by summing the squares of `df` independent, standard normally distributed random variables: .. math:: Q = \sum_{i=1}^{\mathtt{df}} X^2_i is chi-square distributed, denoted .. math:: Q \sim \chi^2_k. The probability density function of the chi-squared distribution is .. math:: p(x) = \frac{(1/2)^{k/2}}{\Gamma(k/2)} x^{k/2 - 1} e^{-x/2}, where :math:`\Gamma` is the gamma function, .. math:: \Gamma(x) = \int_0^{-\infty} t^{x - 1} e^{-t} dt. References ---------- .. [1] NIST "Engineering Statistics Handbook" https://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm Examples -------- >>> np.random.chisquare(2,4) array([ 1.89920014, 9.00867716, 3.13710533, 5.62318272]) # random chisquare noncentral_chisquare(df, nonc, size=None) Draw samples from a noncentral chi-square distribution. The noncentral :math:`\chi^2` distribution is a generalization of the :math:`\chi^2` distribution. .. note:: New code should use the `~numpy.random.Generator.noncentral_chisquare` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- df : float or array_like of floats Degrees of freedom, must be > 0. nonc : float or array_like of floats Non-centrality, must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``df`` and ``nonc`` are both scalars. Otherwise, ``np.broadcast(df, nonc).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized noncentral chi-square distribution. See Also -------- random.Generator.noncentral_chisquare: which should be used for new code. Notes ----- The probability density function for the noncentral Chi-square distribution is .. math:: P(x;df,nonc) = \sum^{\infty}_{i=0} \frac{e^{-nonc/2}(nonc/2)^{i}}{i!} P_{Y_{df+2i}}(x), where :math:`Y_{q}` is the Chi-square with q degrees of freedom. References ---------- .. [1] Wikipedia, "Noncentral chi-squared distribution" https://en.wikipedia.org/wiki/Noncentral_chi-squared_distribution Examples -------- Draw values from the distribution and plot the histogram >>> import matplotlib.pyplot as plt >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000), ... bins=200, density=True) >>> plt.show() Draw values from a noncentral chisquare with very small noncentrality, and compare to a chisquare. >>> plt.figure() >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000), ... bins=np.arange(0., 25, .1), density=True) >>> values2 = plt.hist(np.random.chisquare(3, 100000), ... bins=np.arange(0., 25, .1), density=True) >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob') >>> plt.show() Demonstrate how large values of non-centrality lead to a more symmetric distribution. >>> plt.figure() >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000), ... bins=200, density=True) >>> plt.show() noncentral_chisquare seed(seed=None) Reseed the singleton RandomState instance. Notes ----- This is a convenience, legacy function that exists to support older code that uses the singleton RandomState. Best practice is to use a dedicated ``Generator`` instance rather than the random variate generation methods exposed directly in the random module. See Also -------- numpy.random.Generator standard_cauchy(size=None) Draw samples from a standard Cauchy distribution with mode = 0. Also known as the Lorentz distribution. .. note:: New code should use the `~numpy.random.Generator.standard_cauchy` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. Default is None, in which case a single value is returned. Returns ------- samples : ndarray or scalar The drawn samples. See Also -------- random.Generator.standard_cauchy: which should be used for new code. Notes ----- The probability density function for the full Cauchy distribution is .. math:: P(x; x_0, \gamma) = \frac{1}{\pi \gamma \bigl[ 1+ (\frac{x-x_0}{\gamma})^2 \bigr] } and the Standard Cauchy distribution just sets :math:`x_0=0` and :math:`\gamma=1` The Cauchy distribution arises in the solution to the driven harmonic oscillator problem, and also describes spectral line broadening. It also describes the distribution of values at which a line tilted at a random angle will cut the x axis. When studying hypothesis tests that assume normality, seeing how the tests perform on data from a Cauchy distribution is a good indicator of their sensitivity to a heavy-tailed distribution, since the Cauchy looks very much like a Gaussian distribution, but with heavier tails. References ---------- .. [1] NIST/SEMATECH e-Handbook of Statistical Methods, "Cauchy Distribution", https://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm .. [2] Weisstein, Eric W. "Cauchy Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/CauchyDistribution.html .. [3] Wikipedia, "Cauchy distribution" https://en.wikipedia.org/wiki/Cauchy_distribution Examples -------- Draw samples and plot the distribution: >>> import matplotlib.pyplot as plt >>> s = np.random.standard_cauchy(1000000) >>> s = s[(s>-25) & (s<25)] # truncate distribution so it plots well >>> plt.hist(s, bins=100) >>> plt.show() standard_cauchy standard_t(df, size=None) Draw samples from a standard Student's t distribution with `df` degrees of freedom. A special case of the hyperbolic distribution. As `df` gets large, the result resembles that of the standard normal distribution (`standard_normal`). .. note:: New code should use the `~numpy.random.Generator.standard_t` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- df : float or array_like of floats Degrees of freedom, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``df`` is a scalar. Otherwise, ``np.array(df).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized standard Student's t distribution. See Also -------- random.Generator.standard_t: which should be used for new code. Notes ----- The probability density function for the t distribution is .. math:: P(x, df) = \frac{\Gamma(\frac{df+1}{2})}{\sqrt{\pi df} \Gamma(\frac{df}{2})}\Bigl( 1+\frac{x^2}{df} \Bigr)^{-(df+1)/2} The t test is based on an assumption that the data come from a Normal distribution. The t test provides a way to test whether the sample mean (that is the mean calculated from the data) is a good estimate of the true mean. The derivation of the t-distribution was first published in 1908 by William Gosset while working for the Guinness Brewery in Dublin. Due to proprietary issues, he had to publish under a pseudonym, and so he used the name Student. References ---------- .. [1] Dalgaard, Peter, "Introductory Statistics With R", Springer, 2002. .. [2] Wikipedia, "Student's t-distribution" https://en.wikipedia.org/wiki/Student's_t-distribution Examples -------- From Dalgaard page 83 [1]_, suppose the daily energy intake for 11 women in kilojoules (kJ) is: >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \ ... 