Solutions Manual Probability and Random Processes with Applications to Signal Processing Third Edition Henry Stark - John W. Woods Upper Saddle River, NJ 07458Acquisitions Editor: Eric Frank Supplement Editor: Jessica Romeo Executive Managing Editor: Vince O'Brien Managing Editor: David A. George Production Editor: Barbara A. Till Supplement Cover Manager: Paul Gourhan Supplement Cover Designer: PM Worksiop Inc. Manufacturing Buyer: Tene Kahn PRP © 2002, 1994,1986 by Prentice Hall, [Ee Prentice-Hall, Inc. Upper Saddle River, NJ 07458 All rights reserved. No part of this book may be reproduced in any form or by any means, ‘without permission in writing from the publister. ‘The author and publisher of this book have used their best efforts in preparing this book. These efforts include the development, research, and testing of the theories and programs to determine their effectiveness. The author and publisher make no warranty of any kind, expressed or implied, with regard to these programs or the documentation contained in this book. The author and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising out of, the furnishing, performance, or use of these programs. Printed in the United States of America 10987654321 ISBN O-13-040787-9 Pearson Education Ltd., London Pearson Education Australia Pty. Ltd., Sydney Pearson Education Singapore, Pte. Ltd Pearson Education North Asia Ltd., Hong Kong Pearson Education Canada, Ine., Toronto Pearson Educacion de Mexico, 8.4. de C.V. Pearson Education—Japan, Tokyo Pearson Education Malaysia, Pte. Ltd,Contents Preface 1 Introduction to Probability 2 Random Variables 3 Functions of Random Variables 4 Expectation and Introduction to Estimation 5 Random Vectors and Parameter Estimation 6 Random Sequences 7 Random Processes 8 Advanced Topics in Random Processes 9 Applications to Statistical Signal Processing iii iv 34 59 94 138 159) au 304Preface This manual contains the solutions to the problems in Probability and Random Processes with Applications to Signal Processing, 3" Edition, The solutions to older problems, i... those appearing in earlier editions were prepared by Messrs. F. C. Jeng, R. Wu, and the authors H. Stark and J. W. Woods. The solutions to the many new problems in the 3" edition were prepared by the authors themselves. We take responsibility for all errors in the solutions (hopefully there aren’t too many) and welcome corrections and comments from those that use this manual AS in the first and second edition of this book, many of the problems are of the drill variety. Others extend the discussion in the text and lead to important results that supplement the theory. A number of exercises ask students to write Matlab. files to solve certain problems where the graphical results convey more information than complicated equations. The Matlab.m files furnished by the authors are far from being efficient, as neither of us are expert in programming. Some problems are of the “research” variety, meaning only that they might require serious thought or, possibly, help from other sources, Both authors—having taught this material for more years than they care to admit—are convinced that solving homework problems are among the most important teaching aids in understanding the material. When a student says, “I understand the material but I can’t do the homework,” it usually means that he or she doesn’t understand the material in the first place. Often, the student's lack of understanding is first revealed when he or she attempts to solve the