(412 HW #2 nx a 1.0 Ze ~Nt0.1) ¥oz0 Rec (0) = EeLLeo, $.xy)] = fy t® $Se000)fy90 ob oben = Jes lance Digetx-cv, zece7}) Ty fire obec 2rer- Pl-cosk-9s co) rer - Pl-cmn s 222 < cya) areg - (1- 201- Blom) ) = 2crg ~2&CCn) + I tb) Rye = aero - BI e oly + | 3. Ree) = 20e- fe® Yea2G 70 eR set0>lece,~ ERS.) [ozes= “et He Paso) wong on celo,too) 1: oe y SiR. ) 70 «c) avlg? 9 23 oes Futele) if rv? Jha aS < > C207 Sutte fax) Yerd > Reo) is minimizect of c= 0 By chy, if c2 Felt) ae Rbs6) >0 tox Jel) Psi 2 Cons Flats) zr WW] © -2¢< Rico) G8) FeO) 94 Jp at MEMS 5. eT (2) OelX- =e at Bytc) Ms ik” By cot) 4. FutE) FE) 2 = sep)2 (b) Reto = Exlue.suc] « L1@,5:00) Py (x=) + Lt9, 800) Potx=i) Stoy= ay, Yeaches wrong decision) ol S Polx=1)=Po( $ reaches wrong dlecision) if Pytx-0)= 0 = Pol § reaches wrong decision) if Mar Scry, Ret) = Potx=1)+ Pov) =1 = Pol § reaches wrong decision) >Rste = Pols Yeoches wrong olecision) O Six) obo=B Rs tr aM =z 3 @ Sure eS Raton = “tb, 3.) = Pa S42 ob) = Bic 1) pd) Jal des ol) = Pytx= 0) (©) Als nonranclomized procecutes are just proveolures menbionect in tb G) 81, 8,83 are nob inaotmissible-> 5, be, ore aotmissible- 24 Ui) With priot distribution T, ts(t)= Reryticsy + Reraymca) 3 > Psa) OL Mest =04 Pat = Fo Mgt =So SPs cys "FCM) , Ss Gi) Similarly, MET =0.b ReCT= OF MSs) Lgl) = Ob => MgacT)= Past) = "A" PST) , as 83 Bayes with respect to this TI Pss ober oly = 2 Sunech 4 ta zh 3: (0 Suppose & is Bayes with respect +0 Some prion olistri bution Ti: TMS)=P Ta): rp Ts.) $ Psat) 3 Ps.m) s Psyc) Psi Psi) S Pett) 3 7 2 Siis Reyes with respect t0 any TI satifying TW) =3. tem = RecfoTitiy = Lefa( Scots, semyfiemyctx) = fs2mefnLefi seroyfie obx = Faefn willis tou} fitxy Lot Hann £ SuiLiatanrh ee tet) 4 Sis Bayes with respecd to TS? YES shiny € AMI, hn) ¥% fixed . Sfx): lo or obi > he Uc.ote) = We ANTI, ne Ki trcot) R00 wane ibe 6, ol) = Wefe Wo TItfe) Frum, 0l:) Fe) “wy Dif pele FO wma Fup = ola For) wy ef) Pe[q@ FOO wmf Sys ole on oh fot) vy Teh) poles FO _ wmefy Soo =o, Feta)” wef.) ry > « > Mead ie where o2 Sede abe Pee4 ta) m= fR@Tole = fo 19) (Sg19-s01"Fp rola) oto = Fs( Ei 0-grxalFpto 18) fA) otx = Jy. (JB 16-scxoI"rree1%y ote) ma (50 ola Lot hn tot)? Joip-tmiaia dg, teD Then § Boyes ©? YxeS sue) e DRED hyo) (by Et marx} Jan emotxy ote When P22, hate) =/n0-ds THelnoe sal 2EIML@lxs) ot + La satOlxrl® —quaolaitic in ob 2 Ned hava) = EUTtel x) (©) Gi) Firsh we will Show if Sexy is a mection of THOIX VES, than 1 Bayes Far ise ctaneke. ak Os arte Bred romontt Bey © mis a medion, ¥c#m If mcc, let hie (18-cl-1-ml) - ¢C-m)signm-) whea @sm0 When mMD iff OEtm.C) > J 19-6191) oto ~ [1-migia) oe ZCC-m) [Signcm-O) 9.0108 2 (cm) (Frgemd 41) 08 - Srgrm49L0) of6)=0. If mC, when Bsc0 When Cems hil8) = ((0-C)- (8-m)) - rc) =0 58 hiB)20 and hle)>0 iff GEtc.m) 5 J 16-c1 400) oto ~ J1d-m1g916) 8 2 (c-m)[ Signi) 9.9908 = COMMFroem’ 418) 6 - Sroxm'910) ob6]= 0 D¥ctm, we always bawe [19-c19Le)o > Sig-migi9ol8 For another meotianm, f10-Fi1910) 00> J lO-migqiarote, Sia-migiareee [1 0-F1gt9) 8 D> S1p-migio) oO = S19- KIFUO) AD > Amedianisy of 919) S argmin Sig-ciqup rote @F or 9t@)= Te1x) vx, if proceoture. § Sobifies Stx) € i meotions of THOIZ)Y YES > Son Eogrgin hao) ve S DS Boyes Wiy Next we Show thik if § is Bayes the Sx) iso mection of TOI yxES hncwdy = fe 1@-ctiT@lx0) ke D=Lasb] etnnx@2zott (Oct) THO 1x.) 009 + careanaBeott (A-O) TT te1%) oL8 = PPO-h) MIx) ol + SX (ot-@) TEI) AO . SBayos > WES 500 © Goan ( [2 cot-py Tl redold + orl) TBI xc)0l0 )Mote thet £ hawo) = fem ielxyolb- [S191 continuous in o > Fhalden = Sp miolm0<0 = SRrieteol8>0 an “Yashntxct) = 21 tollx) >D ‘ Sal dx eLosbd Site Sahnrol)=0 anct dotfost ="Veeraind hacrot ) So § Bayes & Sa =oFixy vxeS ‘ : =I tt Since Faphnexidd=0 PMA ay = [hee MBINIOD 3% Suy= AX(x) is a meolion of TBI VxES Here we prove tus ip § is Bayes, Six is ameotian of Tiisix) YxES Sahrlae Byis, we knpw that Six) is ameotion of TUX) VxES 3 § is Boyos Byciid, we know that Sis Bayos> Six) is amection of Train) YxeS, Sowe conchucte that —§ is Bayes €) 8x) is mention OF THOIm YxES.