7515, 8230, 8770]) Does their energy intake deviate systematically from the recommended value of 7725 kJ? Our null hypothesis will be the absence of deviation, and the alternate hypothesis will be the presence of an effect that could be either positive or negative, hence making our test 2-tailed. Because we are estimating the mean and we have N=11 values in our sample, we have N-1=10 degrees of freedom. We set our significance level to 95% and compute the t statistic using the empirical mean and empirical standard deviation of our intake. We use a ddof of 1 to base the computation of our empirical standard deviation on an unbiased estimate of the variance (note: the final estimate is not unbiased due to the concave nature of the square root). >>> np.mean(intake) 6753.636363636364 >>> intake.std(ddof=1) 1142.1232221373727 >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake))) >>> t -2.8207540608310198 We draw 1000000 samples from Student's t distribution with the adequate degrees of freedom. >>> import matplotlib.pyplot as plt >>> s = np.random.standard_t(10, size=1000000) >>> h = plt.hist(s, bins=100, density=True) Does our t statistic land in one of the two critical regions found at both tails of the distribution? >>> np.sum(np.abs(t) < np.abs(s)) / float(len(s)) 0.018318 #random < 0.05, statistic is in critical region The probability value for this 2-tailed test is about 1.83%, which is lower than the 5% pre-determined significance threshold. Therefore, the probability of observing values as extreme as our intake conditionally on the null hypothesis being true is too low, and we reject the null hypothesis of no deviation. standard_t vonmises(mu, kappa, size=None) Draw samples from a von Mises distribution. Samples are drawn from a von Mises distribution with specified mode (mu) and concentration (kappa), on the interval [-pi, pi]. The von Mises distribution (also known as the circular normal distribution) is a continuous probability distribution on the unit circle. It may be thought of as the circular analogue of the normal distribution. .. note:: New code should use the `~numpy.random.Generator.vonmises` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mu : float or array_like of floats Mode ("center") of the distribution. kappa : float or array_like of floats Concentration of the distribution, has to be >=0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mu`` and ``kappa`` are both scalars. Otherwise, ``np.broadcast(mu, kappa).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized von Mises distribution. See Also -------- scipy.stats.vonmises : probability density function, distribution, or cumulative density function, etc. random.Generator.vonmises: which should be used for new code. Notes ----- The probability density for the von Mises distribution is .. math:: p(x) = \frac{e^{\kappa cos(x-\mu)}}{2\pi I_0(\kappa)}, where :math:`\mu` is the mode and :math:`\kappa` the concentration, and :math:`I_0(\kappa)` is the modified Bessel function of order 0. The von Mises is named for Richard Edler von Mises, who was born in Austria-Hungary, in what is now the Ukraine. He fled to the United States in 1939 and became a professor at Harvard. He worked in probability theory, aerodynamics, fluid mechanics, and philosophy of science. References ---------- .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). "Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, 9th printing," New York: Dover, 1972. .. [2] von Mises, R., "Mathematical Theory of Probability and Statistics", New York: Academic Press, 1964. Examples -------- Draw samples from the distribution: >>> mu, kappa = 0.0, 4.0 # mean and concentration >>> s = np.random.vonmises(mu, kappa, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> from scipy.special import i0 # doctest: +SKIP >>> plt.hist(s, 50, density=True) >>> x = np.linspace(-np.pi, np.pi, num=51) >>> y = np.exp(kappa*np.cos(x-mu))/(2*np.pi*i0(kappa)) # doctest: +SKIP >>> plt.plot(x, y, linewidth=2, color='r') # doctest: +SKIP >>> plt.show() vonmises Returns the singleton RandomState's bit generator Returns ------- BitGenerator The bit generator that underlies the singleton RandomState instance Notes ----- The singleton RandomState provides the random variate generators in the ``numpy.random`` namespace. This function, and its counterpart set method, provides a path to hot-swap the default MT19937 bit generator with a user provided alternative. These function are intended to provide a continuous path where a single underlying bit generator can be used both with an instance of ``Generator`` and with the singleton instance of RandomState. See Also -------- set_bit_generator numpy.random.Generator get_bit_generator Sets the singleton RandomState's bit generator Parameters ---------- bitgen A bit generator instance Notes ----- The singleton RandomState provides the random variate generators in the ``numpy.random``namespace. This function, and its counterpart get method, provides a path to hot-swap the default MT19937 bit generator with a user provided alternative. These function are intended to provide a continuous path where a single underlying bit generator can be used both with an instance of ``Generator`` and with the singleton instance of RandomState. See Also -------- get_bit_generator numpy.random.Generator set_bit_generator pareto(a, size=None) Draw samples from a Pareto II or Lomax distribution with specified shape. The Lomax or Pareto II distribution is a shifted Pareto distribution. The classical Pareto distribution can be obtained from the Lomax distribution by adding 1 and multiplying by the scale parameter ``m`` (see Notes). The smallest value of the Lomax distribution is zero while for the classical Pareto distribution it is ``mu``, where the standard Pareto distribution has location ``mu = 1``. Lomax can also be considered as a simplified version of the Generalized Pareto distribution (available in SciPy), with the scale set to one and the location set to zero. The Pareto distribution must be greater than zero, and is unbounded above. It is also known as the "80-20 rule". In this distribution, 80 percent of the weights are in the lowest 20 percent of the range, while the other 20 percent fill the remaining 80 percent of the range. .. note:: New code should use the `~numpy.random.Generator.pareto` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Shape of the distribution. Must be positive. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Pareto distribution. See Also -------- scipy.stats.lomax : probability density function, distribution or cumulative density function, etc. scipy.stats.genpareto : probability density function, distribution or cumulative density function, etc. random.Generator.pareto: which should be used for new code. Notes ----- The probability density for the Pareto distribution is .. math:: p(x) = \frac{am^a}{x^{a+1}} where :math:`a` is the shape and :math:`m` the scale. The Pareto distribution, named after the Italian economist Vilfredo Pareto, is a power law probability distribution useful in many real world problems. Outside the field of economics it is generally referred to as the Bradford distribution. Pareto developed the distribution to describe the distribution of wealth in an economy. It has also found use in insurance, web page access statistics, oil field sizes, and many other problems, including the download frequency for projects in Sourceforge [1]_. It is one of the so-called "fat-tailed" distributions. References ---------- .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of Sourceforge projects. .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne. .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme Values, Birkhauser Verlag, Basel, pp 23-30. .. [4] Wikipedia, "Pareto distribution", https://en.wikipedia.org/wiki/Pareto_distribution Examples -------- Draw samples from the distribution: >>> a, m = 3., 2. # shape and mode >>> s = (np.random.pareto(a, 1000) + 1) * m Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, _ = plt.hist(s, 100, density=True) >>> fit = a*m**a / bins**(a+1) >>> plt.plot(bins, max(count)*fit/max(fit), linewidth=2, color='r') >>> plt.show() pareto This is an alias of `random_sample`. See `random_sample` for the complete documentation. sample weibull(a, size=None) Draw samples from a Weibull distribution. Draw samples from a 1-parameter Weibull distribution with the given shape parameter `a`. .. math:: X = (-ln(U))^{1/a} Here, U is drawn from the uniform distribution over (0,1]. The more common 2-parameter Weibull, including a scale parameter :math:`\lambda` is just :math:`X = \lambda(-ln(U))^{1/a}`. .. note:: New code should use the `~numpy.random.Generator.weibull` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Shape parameter of the distribution. Must be nonnegative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Weibull distribution. See Also -------- scipy.stats.weibull_max scipy.stats.weibull_min