assigned problems. Another warning sign that students are not following is when they say, “I know how to do the problem, but I failed to get the right answer because of silly mistakes.” Very often so-called silly mistakes are not silly at al. We have compiled this Solutions Manual as a service to the instructor. We hope that the solutions are described in sufficient detail to be effective Henry Stark John W. Woods ivaD CHAPTER 4 A large number o people with personalities , backgroids , ne of rere to rit should be exposed to an identical situation in which theft is possible . The fraction of those that steal in relative to the had wt qe some meoning -- in the relative Tepuency Sense only -- to the rase “Ralph is possibly quilty of theft .”. post ey f ef (hz) Nole thet D3 bt 34D ey D dapbies 3 but not th other woe form over wou around. Thos ‘ft ay eshot ? Covd 2 avd Ped a 3 TH was never stoled tht a 2S. Li fewisr with cavd 3. On te other haud if we twra over card + : 3, ther tk vole eee ew : “a mauct tora is violated - Hence aud Ag rie ver card + card A) of cow o 4The number of fovoroble oueoner 1S four: 24,68, a tote! number oF outcomes is g. Hence PL even] = 4/3 lie a's assumphow: all outcomes equally Ki ely A Theve ave. gus outcomes ; HHH GD GTB THD TT THT TH TTT. Th circled outcomes are te dasi- red evewts, Hence P Laheads, stail]= 3/8. The possible outcones gre: Zl 3I 12 @ 32 13 a3 @ The circle! outcomes are when the sane ball Is drawn twice. Hence PL drowsing. ball twice|= 3/9 = 3° A:Let b,,bs., be represent the six balls. Then 4 { bybs, bibs, brba, bbs , bbe, bbs, bobs, baba Bibs, babe, babi, bybs, babs, babs, ube, eb, buds, by, be be. baby, by by, beh, bsby, bly, bsbe, br, bobs, be by dy be, bebs } If the first ball is replaced before the second draw then $25*,U {bb babs, bibs baby, by bs, bebe } ED) PH Let y= height of man , Ay = height of woman, Then a) 2* f hu hy me Ayre} a s bE = fhehw theres by res Auch Ff eS To shirten the notation, the event of drawing ball with the number ¢ will be denoted as é, Then Ae fhan lo}, Bf lass}, Fe f456 78} Thus + E°:f6 T8710}, Fo= $12.3, 7,10} EF =]4,8), EUF ={b23.45.6 78} EFS=f 1,3} , ESF 4678}, LUFSL 23,67 8Z/of (EF IEF) = 162,3,6,78} , EFUCEF) « [4,5-9./0} (EVF)= $810} (EF = fb23, 67 89,00}, where : E°«{ drawing a ball greater than 5} Fo | dreaisg a ball net in the rage 4-8 inclusive | EF*| drawing @ ball greater than 3 but no greater ‘than 5 } The other events are sintlarly described . esFor this experiment , all pessille outcomes are Hl 2) 31 4 St 12 22 32 42 52 13° 23 33 43 33 14 24 34 AA SA 15 a8 35 45 $5 There are 5%5 +26 persililities . The favarable outcomes are 4: (1,4),(2,3), (3,29, 16,1) 30 P2 is indeed correct , “Dim” ignored the fact that (iy) is different from GPs ie. , the outcome (2,3) is distinct from (3,2) , “Dense” ignored the fact that the 7 distinguishable outcomes are not golly Iekely PS a) Qu¢=2 aged Hence PCQU4) = PLA) + PCH) =PLaNe1 Therefore PC#)= 1-Ptay=0 b) B= EFSUEF CEFINCEFI= ¢ Hence P(E)*P(EFItPCEF) or PCEF) = PCE) -PLEF) ¢)BUE*Q EES PE)tPE)~ Pea: mS GID cor: ery FE ; but CEFN CFED=4 Hence P(E ®F)= P(EFI+ POE) ~ aid PCEF)= PCEJ-PLEF) and PCFED = PUFI-PEEF) Hence PLE ®F) = PLE) + PLFI- 2 PCEF) oSG3) qyFor Swap icity arsoeshi as follows: Cat=t, degoa, got=3, pig ea. Then 4b (4) consists of 16 Quast: UGB EM: (2) fa 14) 24: Baba, {he {wg}; {o34}; fa 344; a. ae } Note thet {yz} = {iy Ufzt ama 72 P(E = PTH + PLeegT- Piantet But Pf 2} = OS, there fore A [209-05 204 P[{34t]= ors Pl{et+ Pc t231 = oos+ PL ft] Thevegore Pfs] = 0.05; P[{et]=os P[{4t]- 0.05 PLL] =04. Fron these Probabilities we com compole all ote Proboliities using PauBl= Pat PB] ip AnB=ad. For exane’ PLE wat) = PC Saye PLL PL{s4] = O4 + 0.054 O05= 0.5. Tk underlying expevinent assunes the animals ave ih separate covered cage and 4 cage is closan ot randew.(2) WR Commet uae Te Il Ih evet Field of pot (%) becovse we hae sess pro dab thy inform Tide 70 wl. Fav Cx duple we Con determne PL feat x] = PI fest dog - {dos ] = P [feet a5] - PLtdogt] = 049-05 =0-Y , bub we Comot determine Piri], we ney ae only Aven PL farses) | ad Tree (8 no other velev at jnhimrton. — Thee wort ke spechied a pro botilaty fv every cw | So we dofre ated of quit Q erate: prexbas Sanat pig Qa a Singlet ov elemen tang bet. Use +5 Tha wapp! *5 evant 3 trom pact (a) , Wwe hovey Fe) 5 1245 1245 Bly Lash F234} 5 SL3354%5 6} and each Crt hea a pro babilityP[AJ=6/8; P[B]=6/8 P[AB]=nas/ntr=4/8 is not equal to P[A]P[B]=9/16 Therefore A and B are not independent.) Let ni, T=1,2 Ceprsewt the ovteome of the Uth toss. Ciuce the tosses! are \ndependent : P yng = Piny P(na] = Caen P [rtys?)=3|= P [ns4|n=3] = Ph=3{Piyss\ Paap re Clear PIAI= P [B/= P[AB[= Ps picking ome of two red aces Qk Qs as “2 th 52 cords = a Is Piva’ eat Pre 2 aa = a. oa yes! Hence A owd B are mndependest AdLD All possible oukames are giver below : Wot Bt 4te- at 1222 32 --- 92 13431 ‘ 81 possible met outcomes, sat . 1929 -- - 9% To get E=T, we need one of the following outcomes: Gl, 62,43, 34,25, 16, Thus ‘there 6 ways of gett Z?7. Hence oe 7 = Plz, Zodd) _ PLZT eae) P(zodd) Pt Zodd) The only unknown is Pt Zod). Out of 81 cutcomes, “Ao produce odd Z, 41 even, Thus P(2T |Z edd) = S48! Serer 7 O'S. Now we want to compute Pat least one of NM 7TIZ>10) Sp’. We can urite . Cat LA Zr) P= = e PCz>/0) Frm the table , P(Z7r0) =36/81 . It is easy to see from the table that the only waysMam Zo and both NG ST are the followin Emil s 645+ S46 = +d ht7, Beiaz SHT~ T+ 2046, Z=356+T2T+6 , Z=/4= 77, that is, the 36 possible outcomes are reduced by 10 and Plat least one of AAG >T, 2710) = ey Hence 1, 284 P'" Bifar = O72. Finally to compute PlFodd . Ni>8) (z @) = Pized’. M7 a) Pz odd] N>8) Pinr>ey We know thet PIG >@)= PIN? 9)* I/2 . Also fron the table we see that the event { Zodd, = 7} can happen as follows: 92, 944, 746,948. Thus Pl Zodd’,“; >a) = 4/ai. Hence P( xodd IW, > 8) = $f =$ 2044, PS Oe 7)We ave quuen Het — P [DI= 0.004 Let T=D. denote the evewt that He feck says thal for have D avd let T= Dv dence +e event hot tle test Says that you dowt hae D. Then P frepipj=t ond PLT Pipy|= oes We wish to compebe Pipie a: pprege PO%A. eines PEED] PfolplPpAe Arp] Pa] = 1 x 0.001 | 1x O.col + O.0oSK 6.999 | H 4aRS = OME Thes tu only IZ oo op te cases will a positive test result actally. Con firs that You suffer from D. Th other 83% op He time you will be needtesel y worried ! ASLet S, dense He sch of Oecopahous aud S, the set ee iu kereste and/or hobbies. Then Dy =] office momogan, exgirecr, doctor, feacher..) S.