scipy.stats.genextreme gumbel random.Generator.weibull: which should be used for new code. Notes ----- The Weibull (or Type III asymptotic extreme value distribution for smallest values, SEV Type III, or Rosin-Rammler distribution) is one of a class of Generalized Extreme Value (GEV) distributions used in modeling extreme value problems. This class includes the Gumbel and Frechet distributions. The probability density for the Weibull distribution is .. math:: p(x) = \frac{a} {\lambda}(\frac{x}{\lambda})^{a-1}e^{-(x/\lambda)^a}, where :math:`a` is the shape and :math:`\lambda` the scale. The function has its peak (the mode) at :math:`\lambda(\frac{a-1}{a})^{1/a}`. When ``a = 1``, the Weibull distribution reduces to the exponential distribution. References ---------- .. [1] Waloddi Weibull, Royal Technical University, Stockholm, 1939 "A Statistical Theory Of The Strength Of Materials", Ingeniorsvetenskapsakademiens Handlingar Nr 151, 1939, Generalstabens Litografiska Anstalts Forlag, Stockholm. .. [2] Waloddi Weibull, "A Statistical Distribution Function of Wide Applicability", Journal Of Applied Mechanics ASME Paper 1951. .. [3] Wikipedia, "Weibull distribution", https://en.wikipedia.org/wiki/Weibull_distribution Examples -------- Draw samples from the distribution: >>> a = 5. # shape >>> s = np.random.weibull(a, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> x = np.arange(1,100.)/50. >>> def weib(x,n,a): ... return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a) >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000)) >>> x = np.arange(1,100.)/50. >>> scale = count.max()/weib(x, 1., 5.).max() >>> plt.plot(x, weib(x, 1., 5.)*scale) >>> plt.show() weibull This is an alias of `random_sample`. See `random_sample` for the complete documentation. ranf power(a, size=None) Draws samples in [0, 1] from a power distribution with positive exponent a - 1. Also known as the power function distribution. .. note:: New code should use the `~numpy.random.Generator.power` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- a : float or array_like of floats Parameter of the distribution. Must be non-negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``a`` is a scalar. Otherwise, ``np.array(a).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized power distribution. Raises ------ ValueError If a <= 0. See Also -------- random.Generator.power: which should be used for new code. Notes ----- The probability density function is .. math:: P(x; a) = ax^{a-1}, 0 \le x \le 1, a>0. The power function distribution is just the inverse of the Pareto distribution. It may also be seen as a special case of the Beta distribution. It is used, for example, in modeling the over-reporting of insurance claims. References ---------- .. [1] Christian Kleiber, Samuel Kotz, "Statistical size distributions in economics and actuarial sciences", Wiley, 2003. .. [2] Heckert, N. A. and Filliben, James J. "NIST Handbook 148: Dataplot Reference Manual, Volume 2: Let Subcommands and Library Functions", National Institute of Standards and Technology Handbook Series, June 2003. https://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf Examples -------- Draw samples from the distribution: >>> a = 5. # shape >>> samples = 1000 >>> s = np.random.power(a, samples) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, bins=30) >>> x = np.linspace(0, 1, 100) >>> y = a*x**(a-1.) >>> normed_y = samples*np.diff(bins)[0]*y >>> plt.plot(x, normed_y) >>> plt.show() Compare the power function distribution to the inverse of the Pareto. >>> from scipy import stats # doctest: +SKIP >>> rvs = np.random.power(5, 1000000) >>> rvsp = np.random.pareto(5, 1000000) >>> xx = np.linspace(0,1,100) >>> powpdf = stats.powerlaw.pdf(xx,5) # doctest: +SKIP >>> plt.figure() >>> plt.hist(rvs, bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('np.random.power(5)') >>> plt.figure() >>> plt.hist(1./(1.+rvsp), bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('inverse of 1 + np.random.pareto(5)') >>> plt.figure() >>> plt.hist(1./(1.+rvsp), bins=50, density=True) >>> plt.plot(xx,powpdf,'r-') # doctest: +SKIP >>> plt.title('inverse of stats.pareto(5)') power laplace(loc=0.0, scale=1.0, size=None) Draw samples from the Laplace or double exponential distribution with specified location (or mean) and scale (decay). The Laplace distribution is similar to the Gaussian/normal distribution, but is sharper at the peak and has fatter tails. It represents the difference between two independent, identically distributed exponential random variables. .. note:: New code should use the `~numpy.random.Generator.laplace` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional The position, :math:`\mu`, of the distribution peak. Default is 0. scale : float or array_like of floats, optional :math:`\lambda`, the exponential decay. Default is 1. Must be non- negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Laplace distribution. See Also -------- random.Generator.laplace: which should be used for new code. Notes ----- It has the probability density function .. math:: f(x; \mu, \lambda) = \frac{1}{2\lambda} \exp\left(-\frac{|x - \mu|}{\lambda}\right). The first law of Laplace, from 1774, states that the frequency of an error can be expressed as an exponential function of the absolute magnitude of the error, which leads to the Laplace distribution. For many problems in economics and health sciences, this distribution seems to model the data better than the standard Gaussian distribution. References ---------- .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). "Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, 9th printing," New York: Dover, 1972. .. [2] Kotz, Samuel, et. al. "The Laplace Distribution and Generalizations, " Birkhauser, 2001. .. [3] Weisstein, Eric W. "Laplace Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/LaplaceDistribution.html .. [4] Wikipedia, "Laplace distribution", https://en.wikipedia.org/wiki/Laplace_distribution Examples -------- Draw samples from the distribution >>> loc, scale = 0., 1. >>> s = np.random.laplace(loc, scale, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> x = np.arange(-8., 8., .01) >>> pdf = np.exp(-abs(x-loc)/scale)/(2.*scale) >>> plt.plot(x, pdf) Plot Gaussian for comparison: >>> g = (1/(scale * np.sqrt(2 * np.pi)) * ... np.exp(-(x - loc)**2 / (2 * scale**2))) >>> plt.plot(x,g) laplace__getstate____setstate__ gumbel(loc=0.0, scale=1.0, size=None) Draw samples from a Gumbel distribution. Draw samples from a Gumbel distribution with specified location and scale. For more information on the Gumbel distribution, see Notes and References below. .. note:: New code should use the `~numpy.random.Generator.gumbel` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional The location of the mode of the distribution. Default is 0. scale : float or array_like of floats, optional The scale parameter of the distribution. Default is 1. Must be non- negative. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Gumbel distribution. See Also -------- scipy.stats.gumbel_l scipy.stats.gumbel_r scipy.stats.genextreme weibull random.Generator.gumbel: which should be used for new code. Notes ----- The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value Type I) distribution is one of a class of Generalized Extreme Value (GEV) distributions used in modeling extreme value problems. The Gumbel is a special case of the Extreme Value Type I distribution for maximums from distributions with "exponential-like" tails. The probability density for the Gumbel distribution is .. math:: p(x) = \frac{e^{-(x - \mu)/ \beta}}{\beta} e^{ -e^{-(x - \mu)/ \beta}}, where :math:`\mu` is the mode, a location parameter, and :math:`\beta` is the scale parameter. The Gumbel (named for German mathematician Emil Julius Gumbel) was used very early in the hydrology literature, for modeling the occurrence of flood events. It is also used for modeling maximum wind speed and rainfall rates. It is a "fat-tailed" distribution - the probability of an event in the tail of the distribution is larger than if one used a Gaussian, hence the surprisingly frequent occurrence of 100-year floods. Floods were initially modeled as a Gaussian process, which underestimated the frequency of extreme events. It is one of a class of extreme value distributions, the Generalized Extreme Value (GEV) distributions, which also includes the Weibull and Frechet. The function has a mean of :math:`\mu + 0.57721\beta` and a variance of :math:`\frac{\pi^2}{6}\beta^2`. References ---------- .. [1] Gumbel, E. J., "Statistics of Extremes," New York: Columbia University Press, 1958. .. [2] Reiss, R.-D. and Thomas, M., "Statistical Analysis of Extreme Values from Insurance, Finance, Hydrology and Other Fields," Basel: Birkhauser Verlag, 2001. Examples -------- Draw samples from the distribution: >>> mu, beta = 0, 0.1 # location and scale >>> s = np.random.gumbel(mu, beta, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 30, density=True) >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta) ... * np.exp( -np.exp( -(bins - mu) /beta) ), ... linewidth=2, color='r') >>> plt.show() Show how an extreme value distribution can arise from a Gaussian process and compare to a Gaussian: >>> means = [] >>> maxima = [] >>> for i in range(0,1000) : ... a = np.random.normal(mu, beta, 1000) ... means.append(a.mean()) ... maxima.append(a.max()) >>> count, bins, ignored = plt.hist(maxima, 30, density=True) >>> beta = np.std(maxima) * np.sqrt(6) / np.pi >>> mu = np.mean(maxima) - 0.57721*beta >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta) ... * np.exp(-np.exp(-(bins - mu)/beta)), ... linewidth=2, color='r') >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi)) ... * np.exp(-(bins - mu)**2 / (2 * beta**2)), ... linewidth=2, color='g') >>> plt.show() gumbel_bit_generatornumpy.random.mtrand.RandomState RandomState(seed=None) Container for the slow Mersenne Twister pseudo-random number generator. Consider using a different BitGenerator with the Generator container instead. `RandomState` and `Generator` expose a number of methods for generating random numbers drawn from a variety of probability distributions. In addition to the distribution-specific arguments, each method takes a keyword argument `size` that defaults to ``None``. If `size` is ``None``, then a single value is generated and returned. If `size` is an integer, then a 1-D array filled with generated values is returned. If `size` is a tuple, then an array with that shape is filled and returned. **Compatibility Guarantee** A fixed bit generator using a fixed seed and a fixed series of calls to 'RandomState' methods using the same parameters will always produce the same results up to roundoff error except when the values were incorrect. `RandomState` is effectively frozen and will only receive updates that are required by changes in the internals of Numpy. More substantial changes, including algorithmic improvements, are reserved for `Generator`. Parameters ---------- seed : {None, int, array_like, BitGenerator}, optional Random seed used to initialize the pseudo-random number generator or an instantized BitGenerator. If an integer or array, used as a seed for the MT19937 BitGenerator. Values can be any integer between 0 and 2**32 - 1 inclusive, an array (or other sequence) of such integers, or ``None`` (the default). If `seed` is ``None``, then the `MT19937` BitGenerator is initialized by reading data from ``/dev/urandom`` (or the Windows analogue) if available or seed from the clock otherwise. Notes ----- The Python stdlib module "random" also contains a Mersenne Twister pseudo-random number generator with a number of methods that are similar to the ones available in `RandomState`. `RandomState`, besides being NumPy-aware, has the advantage that it provides a much larger number of probability distributions to choose from. See Also -------- Generator MT19937 numpy.random.BitGenerator __reduce__ logistic(loc=0.0, scale=1.0, size=None) Draw samples from a logistic distribution. Samples are drawn from a logistic distribution with specified parameters, loc (location or mean, also median), and scale (>0). .. note:: New code should use the `~numpy.random.Generator.logistic` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- loc : float or array_like of floats, optional Parameter of the distribution. Default is 0. scale : float or array_like of floats, optional Parameter of the distribution. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``loc`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(loc, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized logistic distribution. See Also -------- scipy.stats.logistic : probability density function, distribution or cumulative density function, etc. random.Generator.logistic: which should be used for new code. Notes ----- The probability density for the Logistic distribution is .. math:: P(x) = P(x) = \frac{e^{-(x-\mu)/s}}{s(1+e^{-(x-\mu)/s})^2}, where :math:`\mu` = location and :math:`s` = scale. The Logistic distribution is used in Extreme Value problems where it can act as a mixture of Gumbel distributions, in Epidemiology, and by the World Chess Federation (FIDE) where it is used in the Elo ranking system, assuming the performance of each player is a logistically distributed random variable. References ---------- .. [1] Reiss, R.-D. and Thomas M. (2001), "Statistical Analysis of Extreme Values, from Insurance, Finance, Hydrology and Other Fields," Birkhauser Verlag, Basel, pp 132-133. .. [2] Weisstein, Eric W. "Logistic Distribution." From MathWorld--A Wolfram Web Resource. https://mathworld.wolfram.com/LogisticDistribution.html .. [3] Wikipedia, "Logistic-distribution", https://en.wikipedia.org/wiki/Logistic_distribution Examples -------- Draw samples from the distribution: >>> loc, scale = 10, 1 >>> s = np.random.logistic(loc, scale, 10000) >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, bins=50) # plot against distribution >>> def logist(x, loc, scale): ... return np.exp((loc-x)/scale)/(scale*(1+np.exp((loc-x)/scale))**2) >>> lgst_val = logist(bins, loc, scale) >>> plt.plot(bins, lgst_val * count.max() / lgst_val.max()) >>> plt.show() logistic lognormal(mean=0.0, sigma=1.0, size=None) Draw samples from a log-normal distribution. Draw samples from a log-normal distribution with specified mean, standard deviation, and array shape. Note that the mean and standard deviation are not the values for the distribution itself, but of the underlying normal distribution it is derived from. .. note:: New code should use the `~numpy.random.Generator.lognormal` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : float or array_like of floats, optional Mean value of the underlying normal distribution. Default is 0. sigma : float or array_like of floats, optional Standard deviation of the underlying normal distribution. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mean`` and ``sigma`` are both scalars. Otherwise, ``np.broadcast(mean, sigma).