*{ nature degence, books, muste, cooking, ) Let X deude Henrietta’ occopabon and Y her interest. Then PX = opgice magn Le rab detente| = PL Y= nat degense| K= ofpice wngr] x PLX= ofgice mnge| S PEK= oppice mngel siuce OS Pe vat. degense| X= Ab off rmngy <1.PLXe]= 3 PL] P[X=t]= 2- PLA] P [X=3]+ p[X=a2|+ P Le =! P[xXail= ¢ PRGA PlRSz Pixei|yayf= PCY eed 2 Pe PLX=i] = ¢9% (-Np + Bet ee [— C2) Let As aod Tha examinee Fnrous Bs = evant that examinee quesses CS eveat og geting wight answer P(Al= P Pial= IP PIAY=1 PR |S l=ae PIAL PA] P(Alcj= mal aP PLAT PAl+ reielrel 5 eee ~ pa si (HA) mM PH IHP ALFEF AE AE FE AE HH (C22) There are N cowkdhauts ond only owe Most beoulipel. Hence PCpick vost beautiful) =47-. tH ©) Let Ref eveut Hat rand. drown chip is & at Let Bef eveut that roud. drawn chip is € BY Let f= 5 eveut thet rand. drawn chip ic € S Let D=Gevert that rand. dean chip ic depective PLD]= P(DIATPAJ+ pLoleyrisy+ Pbia)eel 0.05 xX 025 + 0.04% 0.35 + 0.02 x 040 0.0245 Hence 0.05 x 0.25. P(Ma= Pir Plal= TMT ious PLS D= PDE) PIB) _ 0.04. 0.35 « aaeg Pip) 0.03845 Pe [lo] = PPIZ] PE] _ 0.02x 040 Pp] do34y 794 orPUD) =~ ly B Let xa | Then Bo x x (D) ° ° 5 346 Al +23 & 31 2 +32 “7 +25 3 361 8 18 4 367 iu le The peok is guite shallow. Therefore choice of x not critical near peak. TT EBD The number of fe varable wags iS rer-yer-ay~ 124! , The total num r of wags is nenwnenh. Hence Pert/nt TF times a As in 1.25, the number of favorable wags is Tr! However, total number of ways is no nt since cells can at mest hold only one ball . For first ball there are n cells ; for second mot cells ets. Thus Ny = non-)--(n-ren = nt fn-r! Thus earth _ optenen © nyt P* aifaeni =~ ne = (7) TSEr) For oN pockets tere ave nl wage of arranging themacelves but ond one way of doing Bb covvectly, Hence aP lowed ascembly [>t J =Ger2604 2.26 x 167 for N= 106 ava For 3 pockets there are 6 chigperent- arrangements (3:241=6) bub ond, one correct one, Hence. on - try: pe P [sucesd]= + ) BS + Ps For 0 rest corvect ven assenbl, on the nth mH, there moth be n-| foilure , followed by a svecess on He wth try or Pils a nal, First ole that S Pes be 28) ye To Eid a goch "phat SPwzoes solve 0.95% 248 ° ~ - @. Hence nd Ln (0095 6.5Eb hin) = B (2) php 2 Cpe gy = Cprcr-poys 194, TJ £304) The prob. that a BM gee destroyed is 1- preb.f both AMM miss} = 1- (0.2)(0.2) = 0.96 , Hence for all BMs te get destroyed , We need six wins in six tries: (E) (0.96)6( 0.04)" = (0.96) = 0.78, b) Pfat least one BM gets threash t = 1-Pfall are destroyed } = 1-078 = 0,22 c) Pfexactly one gets through } = (£) (69eF cay’ = 6x(a2shx 004 = 0.188, mwas CSD We want to compute Plone BM got thrayh | target destroyed } =P fone BM throagh| at least one BM through | = Plone BM through ,at least one threw P fat least one through | . Pf oe BM through | 22 ” Pf at least one “throwh} © 0.28 wT = 0.9!>) Let Az f event that o chip meets specs. Let Bzfeveut +het o chip heeds weworkeh Let cz t event that a chip is discarded} PlAl= oes PB]= 0.0 Ple]= 0.05 The multinomial law applies here! a) Pfall to chips eet spees.| = _lo! lo ° ° infetol (0.85) (0.10)° (6.05)°= 0.19% » Pla or more discards{ 4 B s1- Po discards[- P[4 drcoral| P[discord|= P(C]= 00s P[c*]= 0.95 Therefore a BI Be (‘SJesseos™ ('°) @asioos) = |— 0.6-0315= 0.085 QP Is meet specs, d weeds work, 4 Aiscord | tol 8 t 1 = snnr ©.8D G10) @os) = 0.122 aa(33) Slot A=Levest that call ic to report five energencat 5 Be Levent that call tr to policet 3 CeLevent thet call is for anbulanse} PIA]: os P[B]= 0.60 Plc]= 0.25 The seyuewe 0202 0202 030102030202 Contains one A, six Bis and three Cs. P[BCBECABCRE]= (als) Gio) (026) = Il x16" b This is the multinomial coegetcrent lo! _ [0-9-6-# Clalit Bal = 840 Rr F— We use the Poi cson the bine Approxmation Yo 3 al: Eq. (09-2) 7 4 7 qos = 16 ) N=loo p= io! Pat Ranct ona dkamond found)= + P[mo dianonds Pound | = d- (15')"/o! - ene 1-agx= oL(35) Use Ep 1.1OD with tro, t+ Tol and Att as geen This gives Lt oF ha5 = 06! = 3.68. Thus PCA clicks in first 10 seconds) -368 {| 4 ze a 3.68", Lf all tickets in one lottery , then Plwin)= $3 =f, If on ticket in each of So lotteries , then probability of a win per lottery is Vieo P but we have 50 chances ! Hence Plat least one win) = 1- P (no win) Pino wind = (82) (Joa) (Fae) = 0,608 Hence PCat least one win) = 1-0.605 =a3vbR-o01, The probability of patient dying with the monitor System » Pia- Fa = (0.2)" (01) * 20d, St shows that Prof Xz agument is wroyy f PrGAD a) Mt each attempt, the probability of succesehl tronsmission = Pl)? Pfat least one successful transmission eccures in m attempts } 1-P{no successful transmission oceures in'm attempts } 1- (3) pr’ C4 pO” = 4-¢1-p%)” b) PL Fr each individual receiver, at least one successful tonsmissim oceares In an attempts } 2 1- Pf Fir each individual receiver, no succesefal transmission occures in m attempts } = 1-(9) p'a-p” = 4- (1-p)” Bim) 2 Pi Forall receivers, at least one sucessfil transmission Occures in m attemp:} = it- ceprye af 7209 ,WS , wee PO) = 0,832 < Brad= 0.96) PsThe sample space for thir ceeerimert is £ %; S12, ttyl é PL geting lo sevens th 100 tries = (SO) ROR note (10 ©) and Cp” ave cbigpicu + do evaluak. Hence vse Poisson approximation : lta. =np= +e... The PLaeths lo see Wm tee abies] x @ a. 2 (eyed = 2.08 Hr eTWe do partchy givet: From the landlerd’c Point ot view the following applies, Te lease includes fete vepaits | then tha cost of two “Cloggers” oe ohe “Nevergail” is IL sane So tt deesurt matter. Te we pairs are not ree avd are te sane for th "Cs" as He Ni? then clearly the "Never fail ” is cheaper to lease. a) From tle Lenauts’ powt of view: Pet leact one’ Clog gee” on] = I~ P[bath gail] = i- (04) =0. 84 PL Nevergal” onl= 0.80 < 0. Theregors tle too" Clog gers “ave betler sce 84% oF tle tine ect least one urll be tuo King JOHThis pecbien ty piges the fossthy Feaconing exhibuled by Mana peeple wot fomiliar with probability owl, in particular; tle idea of independent events. While iP is true thet PL2 bombs on board]s ior th issue is PL beworist bomb| "nervovr” has bomb |= PL terrorist bonb| = to” since th eveuts: As {Eerrorist bomb on beard, Bal mewvors has bonb on beava} are mdependeut. Heme PAIBJ= PIA]; therepore no prolection! Ry Let E=fevest op successfull +tyansnission on chort patht 5 Fafevest of. soceess full trans, on long pothy. They P [success. trancm.[= 1- P[ESF*] PLE‘]= '- p°5 PLFS]= ps. Theregers P { soceess. tromsrs.| = \- C-p-PD = ptephp®. Nole pene ASTh ths core tHe Tel Com antg hf fiud th umion clivective cnreaconable. Here is wha: _ © ~(#20x«8 4 Phovetine[= > © ) ee) nz 5 Fl ul! 4 ~ i EP Var’? J © 4 where £)= SHEI— SF Hb0-08 \V S#60 dy= om ~ 54604058 V5360 mo = 0. Hence 7 - 7 = P [ overtime] = | ° Se halethe-time Cuortc will collect overtime. Not bad! AS a= a ae dx= exped=s.Un portvna bel very. small. The Feader shold Pecoguige that thee te an Occoparey, problem with the candies bed He bails’ and He stedewks béwna He cells” The exnproprsts formula, is 1.244 which — gives Ho provabr lity thet no cell \g empl Hence with = 15 Malo Pwo student is withost o Candy te oo" i os a |) (i-t) & 08§ 2 GJ Cre) A Matlab. m file to do thic problem 's ‘occupancy (balls, cells) k) *prod(1:cel1s) / (prod(1:k) *prod(1:cel1s~k) ) 7DISTRIBUTION OF CANDIES AMONG lo STUDENTS ty a sas neg et aca & : & tay a ttre aan cy AC48) Thc Is a classic problem and Ahe caution is onexpected. Let As=Levewt that no two people have their bwthdays on th some dale}. Let B = [ event that ot l\eart two people bave theiv birthdays on Sane date). Then B= x ond Ple\= PAT}= 