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized log-normal distribution. See Also -------- scipy.stats.lognorm : probability density function, distribution, cumulative density function, etc. random.Generator.lognormal: which should be used for new code. Notes ----- A variable `x` has a log-normal distribution if `log(x)` is normally distributed. The probability density function for the log-normal distribution is: .. math:: p(x) = \frac{1}{\sigma x \sqrt{2\pi}} e^{(-\frac{(ln(x)-\mu)^2}{2\sigma^2})} where :math:`\mu` is the mean and :math:`\sigma` is the standard deviation of the normally distributed logarithm of the variable. A log-normal distribution results if a random variable is the *product* of a large number of independent, identically-distributed variables in the same way that a normal distribution results if the variable is the *sum* of a large number of independent, identically-distributed variables. References ---------- .. [1] Limpert, E., Stahel, W. A., and Abbt, M., "Log-normal Distributions across the Sciences: Keys and Clues," BioScience, Vol. 51, No. 5, May, 2001. https://stat.ethz.ch/~stahel/lognormal/bioscience.pdf .. [2] Reiss, R.D. and Thomas, M., "Statistical Analysis of Extreme Values," Basel: Birkhauser Verlag, 2001, pp. 31-32. Examples -------- Draw samples from the distribution: >>> mu, sigma = 3., 1. # mean and standard deviation >>> s = np.random.lognormal(mu, sigma, 1000) Display the histogram of the samples, along with the probability density function: >>> import matplotlib.pyplot as plt >>> count, bins, ignored = plt.hist(s, 100, density=True, align='mid') >>> x = np.linspace(min(bins), max(bins), 10000) >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2)) ... / (x * sigma * np.sqrt(2 * np.pi))) >>> plt.plot(x, pdf, linewidth=2, color='r') >>> plt.axis('tight') >>> plt.show() Demonstrate that taking the products of random samples from a uniform distribution can be fit well by a log-normal probability density function. >>> # Generate a thousand samples: each is the product of 100 random >>> # values, drawn from a normal distribution. >>> b = [] >>> for i in range(1000): ... a = 10. + np.random.standard_normal(100) ... b.append(np.prod(a)) >>> b = np.array(b) / np.min(b) # scale values to be positive >>> count, bins, ignored = plt.hist(b, 100, density=True, align='mid') >>> sigma = np.std(np.log(b)) >>> mu = np.mean(np.log(b)) >>> x = np.linspace(min(bins), max(bins), 10000) >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2)) ... / (x * sigma * np.sqrt(2 * np.pi))) >>> plt.plot(x, pdf, color='r', linewidth=2) >>> plt.show() lognormal seed(seed=None) Reseed a legacy MT19937 BitGenerator Notes ----- This is a convenience, legacy function. The best practice is to **not** reseed a BitGenerator, rather to recreate a new one. This method is here for legacy reasons. This example demonstrates best practice. >>> from numpy.random import MT19937 >>> from numpy.random import RandomState, SeedSequence >>> rs = RandomState(MT19937(SeedSequence(123456789))) # Later, you want to restart the stream >>> rs = RandomState(MT19937(SeedSequence(987654321))) seed rayleigh(scale=1.0, size=None) Draw samples from a Rayleigh distribution. The :math:`\chi` and Weibull distributions are generalizations of the Rayleigh. .. note:: New code should use the `~numpy.random.Generator.rayleigh` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- scale : float or array_like of floats, optional Scale, also equals the mode. Must be non-negative. Default is 1. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``scale`` is a scalar. Otherwise, ``np.array(scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Rayleigh distribution. See Also -------- random.Generator.rayleigh: which should be used for new code. Notes ----- The probability density function for the Rayleigh distribution is .. math:: P(x;scale) = \frac{x}{scale^2}e^{\frac{-x^2}{2 \cdotp scale^2}} The Rayleigh distribution would arise, for example, if the East and North components of the wind velocity had identical zero-mean Gaussian distributions. Then the wind speed would have a Rayleigh distribution. References ---------- .. [1] Brighton Webs Ltd., "Rayleigh Distribution," https://web.archive.org/web/20090514091424/http://brighton-webs.co.uk:80/distributions/rayleigh.asp .. [2] Wikipedia, "Rayleigh distribution" https://en.wikipedia.org/wiki/Rayleigh_distribution Examples -------- Draw values from the distribution and plot the histogram >>> from matplotlib.pyplot import hist >>> values = hist(np.random.rayleigh(3, 100000), bins=200, density=True) Wave heights tend to follow a Rayleigh distribution. If the mean wave height is 1 meter, what fraction of waves are likely to be larger than 3 meters? >>> meanvalue = 1 >>> modevalue = np.sqrt(2 / np.pi) * meanvalue >>> s = np.random.rayleigh(modevalue, 1000000) The percentage of waves larger than 3 meters is: >>> 100.*sum(s>3)/1000000. 0.087300000000000003 # random rayleigh get_state(legacy=True) Return a tuple representing the internal state of the generator. For more details, see `set_state`. Parameters ---------- legacy : bool, optional Flag indicating to return a legacy tuple state when the BitGenerator is MT19937, instead of a dict. Raises ValueError if the underlying bit generator is not an instance of MT19937. Returns ------- out : {tuple(str, ndarray of 624 uints, int, int, float), dict} If legacy is True, the returned tuple has the following items: 1. the string 'MT19937'. 2. a 1-D array of 624 unsigned integer keys. 3. an integer ``pos``. 4. an integer ``has_gauss``. 5. a float ``cached_gaussian``. If `legacy` is False, or the BitGenerator is not MT19937, then state is returned as a dictionary. See Also -------- set_state Notes ----- `set_state` and `get_state` are not needed to work with any of the random distributions in NumPy. If the internal state is manually altered, the user should know exactly what he/she is doing. get_statemtrand wald(mean, scale, size=None) Draw samples from a Wald, or inverse Gaussian, distribution. As the scale approaches infinity, the distribution becomes more like a Gaussian. Some references claim that the Wald is an inverse Gaussian with mean equal to 1, but this is by no means universal. The inverse Gaussian distribution was first studied in relationship to Brownian motion. In 1956 M.C.K. Tweedie used the name inverse Gaussian because there is an inverse relationship between the time to cover a unit distance and distance covered in unit time. .. note:: New code should use the `~numpy.random.Generator.wald` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- mean : float or array_like of floats Distribution mean, must be > 0. scale : float or array_like of floats Scale parameter, must be > 0. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``mean`` and ``scale`` are both scalars. Otherwise, ``np.broadcast(mean, scale).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized Wald distribution. See Also -------- random.Generator.wald: which should be used for new code. Notes ----- The probability density function for the Wald distribution is .. math:: P(x;mean,scale) = \sqrt{\frac{scale}{2\pi x^3}}e^ \frac{-scale(x-mean)^2}{2\cdotp mean^2x} As noted above the inverse Gaussian distribution first arise from attempts to model Brownian motion. It is also a competitor to the Weibull for use in reliability modeling and modeling stock returns and interest rate processes. References ---------- .. [1] Brighton Webs Ltd., Wald Distribution, https://web.archive.org/web/20090423014010/http://www.brighton-webs.co.uk:80/distributions/wald.asp .. [2] Chhikara, Raj S., and Folks, J. Leroy, "The Inverse Gaussian Distribution: Theory : Methodology, and Applications", CRC Press, 1988. .. [3] Wikipedia, "Inverse Gaussian distribution" https://en.wikipedia.org/wiki/Inverse_Gaussian_distribution Examples -------- Draw values from the distribution and plot the histogram: >>> import matplotlib.pyplot as plt >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, density=True) >>> plt.show() wald triangular(left, mode, right, size=None) Draw samples from the triangular distribution over the interval ``[left, right]``. The triangular distribution is a continuous probability distribution with lower limit left, peak at mode, and upper limit right. Unlike the other distributions, these parameters directly define the shape of the