1- PIAL P[A]= Meck wes A com occur 2 a. ho. of all possible ovteonus “r Wn= 365 (365-1) (365-2) «+ (365- CH-)) nes (365)" n pes POAI= es Om orG- 36s)" = or Dts Len (Im eg) tot An (i- Bb) = -07 For omy small % Lu (i-#) & -% Se 1 oa do for th solution fo have nea~ wing, we obtain ney sil 22.6 n=23, Since wo mort be an integer Thos ouly 22 people need to be tm a Yoon before thy probability is one-hale that ot leat two will have thew birthdoyr on th some dote eTThe Progranning ot hit problem Ts quite simp os ch Vequines onnlys & Towden wonber genembor N% timer for each veoligation... thats basteally it The herd part is the search for a Percolocting path. The latice contains beth branches and loops. Thos tt ic nether tree, nor areeh. The furct decision is to detine a conducting. poth. conduction allewed conduction allowed thvovgh dkagonals only. through hor, vert. Tis condoching clenents Two choices of Conducting paths Te Hine - comsurting, port is the search, Thos ig You come ts a node Cyvuction) awl th path Yee choose doesnt lead anyguhere , you rmuct be ecovepel to relurn fo the node and try, the other poth.An_exangle A. possible poh is LAB@E2. Note the dead-end ot D ond te possibilty og endless looping tf you ave wet careful, Siwce NS 50, relatively. simp search techntqves should work. A qood MS thesis, Good ek!Let A= door picked by you let B= door picked by ‘Me let C= renaini ace Let D= door that” leads to Rexis Then AD = evant Hat door picked by you leads to Rexis; BD= evet thet door picked by Me leady to Rexic; Cds evewt that Te mating door leads ty Rexd. Then Pape; P[sBpu ep]=2 But sme BOACD=¢p p[epucpl= P[Bpl+ P(en]= > Bt me alwoys chooses wrong door! ee Plgp|=o ond pPlep]=+. There gore you shold switch +o door C. This call double your probability of Winning tLe Rexis! PeCHAPTER 2 2D poen-2 &o s o . PIK=2) = Z,(/?) aay cory? = 057 (12) coaco7y?™ = 0.383 PIX=31= z£ fy (oad (0,79 = 0.0472 Piteai= Z (PY road cars = 0. (envy) — b) By differentiating : K/2 , o8xsd 1/4, 12X82 fo= Oo . asxsto ho Yao , Is x20 7 30K ©, xB20 Se Lfiodn =ff'% dx +S kdxe fo dx mt, co) Let X be te waiting time . Then |. PLXzs0) = 1- Pixs/o] = 1- Fi) = 4/2 2. P{xsS] = Fs) = ¥2 3, Plé 4 ©) fay = C143 €% Ux +h SH-9) + HEMI UY Piasx<3] = fia dx £2 sixndx d)P [axes] = (2 /,432%dx +4 (2 staxmdx ©) Fs) = Packea] +f e%dx + Li sxadx Pom *P [X = PUA. Xeven) _PUXeren IA) Peed) PX | X even) PiXeven) = Pl Xeven) =o A= 4,3 ~ Pte hy P(X even) P (Keven) = P(X*0)+ PiX=2) +P(X=4) = (23. f+ (Sy +(Jy Pike 0) = (4)(e*= % > PIx=0 Xeven) = i = Poxsa) =(Q(df= ff & PcxealXeven) = P(xsd) = (Q)(bt = 7h > Psd) Xeven) =H eS GD « bulbe produced by Born ~ APM Ny = 37% Mt 3sm=n Amen mr 92 AE 2 Pde MWh = & PCB)= 34 Fonaye (= 2 ue Buia) =(1- 2) wen Riw= Feaia)) PUA) + Fi (X1B) PCB) P(e tac POF Fay=.860 Fis) = 85 FCT) =.92 Pburns at least 2 months ) = 1- Fra) = 0.44 Pcburns at least months) = 1- Fis) = 01S Plburns at least 7 months) = 1-F (7) = 0.08 CSGad Given At fbexsa}, we calculate“ Fe(X1A), @) Xb, i) x>Qt Fe txtared, since Xex for X74 is a certain event 8. ti) bexcat Fy(xtad = POXs%, A) /POA). But for bexI since S tecoete!Pattiny all te ere gate we qet x= ) | ) a . PLX>b|= 1- RO + RO or Foes > Hovepere be (64). ON agbs Etb-Es & on abk- bai ao bye! +23 Sine bE (a4) ba IH VAR T= O18S RtWe ve Bayes gorsula foc pags: Cal) £ (C)) (aly = bye x fare fy oy fy bm fyaa= I fey Cone = Pri Ca? fy om Let g= at ae die, obtein a +F e dg = beg (4) - erp (34). But ere (Ws ~ eve C2. Hence fy a= Lee GA)- )- Cod, ae” ae ae) Then Lape Crld= SE ep Cae OD oTLet T be te preg.’s arrival time. ne We Ot 8:00 AN be origin. ° oo t 0, t60 6 P[A]= P[Tase]= 1 F(a P[gl= Plrssil= 5G) P [ABJ= P e eras 2 wat +O, Let au® x dus doy2 usoc/r. Ther PLX> xe = eae ores) 2(x+0) for XD0, A So. 3° pl xs wral Xda 4 P[Xana, Kal P| xX> But ude thet for X>a, the es 4 § Kd x+0, K>ahS[Xe xh n{ X>oh = {X> meat since [ X> xray is a subset ot bx >a}. tlence oe | ~2(x+a) _ 2xXtal PLxP el Kol Tigo om = e” inde perdeut of a)! Since, this cond. probabi lity is ud. ap a, the memory of @ has been lost PTWe need fo solve for 4 ie i- € TS 09s. But this iuep Ire thot 4d — bn, (0.08) ¥ 2996. Thus 4=3 should do. Age This ic a rather clacsic prodlers lu dletection theory Loe oy P fapd= P ExzeslM| =e ele a Pr e Ey = Le enpo5= 0.63 Palm] = al ene =-L-erpes~ 03 PRINT ae A eda = tet erped= ost om Pixinl> ga ez pale —erg ne 0.3L AH aA ~ 2Gad From Bayete theorem, Pfalaj= PAN PCS POU Pay = 068 ————___ IAI 0.69 Pil] +03 1[1- Pil) o3t__- PM 0.64 Pfd]+ 0.31 [- Pi] PiMia‘= ep rod oa___PLHI P facla)= Phi ee . PS oar Phe 068A] Pinca‘ = PBAMTPIS ggg (— PM Pad 0.31 PIM|+0.69 f-Pe] Ae a check we note the PIN)AT> PRCIAJ=1 and Lduwier for P [JAC + Periage. For PIME 162 For PME 10° P [MIA] = 0.002 P Mla] 2.2 x10" P IMAI ™ 0.498 Pivelaq= 1 Petal P [MUAT] & 045x167 Pp (Ml A= oS x 10% peactay = t= eM [Decavse of He uncertowty wm PIX \ettle value tm P [M*|A{ = 0.9996 these viumaberg hae decry avon Amotring : Aog correct " yao function [alpha beta)=roc(r) Noo by \i B, Pre he ie : (alpha (2: 100) ,beta(2#100) oun ot 02 03 04 05 06 07 08 08 1 aL, probability ef a folie alavon,GZ2D Fron te dela A= doxto%sh/sec Doring the couwbing mlerval Abs [o-Ssec. Hence at = 30 P [Palee lara tin ex|= Ploph. er] + P[L ph» cr} = @ Ger, EGs! { wy ° P [at Bod 4 false alarm i 10° ties] u (= p [fo fale olarms vw lof tues] 6 : >| =\3,\10 a (<) (aa6x18°) (\- 24.6x10 ) ~% 03x16" " Aa De {S24 where G= green Rs ved Y= yellos. The OH field og eveuty ore LOR 1RT, INF, FS, RF vey Bath te green ovved {G, Y} Bad 1s green ov yelles, {XG Regi is UA ov yellow, b (ome set) SL ( ertan event ) b XE)--1 X@=e X (elloa)= 7 PIel= PHelloul= 05 PLR| Hence P[Rl+ 05 PIRI+ os PIR]=! Pike + P {61= P [velleul=2- Lo in) — ge 075 tye 1% 025 ee 1 ——_ +4 --__4—— 7 ° Sar x ' I ! RQ@= OFS Ao) een = Bx5= 40 minctes Trin = 8XO= 0 mrinutes Note thet ie eack ctetion sends a Amescage thoh is im pinitesimally short yon ger Taranto. Leb 1 denole the warting Hime ond let P= probabil of a stoden bema busy. Thon, E(r]=- [2-5 p+ a-pr4la G minvles Bor p=o 20 minvtes for pl [4 matucks for p= 0.5 wouopWAITING TINE Matlalb m.file for computing 215 etd) 50 Monte Carlo simulations of Token System enti: Senna ennnnnR DERN PnEEEEE [ ° —————— — ——— LY eS — —————] So _——— _—— 2 | oo ——————— | a © SS ————=—— ———— | OO — | — oo ° Oo —— ° — — — sajna own Bunjem trial number: Ew) P[X+Y"s e[= Wa e “dy Coy): eyesee let rs (eye dx days raed C Ne) surrteh te polar co-ordinates) , Then cm, PE XY sef= Le ( [et rdede Yo 6=0 CG LAL = t e de Y=0 Let us rh daz rd Then 2 a7 -c P[X* Yeas (ete ee = In = 0.95 -h € = 005= ts Ch = Au 2023 c= fe '= 245 optQ2%) The pde of the failore time X is fe fy b= ewe = B ot in thie case Assume pis measured in (hours): Tp A= { evest of failure in 100 hee or lesrt then loo 4 PLA}= P(X wsl= [pet ae —~Miloo = 1-e iS S005 —Hloo e > 0.95 or, to King \oqs avd Saluing . HS 5.13 x10 PptTr geveral for ong ox (€) t lise b= x“@ exp iE £ act dt] U) for t Ks g6 yt (t- asjerpl-{ [tar feraga| (¢- 95) enpl-{s + €- ast)- (se-s5}] (t- 95) exp [- fH-ast+ 50} for t>10 u u