pdf. .. note:: New code should use the `~numpy.random.Generator.triangular` method of a `~numpy.random.Generator` instance instead; please see the :ref:`random-quick-start`. Parameters ---------- left : float or array_like of floats Lower limit. mode : float or array_like of floats The value where the peak of the distribution occurs. The value must fulfill the condition ``left <= mode <= right``. right : float or array_like of floats Upper limit, must be larger than `left`. size : int or tuple of ints, optional Output shape. If the given shape is, e.g., ``(m, n, k)``, then ``m * n * k`` samples are drawn. If size is ``None`` (default), a single value is returned if ``left``, ``mode``, and ``right`` are all scalars. Otherwise, ``np.broadcast(left, mode, right).size`` samples are drawn. Returns ------- out : ndarray or scalar Drawn samples from the parameterized triangular distribution. See Also -------- random.Generator.triangular: which should be used for new code. Notes ----- The probability density function for the triangular distribution is .. math:: P(x;l, m, r) = \begin{cases} \frac{2(x-l)}{(r-l)(m-l)}& \text{for $l \leq x \leq m$},\\ \frac{2(r-x)}{(r-l)(r-m)}& \text{for $m \leq x \leq r$},\\ 0& \text{otherwise}. \end{cases} The triangular distribution is often used in ill-defined problems where the underlying distribution is not known, but some knowledge of the limits and mode exists. Often it is used in simulations. References ---------- .. [1] Wikipedia, "Triangular distribution" https://en.wikipedia.org/wiki/Triangular_distribution Examples -------- Draw values from the distribution and plot the histogram: >>> import matplotlib.pyplot as plt >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=200, ... density=True) >>> plt.show() triangular set_state(state) Set the internal state of the generator from a tuple. For use if one has reason to manually (re-)set the internal state of the bit generator used by the RandomState instance. By default, RandomState uses the "Mersenne Twister"[1]_ pseudo-random number generating algorithm. Parameters ---------- state : {tuple(str, ndarray of 624 uints, int, int, float), dict} The `state` tuple has the following items: 1. the string 'MT19937', specifying the Mersenne Twister algorithm. 2. a 1-D array of 624 unsigned integers ``keys``. 3. an integer ``pos``. 4. an integer ``has_gauss``. 5. a float ``cached_gaussian``. If state is a dictionary, it is directly set using the BitGenerators `state` property. Returns ------- out : None Returns 'None' on success. See Also -------- get_state Notes ----- `set_state` and `get_state` are not needed to work with any of the random distributions in NumPy. If the internal state is manually altered, the user should know exactly what he/she is doing. For backwards compatibility, the form (str, array of 624 uints, int) is also accepted although it is missing some information about the cached Gaussian value: ``state = ('MT19937', keys, pos)``. References ---------- .. [1] M. Matsumoto and T. Nishimura, "Mersenne Twister: A 623-dimensionally equidistributed uniform pseudorandom number generator," *ACM Trans. on Modeling and Computer Simulation*, Vol. 8, No. 1, pp. 3-30, Jan. 1998. set_statenumpy._core._multiarray_umathnumpy.core._multiarray_umath_ARRAY_API_ARRAY_API is not PyCapsule object_ARRAY_API is NULL pointermodule compiled against ABI version 0x%x but this version of numpy is 0x%xmodule was compiled against NumPy C-API version 0x%x (NumPy 1.23) but the running NumPy has C-API version 0x%x. Check the section C-API incompatibility at the Troubleshooting ImportError section at https://numpy.org/devdocs/user/troubleshooting-importerror.html#c-api-incompatibility for indications on how to solve this problem.FATAL: module compiled as unknown endianFATAL: module compiled as little endian, but detected different endianness at runtimenumpy.PyArray_MultiIterNew2numpy.PyArray_MultiIterNew3numpy.import_arraynumpy.random.mtrand.int64_to_long__init__numpy.random.mtrand.RandomState.__init__numpy.random.mtrand.RandomState.__repr__numpy.random.mtrand.RandomState.__str__numpy.random.mtrand.RandomState.__getstate__numpy.random.mtrand.RandomState.__setstate__numpy.random.mtrand.RandomState.__reduce__BitGeneratornumpy.random.mtrand.RandomState._initialize_bit_generatornumpy.random.mtrand.RandomState.seednumpy.random.mtrand.RandomState.get_statenumpy.random.mtrand.RandomState.set_statenumpy.random.mtrand.RandomState.random_samplenumpy.random.mtrand.RandomState.randomnumpy.random.mtrand.RandomState.betanumpy.random.mtrand.RandomState.exponentialnumpy.random.mtrand.RandomState.standard_exponentialnumpy.random.mtrand.RandomState.tomaxintnumpy.random.mtrand.RandomState.randintnumpy.random.mtrand.RandomState.bytesnumpy.random.mtrand.RandomState.choicenumpy.random.mtrand.RandomState.uniformnumpy.random.mtrand.RandomState.randnumpy.random.mtrand.RandomState.randnnumpy.random.mtrand.RandomState.random_integersnumpy.random.mtrand.RandomState.standard_normalnumpy.random.mtrand.RandomState.normalnumpy.random.mtrand.RandomState.standard_gammanumpy.random.mtrand.RandomState.gammanumpy.random.mtrand.RandomState.fnumpy.random.mtrand.RandomState.noncentral_fnumpy.random.mtrand.RandomState.chisquarenumpy.random.mtrand.RandomState.noncentral_chisquarenumpy.random.mtrand.RandomState.standard_cauchynumpy.random.mtrand.RandomState.standard_tnumpy.random.mtrand.RandomState.vonmisesnumpy.random.mtrand.RandomState.paretonumpy.random.mtrand.RandomState.weibullnumpy.random.mtrand.RandomState.powernumpy.random.mtrand.RandomState.laplacenumpy.random.mtrand.RandomState.gumbelnumpy.random.mtrand.RandomState.logisticnumpy.random.mtrand.RandomState.lognormalnumpy.random.mtrand.RandomState.rayleighnumpy.random.mtrand.RandomState.waldnumpy.random.mtrand.RandomState.triangularnumpy.random.mtrand.RandomState.binomialnumpy.random.mtrand.RandomState.negative_binomialnumpy.random.mtrand.RandomState.poissonnumpy.random.mtrand.RandomState.zipfnumpy.random.mtrand.RandomState.geometricnumpy.random.mtrand.RandomState.hypergeometricnumpy.random.mtrand.RandomState.logseriesnumpy.random.mtrand.RandomState.multivariate_normalnumpy.random.mtrand.RandomState.multinomialnumpy.random.mtrand.RandomState.dirichletnumpy.random.mtrand.RandomState.shufflenumpy.random.mtrand.RandomState.permutationnumpy.random.mtrand.seednumpy.random.mtrand.get_bit_generatornumpy.random.mtrand.set_bit_generatornumpy.random.mtrand.samplenumpy.random.mtrand.ranfbuiltinstypeboolcomplexnumpydtypeflatiterbroadcastndarraygenericnumberintegersignedintegerunsignedintegerinexactfloatingcomplexfloatingflexiblecharacterufuncnumpy.random.bit_generatorSeedSequenceSeedlessSequencenumpy.random._commondoublePOISSON_LAM_MAXLEGACY_POISSON_LAM_MAXuint64_tMAXSIZEnumpy.random._bounded_integersPyObject *(PyObject *, PyObject *, PyObject *, int, int, bitgen_t *, PyObject *)_rand_uint64_rand_uint32_rand_uint16_rand_uint8_rand_bool_rand_int64_rand_int32_rand_int16_rand_int8int (double, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type)check_constraintint (PyArrayObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type)check_array_constraintdouble (double *, npy_intp)kahan_sumPyObject *(void *, bitgen_t *, PyObject *, PyObject *, PyObject *)double_fillPyObject *(PyObject *, PyArrayObject *)validate_output_shapePyObject *(void *, void *, PyObject *, PyObject *, int, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyObject *)contPyObject *(void *, void *, PyObject *, PyObject *, int, int, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type)discPyObject *(void *, void *, PyObject *, PyObject *, PyArrayObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyArrayObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type, PyArrayObject *, PyObject *, __pyx_t_5numpy_6random_7_common_constraint_type)cont_broadcast_3discrete_broadcast_iiiInterpreter change detected - this module can only be loaded into one interpreter per process.name__loader__loader__file__origin__package__parent__path__submodule_search_locationsModule 'mtrand' has already been imported. Re-initialisation is not supported.cython_runtime__builtins__numpy.random.mtrandinit numpy.random.mtrand4294967296name '%U' is not defined while calling a Python objectNULL result without error in PyObject_Callraise: arg 3 must be a traceback or Noneinstance exception may not have a separate valuecalling %R should have returned an instance of BaseException, not %Rraise: exception class must be a subclass of BaseExceptionexception causes must derive from BaseExceptionkeywords must be strings%s() got multiple values for keyword argument '%U'%.200s() keywords must be strings%s() got an unexpected keyword argument '%U'at leastat mostexactlys%.200s() takes %.8s %zd positional argument%.1s (%zd given)hasattr(): attribute name must be stringjoin() result is too long for a Python stringcannot import name %S'%.200s' object is unsliceableMissing type objectCannot convert %.200s to %.200stoo many values to unpack (expected %zd)need more than %zd value%.1s to unpackassignment'%.200s' object does not support slice %.10scannot fit '%.200s' into an index-sized integer'%.200s' object is not subscriptablebase class '%.200s' is not a heap typeextension type '%.200s' has no __dict__ slot, but base type '%.200s' has: either add 'cdef dict __dict__' to the extension type or add '__slots__ = [...]' to the base typeinvalid vtable found for imported typemultiple bases have vtable conflict: '%.200s' and '%.200s'%.200s.%.200s is not a type object%.200s.%.200s size changed, may indicate binary incompatibility. Expected %zd from C header, got %zd from PyObject%.200s.%.200s size changed, may indicate binary incompatibility. Expected %zd from C header, got %zd-%zd from PyObject.No module named '%U'_cython_3_1_2Shared Cython type %.200s is not a type objectShared Cython type %.200s has the wrong size, try recompiling__name__ must be set to a string object__qualname__ must be set to a string objectfunction's dictionary may not be deletedsetting function's dictionary to a non-dict__defaults__ must be set to a tuple objectchanges to cyfunction.__defaults__ will not currently affect the values used in function calls__kwdefaults__ must be set to a dict objectchanges to cyfunction.__kwdefaults__ will not currently affect the values used in function calls__annotations__ must be set to a dict object%.200s() %s (%zd given)%.200s() %sBad call flags for CyFunctiontakes no argumentstakes exactly one argumenttakes no keyword argumentsunbound method %.200S() needs an argumentneeds an argumentdoes not matchcompile time Python version %d.%d of module '%.100s' %s runtime version %d.%d__pyx_capi__%.200s does not export expected C variable %.200sC variable %.200s.%.200s has wrong signature (expected %.500s, got %.500s)%.200s does not export expected C function %.200sC function %.200s.%.200s has wrong signature (expected %.500s, got %.500s)__int__ returned non-int (type %.200s). The ability to return an instance of a strict subclass of int is deprecated, and may be removed in a future version of Python.__int__ returned non-int (type %.200s)an integer is required:0yE>??qh?j%= BT ~Q~U~KD Ga7\%aFOaSuzpD(|Wc %WM$ t`K[oT`gtSwf# Wl`0H7[z1z(zK^2#9MM0MFPrOxS왎2ȩn{TH,ҭ^p .]M[\}r;/4d6dcNQp.t@e$oX%L(xo6xu{fUY>9>{ppCBwS(:5^dܓAN}8) YfHqն&|s f2,2Ztզޗ .n ZR'ӯB)[l@u Pҍ'TȈt(5wI'L/$;nXMØT`OArW,+jtȳRfARnqӊ<KZW$eKs) 4<=>)G'QA@Y.(5bX jz>lq{2Xx{~JH҄Cc`Qz%~ )Q\HsrUb'Bkq-hnק Ψ;3Kd)P^٨Tv$Hx"$ 5..&$ŗ: L< Ŀk<4xV<=A[<'?}y<NG<~;[?l?*?%z?PՋt?4?e;?$"?zaWF}?Gz‘B?Oq1? OU?ߺH?7a?nV,? K?Xhw?հ<?Vp\?m?)?zP?ZcX?*;Q^?#*'g? U7?e&$ ?jJo?\Ȭ)?L&?FS?leZ&?g ?NIO??xRr!?P_hy?y6IJO?_5%?[X~?1>?bU?+À?PX?5:pɗ0?8d?;U?J?͓?)m?ېZ]G?/|!? ?iT??Wq?PF9 ?ߓ^??ۮY?3???i?Z8o? O5?ٸ?P?R9?igP?La;?L?!ވ?%o?{7=8?Ҁt?DvC?6?=p\?;So&?mj?W?j?$O?z5Ѽ?Ҏ?C|P?yh|?%H?/ZM?f!w;??>ǭ?MAz?G?y?.?P9կ?TT}?g4K?#$O? Y?BM?6C;?B"_U?~t$?œ߉?52?Ҙl'?DɤT?<(i?qE8 ? Uī?OQM?o^?Sq͒?Gط5?zx?1zd}?:R!?Wg?~& ~k?=~-2?ZҿҶ?'|j_]?it?[?8R?uqb?#h?z|J?G~`?\!>?GF?vJ?l󈬚?5hȩmE?㭍?-l ?uG?1i%?調?M?e*|?zè?^V?4<%F?B}u?c-@c?n? R=?Kr?*}T#?,"k>?R) ?K{o?vaӽ?命8? t;I_? h?3xk?3Ӻ?b3?vZ9S?LJisk?M$a.?ftW?+ ?"@|?&#?p>_?1fҲ? DE?} ?/?%,?0?5nl+,&?QG?b. ?,*(>?p_8?cU)?h*?'wާ?dИۦ?ԭ<ڥ?]']ۤ?ݣ?=|?j?.?ĥׁ?u? ̓0?"NR? y? ڥ?d֔?^8 ?0`4I?IrO*?O'?x A?B?/)?7h`|?] ٨v?p?gC_e?T?wE`mru\zw8xky5zz/ {ԃ{{7|3}|&|H|}C}g}ۇ}}a}g}]}~~4%~5~C~Q~g^~ij~u~>~2~~r~դ~Ƭ~N~u~C~~~~k~~~~~~t~~~6 < :#%](*.-z/13579;=?EABD:FGNIJ8LMNLPQR T=UdVWXYZ[\]^~__`;abbcod.eefLggh~~7~~/~7~~ ~ ~w~G]~>~Y~,}6}b}|O|06{xI4_h2z3+3@3aQ3i`3{am3Ay3i3*353=3r333|ϡ3ڍ3+333^33׶3iż3-¿3c3%3uY3<3L3gv3;3k3-3$3!333P3P33<3p~3չ3^3J3I34V?4=3@4A4A4qB4C4D4udE4-CF4K"G4H4H41I4J4vK4\fL4HM4+N4aO4O4bP4ٽQ4R4ԊS4crT4ZU4CV4-W4ZX4Y4UY4Z4[4(\4_]4^4_4C`4va4alb40cc47[d4~Te4Of4Jg42Hh4Fi4Fj4Hk4Kl4MPm4Vn4^o48hp4sq4r4s4 t4u4v4Cw4x4 z42{40S|4u}4~44v4@ 4L4>4ق4v444lV44R4F44p4 I44"4_44Ќ4l4L4`4ԏ4坐4y4ݖ4%44r&4k44(4444.4Q4N4t44\۶4H94̻4p44~X4w4p_4~444?V#z?u?q?}n?k?Lh?e?Rc?`?Zw^?*+\?Y?RW?U?_S?XQ?߱O?M?3K?J?GH?F?jD?`C?(`A?j??>?x,>N>>q>>>j>>k>>Π>>F>>>'>\>#>u>J>*>_F>d>+>$>w>>>JK>y>|>iݿ>>I>;>ʾ>t>5<> ~>>>O>>>~3>T>ե>(>g~>ՠ>G/>>>F>J> >:n>bԓ>Q<>>x>~>>>^>Ј>D>l>1>>%>\D>@|>?y>Bv>Hs>Qp>#^m>mj>|g>md>a>^>$[> Y>=3V>[S>P>M>J>~H>UE>B>?>=>S:>7>"4>=22>T/>d,>m+*>m'>c$>N?">,>>m>t>F>>1*> > >Y>>ʗ>>I=_={==^==&=_=g=='0===P6=˙=\= s==d= =yo=/=6=.=fЍ=x=i'=܀=a1y=p=xIh=_==W=TO=G=>=N6=.=&===-H==<א<̀<<<. NVeΙVn6nvK zicp%E tQ)2U1WQ9Lin?23F:L"3\LQ V f[_rWDdx h+*k2=Ko:qr Mu\x?A{FS~8;b=ZV`bBtu9=JE>XدGwdO 8cx AFẙi&zqVYםΡag6 X83:뇡koɣj_ۤ| Mg^ݧt|Ψ_ΓXp2X^ttH蟿W;ޭl~$\z[߁İPp:J+N!X ɦ֬ ᆴX7(. Ɉ?5}h.G{tr&oya=cA/˺DH0⤮<)9O@ᣩTrVj֋@?˷dsI^i@(0߾ta&⊂l1EA1T[n&mi#d)B}QJwt}B < EOvpc/F<Ң"Ae އ0~ Rfq(*QtH3D@M`P}hwx%ƿ8*JG+[EliPIw+ E>ҙ02yΩ4A (Nt.Ȱ--̕^&܌z#;ޖu~g6X .pmF 3n bH޵LaEZvpR(-x_b˿ӰdyQӶVg<7܆ut7$MH𯋉ld"rqտH)݄ /0 wپ}2}K D5z&R cM,}uc?Ѡp5.bJ3ʸT[vv+\[U@ضBi"7oLeiFγ>SR(D2Z> B0$y1gWr-ެ @樫(afoeW-|&aY +M]d<A]X`<+M[Ij<[5q&<.8eG< h#ឪV <;LC%K<ꆭh?{s?%;?omo?3;?J9?++?*T[?};1s?HeC?$`?vE!=?ſ-r?MBц?K=?Q}6Ei?7u? !?z}k? ~?@?`x?*?8? Qi?oTC?_(4?ָ?@je?!u v?7Zi?{ ?I?]T?9]??}?8aD:?Yζi?Ɲҷ?r^sSw?ꍰ07?d>[?%۹? Ə{?'HB>?vX#?l1&?:l?磽!O?ލ?&?ڋ?ﾀ+ j?A1?N0Z?0H?}G?(V?5$1!?pB9 ?b"FS?)vEW(?vG}rO?~ /? {^?Z? ?ބS?i"?lR?3Sn?>N?Ґ]b?,|y2?jG>?TLҫ?~>\O??@YH?/֎@?9O"H?>?1 7?8?Ox?]4?5D9g?r|?>ܸ$8? [B/?I䠟?O?y%d?bPޱ?c?PR?j?F}?9(Q1?c?(ڦ^w?0U^Q?1j?T ξ?x.BTv?Imb.??6YJ?)ِ?\C}?%d?w?SN?эv?pa?,Q&?@oű?SuFe?PV?;?I?viׯ?4D?.g?X1Iα?Jy?!dJ?پz?j»?8G;?L|{ʎ?mwn?k9:9??Ry?A&E?U?Ŗ<?k&_?G??~#? V#?_?S?Q| z? Y&?$?htQz? 3Tݜ?pXP?N梚?H*g?gS(u?1c?w@rT?Q=I?QA?]1%? RD?lj?W'n?-BU؊?h?t4? n?boQ?qvi?_)N?]tQW}?6H#z? 67w?"Ηs?C@Wi=q?ḰXl?f?$ka?%> T+Y? O?K 2=?QHqoMֻanjDotTrotou$w'xx,jyy7\z׻z{W{S{{.|3|]|ȃ|||I||}C0}F}Z}m}}S}(}}-}}"}}|}M}~i ~~~B(~o0~C8~?~F~M~T~Z~a~f~l~r~]w~v|~`~ ~~$~m~~~w~:~ަ~f~ѭ~#~Z~y~~q~K~~~^~~a~~~`~~~~~~~~f~*~~~-~~J~~=~~~\~~~$~U~}~~~~~pH`  i   6  H  A!B+m 5XttW3 `wK\ L   s   G {V~~~d~~x~~K~~~~~~~~~)~~~a~~~{~;~~A~~~m~~z~~~"~k~]~~~ԃ~|~s~j~Ua~W~K~?~2~$~~~ }}} }}i}A}}|Q|D|{3N{zeyww7mse'52V227222f23ن 33H3(&3o.3z63oN>3E3lM3FT3/[3b3i34p3fw3&~3[3B3ψ3g37!3>3T3d3n3r3Fq3j3_31P3r<3$3k 33ȸ3q3|{3P3#3C333dY3"3+3®3r35333x37333p3-33731b33l44(44h4C44 4` 4M 47 44?4nB44L4 i4a4T044542`44p. 4!4"4i$4%4@'4t(4>*4+4,4j.4/4'P1424):4454&)7484c:4;4$=4+>4@4A4KC4vD4B(F4G4:I4J4rTL4M4GuO4Q4R44T4U4EiW4Y4 Z4G\4]4_4:a4b4d4Bf4\g4ji4bk4m4n4p4\r4}"t4Yu4Hw4[y4X{46.}4 4q4a4]S4F4N<434,4+(4{%4$4o&4,*4'04m84 C4P4_4q474{4w4>ԕ44s4<4d444$4 (4a44lߢ4$4l44x 4_444{4 4EP4±4{:4귴4);4nķ4S444<*4տ44A4.44ע44f4RW4R4*Y4Fk44δ444444g44k4<444y44u4_45555@5ó 5 5]5^555q5v 5!5%5V*5s/5;S55:<5D5NO5^5Nv5?/*p?3f?(_?xY?յS?N?J?F?DB?Qt??u+ ?A?Į?"?ʝ?G??i>l>7>>*J>>>^>>F>>7P>>K{>>>u>;->>>|>eO>4(>8>L>N>ȿ>>>>·>ε>߳>>>0>U>~>|>ަ>Y>IP>w>ҟ>B>e>)>~>_>>C>{>>J>﷍>X(>'>N>Í>x >b>x>>!>^}>;z>Хw>@t>wr>byl>i>g>Sd>3a>^>]\>&Y>z)W>T>P R>O>L>5~J>3H>E>nC>@>VK>>;>9>07>4>2>>0>p->+>s)>7'>%>"> >s>L>*>= >T>T>4>>y >ϣ > >>L>>l>=+==0 =C==8==hp==0== =n==|===;=ں=Z=o,=ް=ߗ=.W===%==r=W= C=4=M,=4*=D.=y8=H=~=x=h㈵>??9v?[ m?m{?h|?5?Q?x&?UUUUUU?UUUUUU?Q???dg?r鷯?n?$~?r?@(\@= ףp=@@-DT! @333333 @B>٬ @3 ; @@ffffff@-DT!@@~)@"@$@.@4@>@X@`@a@i@|@@@N@MAA.AACS lѿ-DT! -DT!q?3?r?0@  @ & Uh 0l& l & 4L&      |   F.text$mn6.text$mn$00F.text$x.idata$58.00cfg.CRT$XCA.CRT$XCZ .CRT$XIA(.CRT$XIZ0.CRT$XPA8.CRT$XPZ@.CRT$XTAH.CRT$XTZP.rdata& l.rdata$voltmdl& 4.rdata$zzzdbg( .rtc$IAA( .rtc$IZZ( .rtc$TAA( .rtc$TZZ( (.xdataP .edataX d.idata$2,Y .idata$3@Y .idata$4` .idata$6 .data `%.bss (.pdata  t20  4 2p  R`P!) ) t 4 Pm( !Pm( 11d B!TP) !4$) !t8) !8) !$) !P)   4 2p td4 rPdT42pd42 ptT424 2 p`P! ) !) ! ) d T42p 4rp`PT 4 R p `2pP0!d Px* ! * !* !Px* d T 4Rp t dT42 d T 4 2pB2`!4 + !t ) + ! ) + ! + pP!  d4 ph+ !t+ !t+ !t+ !d4 ph+ d T 4 Rp//t20Q QLdGT  p0dT 4 p d T 4 Rp Bp ` P 0B4 2 p`P! Rd, ! Rox, !Rox, !Rd, R020  4 Rp t d T 4R  4 2p!T, !T, !dak- !ak- !, 4< <4,t%T ` 0!4+ !t4+ !+ !4 4+ ! 4+ !t4 4+ !d4t+ !+  dT4 r!t 0(. !0(. !t p(. !p(. T 4 R p ` 4 ` [  p`Pd4 p P  p`!) ) %  T4. !. ooT d4 pF Fd >4 T 2 p+ dT4p PT4 r p `  P!a at *" d 4 `y|/ !`y|/  P!;;h4&t d4p/ !p/ !p/ * 4" p`P `P!4@0 !   tL0 !L0 !L0 !@0 ^ ^VKT rp`0$pP !d40 !Č0 !&&ԋ1 !xBhC1 !1 !1 !0 !d0 !0 !.0P !d-881 !t,881 !P P9x) *+881 !881 !881 !881  pP !( ($ d4#`FF42 !`FF42 !`FF42 %>`0P !ffKx.h 9:; <t=`RR2 !:`RR2 !`RR2 +0P !WWh$%&'( t)d*ff3 !ff3 ' + p`0P H&@pP !4??p3 !**;<=d>?C3 !;<=?C3 !;=?C3 !;?C3 !?C3 !?p3 ! ;<=d>4??p3 ' ' p`0P (,p`P !43UUh4 !() *+UU4 !xhU e4 !U e4 !xhU e4 !)UU4 !)UUh4 !UUh4 * 40& p`P ((P !#4,F"$%t&d'P5 !4,P5 !"#$%t&d'4,P5  4r p`P d T 42pp ` 0 P! 6 !6 `0 `!t086 !086  p0P!dp6 !p6  p0P!  di6 !i6 !  d 6 ! 6 !  d`ԕ6 !`ԕ6  p0P!d |<7 !|<7 &  p0P !dCt7 !Ct7 %  p`0P !7 !7 %  p`0P !$o%7 !$o%7   p0P!d **@8 ! **@8 !  d/y/6 !/y/6 $ p0P !d@3(48 !@3(48 !d UZV<7 !UZV<7 !d Z[8 ! Z[8 $ p0P !d vv49 ! vv49 p`0 P!Tx9 !x9 !T҆x9 !҆x9 $ p0P !dH9 !H9 !dq@8 !q@8 %  p`0P !͘<: !͘<: %  p`0P !Y: !Y: !`,7 !`,7 !Y7 !Y7 % " p`0P !!`I ; !`I ; !: !: !i7 !i7 !d1@8 !1@8 !I: !I: !P7 !P7 !): !): ! : ! : !P: !P: !p=<: !p=<: !0<: !0<: !p = <: !p = <: !0<: !0<: !07 !07 !07 !07 !d8 !8 !d9 !9 '  p`0P '  p`0P '  p`0P !dPkQ8 !PkQ8 $ p0P !def > !ef > %  p`0P !kld> !kld> $ p0P !dqq> !qq> !  d}~6 !}~6 !  d6 !6 !  dp6 !p6 !  dp6 !p6 d42 ptd424 d 2pT4 r p `yyd"p0PR`!tT 4#<$? !tT 4#<$? "Vp0P !_^d]p&&0@ !p&&0@  p`P0!=x@ !=x@  t d T 42T `!  t4p@ !p@ p ` P 0!A !A 44t /T *4 R`zz4 d T R pdT42pd T 4Rp!hpA !pA xd 4 r p!h`A !`A !h`A !`A 4pP!; ;h)x d`xB !`xB 4pP!> >h1x d`B !`B !hA !A dT 4 p!++xhpB !pB WWh@(xdT4pPPhC5$xdT4p` P!RRD6/x$h t4\C !\C Tp!UUG9+$xh d4 C ! C && 0!0hD !x h hD !  O,D !O,D !hD !0hD $$ 0!D !% % x h D !  D !D !D !D x h  hbxh# #h 0!!!x 5TE !5VlE !5VlE !5TE !x 5TE  x0  hR0  hb  xbxhr0##xh0x h0hx0!!x 0!h @hF ! |F !|F !|F !@hF 5 5+"xhIIxDh>4T 2 `!t 0G !0G 884T 2 `!t 6 H !6 H !4и+ !4и+ 4 2 p`P! H ! H !H !H 4 2 pP!d `H ! YH !YH !`H 4 2 p!d P8I !T LI !LI !P8I   2 P0!t (I !d I !  I ! I !t (I !(I td2!4J !J ==6.&x hdT4ptt l d S K C>0(x!h& p ` 0!T-pxJ 990(x"hdT4p!mJ !mJ !mJ !mJ UU4! ! xh4K !x4K KKE;5/*% x h p`0 d T 42p0!  xhK !xK   xb  x!h@Z L !@Z L x hhb\\h x06 60&xh 0!YL !YL , ,#h 0!xPL !""L !L !PL !xPL && 0!pPM !x h dM !  xM !xM !dM !pPM ## 0!h  M !/ /)#x M ! M ! M   4 `P!x h@qXN !tqhN !qhN !@qXN @@.) xh 0NNHC>4+xh 033hxysLC91 )x !h 4#p ` P0t d42~XF*5F2P  4 Rp~o]dv]v 4rp`~RP  brp`0  4 RP "~  P4PdT4p^S ??P Q R pP`ги00`  @@ @`0```p0P p@pyS S S S S S S T #T =T ZT yT T T T T T T T U (U 6U FU UU eU vU U U U U U U U V "V 8V NV [V kV V V V V W %W LW bW zW W W W W W X (X EX dX vX X X X X X   !"#$%&'()*+,-./0123456789:;<=>mtrand.cp311-win_amd64.pydPyInit_mtrandrandom_betarandom_binomialrandom_bounded_bool_fillrandom_bounded_uint16_fillrandom_bounded_uint32_fillrandom_bounded_uint64random_bounded_uint64_fillrandom_bounded_uint8_fillrandom_buffered_bounded_boolrandom_buffered_bounded_uint16random_buffered_bounded_uint32random_buffered_bounded_uint8random_chisquarerandom_exponentialrandom_frandom_gammarandom_gamma_frandom_geometricrandom_geometric_searchrandom_gumbelrandom_intervalrandom_laplacerandom_logisticrandom_lognormalrandom_logseriesrandom_multinomialrandom_negative_binomialrandom_noncentral_chisquarerandom_noncentral_frandom_normalrandom_paretorandom_poissonrandom_positive_intrandom_positive_int32random_positive_int64random_powerrandom_rayleighrandom_standard_cauchyrandom_standard_exponentialrandom_standard_exponential_frandom_standard_exponential_fillrandom_standard_exponential_fill_frandom_standard_exponential_inv_fillrandom_standard_exponential_inv_fill_frandom_standard_gammarandom_standard_gamma_frandom_standard_normalrandom_standard_normal_frandom_standard_normal_fillrandom_standard_normal_fill_frandom_standard_trandom_standard_uniformrandom_standard_uniform_frandom_standard_uniform_fillrandom_standard_uniform_fill_frandom_triangularrandom_uintrandom_uniformrandom_vonmisesrandom_waldrandom_weibullrandom_zipfZ m Y 4n Z fo Z o H@Y q p p p p p p hp Lp 8p $p p o o o o m m m m m m n Fn Nn Vn n n n n n n n n |n vn pn hn `n n n n o (o Do \o n e e "e 2e De Ze je ~e e e e e e e f f *f 8f Hf Zf hf f f f f f f f g g 4g Jg ^g rg g g g g g g g h h ,h Nh fh ~h h h h h h i ~a d Ri pi i i i i i i i j j .j >j Tj rj j j j j j j k k PyTuple_TypePyList_TypePyDict_Type1PyCFunction_TypePyMethod_TypeBPyCapsule_Type4PyTraceBack_Type_Py_EllipsisObject%Py_Versionpython311.dll@strchrAstrrchr;memcmp<memcpy>memset__C_specific_handler%__std_type_info_destroy_listVCRUNTIME140.dllexpm1log1plog1pfJacos|ceilcosexpexpffloorfmodloglogfpowpowfsqrtsqrtf6_initterm7_initterm_e?_seh_filter_dll_configure_narrow_argv3_initialize_narrow_environment4_initialize_onexit_table"_execute_onexit_table_cexitapi-ms-win-crt-math-l1-1-0.dllapi-ms-win-crt-runtime-l1-1-0.dllRtlCaptureContextRtlLookupFunctionEntryRtlVirtualUnwindUnhandledExceptionFilterSetUnhandledExceptionFilter@GetCurrentProcessTerminateProcessIsProcessorFeaturePresentQueryPerformanceCounterAGetCurrentProcessIdEGetCurrentThreadIdGetSystemTimeAsFileTimeADisableThreadLibraryCallsInitializeSListHeadIsDebuggerPresentKERNEL32.dllP .4.(?$4xC *@?I/CV@3I`e9Vo ?e{DoJ{P PUl Zx_e kq0 v {I Ђ }Dx#P#0{{#0{{#{|#{|#p||#| }#| }##$$($8$H$pX$ph$px$@$0 /@/0/(/$gp|@;2N>4BБ2ЖG3@` Hp Ip 6ЁPM d Y`80` px P   hp x  0 .  X/.P=P`/@K0PKg @0` P 4.(?$4xC *@?I/CV@3IPUl Z vDx#P/$@;2N>d Y`` x Px .  P=P`/8Kp`=^0`KHtp g00`t0l0.`e9Vo ?e{DoJ{P x_e kq0}808Kp`=@K0PK^0`K^^`D^p  `g Htp g00`t@0#p&x  l00` UUUUUU?llfJ?88C$+K?<ٰj_AAz?SˆB8?5gG2-+] fu|<7 |!P7 !d7 Ct7 C7 7 9 Hp= H= 7 e"7 e"$7 $o%7 o%(8 ( *08 **@8 *D-X8 D-.l8 /y/6 y/$2|8 $2;38 @3(48 (468 698 9?= ?D= DJ= JxP= PkQ8 kQ(T= (TU> UZV<7 ZVY8 YZ8 Z[8 [] 9 ]_$9 _e= ef > fi<> ikT> kld> l`o> `oq> qq> qxt> xtv> vv49 vyP9 y{h9 0{{, {{+ { |, |p|+ ||+ |}, }}+ }~6 ~> ? ) x9 X9 Xm9 p, h) p, 3+ @v, ҆x9 ҆9 -9 0vHA H9 H9 ߎ: r( , Ƒ) q@8 q: Ö,: Ж\A ͘<: ͘r\: rp: Y: Y: U: `,7 ,: s: 6 ? ,? Y7 Y: `: `I ; I,; `@; `+ : EP; Ed; i7 it; p; p6 * >x* xg$* a) p@ F@ FX@ `+ s+ ) ) ?* @+ , $? 0{? \- $. Pm( m( ) . . ?p3 ?C3 C93 9(:3 (:I:3 I::3 ::4 ::4 :B$4 BuUH4 UUh4 UU4 U e4 eg4 g,h4 ,h4h4 4hh4 hh4 hh5 hh5 h,5 P5 yd5 y5 5 ۪D/ Zh/ `y|/ yP/ Pk/ p/ O/ Ol/ l 0 0 / @0 L0 `0 0 0 /0 00 0 0 1 1 601 677H1 777\1 7>8l1 >8^8|1 ^881 881 881 881 8E1 EF2 F/F2 /FZF$2 `FF42 FOL2 OOp2 OP2 P`R,/ `RR2 Rmf2 mff2 ff2 ff3 f3 @3 xP3 . . ِ/ t5 ?) @) 5 ) a) pDK Pß) ПK 5TE 5VlE VܡE ܡSE SpE pE E d, pxJ J mJ mJ K K $K J C$J CD8J P8I LI H`I HMtI M`I `H YH YI I ȳ(I г) H H  H ϶H ϶H (I (I  I I I I ¸J и+ XH -lH 0G G "H 0) 6 H 6o4H oHH + ߻8E VE `8E 8E ǼLG Ǽ XG lG |G ,F B( BG ԾG F 4K HK 5hK @,F (E 6 G @HJ E TF @F (E 8E  G :K @W+ `t+ + F ,8E 0`E `A ;A ;^A `A <A <_B `xB x(B PB `B pB B A mB mnB pB  B !B 0hD hD O,D OILD IhdD hntD nD D D D D D E  E  C 4C \C lC C  C DC DEC P@F + ppA pA A A F + 4DE @hF |F F F }F }F a@F pF AG PL L L 7 M 7T$M T4M $M + 1F @E 8E K K K 4(E @lN pE lL YL YL =L @p(E pL +PL 0, 7O @qXN qhN N 2N 2GN `p+ pN `L @E @p+ pPM dM xM M M M M @F  M M N B0N BbHN p4@F @Z L Z,L @L  O =+ @dO pO  + 6tO 8O  P !) $X, X+