For2.29) Let ax @ AY 0 PL xc X< xray “ 4 | : ow (rc eary you) — ey (4404) Bey (yt ay) + 5. Cm) 5 iz ne (te eet 444) ~ FE Olay Ax ~ | Fg Costa Bee CHD] ax mE Cota Fig Coa) Ox OK le Say Om — Ar dx eon Deby, > Ley (my) axa (c) From @ follows thet fay (G7) dx dy. >So since it is probability. Hence since d¥>oand dy >o it follows thet on (m4) 20, (b) Thie expvecston ic ae (<0, +9) £P[ Xs, Ysols 4. Fouw- — Foy aro*(- py exp laste Oe aezpmp |. Ig pee avd OF! ee 2 ¢ “On Loex eee me ™ ae (a)P[-4 33 | Base. For any. *! P[X2x]+ PL A0; fa y PL aX sal P[x0, only solufion +o 4-gQv=0 is yl o X= ye For 44.0) no real sduttons te Y-gtr=o. Hence fy Cd= a | gien= 4 Ke BON |e ( a age OL ae ay Thos -4i fy Gn= fz 4e uly) Ye = 0 fa ySo. Gt y=o: PL¥ad= P[X x4 Hence Iq /dx|=I and fe pe fe For = 0, Peo Pix 2l aeyrd-ep@=as23 Por yea PfYail= 05% (fires abou) Hew s ee eh -)-u S fy @ a G4) | + 0.523 ce a 401345").Gwen: Se re For oSx<4 gWqzx For Ls x< r sme 22K. Elsewhere giao. / For 4h no real rests to 909-459 pe is 0; Por o<4<4 glx)-y 2d has 2 ete, RX-H=O or X= H/2 Od A-RK-YRD OF we im gl. a PRO deere () Sol + fy (a= ol yr) Y-x=0-D Kea Cone fra g] = Pires 2 Pee HL 2 ee ee fe of ~ ae x4 (Ay ol E+ ae a. | aT Wee (ay VA For ea) two roots in (-T, 1) (cont'd)Too roots ore x)= + cos (a ele?) Also go)= (cos x) ‘=> qioy= Se ‘ene _ Use A NR to show [9/Cx)]= aE ¥ - x) Hence, fy O)= 2 oe = MYL \ = = For yé-4, again teso rook oud oblain(2<). For fees, no solufions fo Y-9=0- Therefore fy (we O° (*) ee . GB 2x fol, oleae uy Let Vs-¥, Ply 0 e& wAlel=4 7 Let w4lzl (ees Qonsider 4@= 2 we 4G\=0 | | two oaks Bae aa Ge © At Q yO, = 4; ot B ¥@L= + ty (wy = £09 /4+ fe Co9/t fa #>0 =o for arco Hewe ig £2 @4={,69) fe 2f,C3 ee. Let Z= min (X,Y) Pizsl= I P(X>3) Pixos] Hence. B@= li O-&@)) O- Fe @) = Rat &@-k@ FO) f20= f0+frO- LO Re@- fr0k@ £263 ° L AFTha event {2 cab is the sone as {mx (X, Ky K) ef a { Xsz, X sa, K sah. Sinen the XX; one mdopendests B@= BO Go. Smee th Xs are tid E@= Bao = Fy @ Hence 4 E@= [Rel AGay The eveut {2 pal { min(X, Ka) >2f Ys tdewheal with [Z>e}= f Xra, Kary X,>z Pl Z>e]= P[X >, X.>z,---, Kr>2| S P(X >). - P[X,>2| =f Ro: + [Rel Because th Xx; are Lid P[z >e|= [!- ee a? zis 5. Hence F@= re [i- Roy x, atGS = emax (Xyo7Xe) X's dep, P[2, al. Bo Plz. de> PRC max < Pd] = (- Ale “[F El , = [iO eae = a = 20 Hence Plz, = Ao = 0 # ARO neueWe ave giver W= X+Y, where X,Y ave ind. ident. dicte’ buted (td) with Xs UCWOSY. Then Pult= Lome pe9 fy Ji =z =] w A good wo to describe fue is ag fuer let |ret [1 Next, consider (v4) oe , “ 2* 270 Rests 0 2-qw)=0 ore WH ae i i. y * no geal vee) Hee fr 40% Lgl. awd [atghl= g-. Hence fya)= 2-H for O Lor fer 6 z to ee = P[2esl= - BO= I- g(a) Se totes 0875 ‘ a o “4th ‘) Gated area) oy Y P(gsn|e 5 Pix] PLY Al where Z= xr Heres hon a Plzou= Fe" z i ar? 3" - ar 2 (4) 2 rte 5” negl Herne 2 is Poisten with De= Ay + Dy. ae <= P)z=ol+ P= ws ?fee] +Plz=s & STs sat 5% + Sirat shir] e S[s+ 12.54 20,83 + 260442604] 0. 6) a RouGD a) Z=XtY where fara d rect (4) fp a rect (H) hans Choon hep ay Meter Fa-peg LPs & fees only for gr! < F< 5+ region 1: jtl<-2 or 3<-3 £9) 20 C70 overlap) 1 a . 3 2 region 22 -3<3<-l fay = day = oars) —__ a “eo Ng+t 2 region 3: ~249-1 but gei<2 p -1<3<1 fone PN gaged hs 2 33 Se (contd)in 5 S-1K2 0 3tl72 DS /e3e3 oF a a (ney) fia: FLAS OI rpm $23 Bars e b) 2=2Xx-Y let V#2X W2-Y Then £03) = hep hpet Piven) =PUvs 4) = RK) & hone they Pcwew) = PCy >-w) = 1-R cw > fis few) Hence theo Now proceed as in (a) ty ebtain tet»The real roots BRi,Ra te gex'ty wex are Rit xew , ye Sy-w wi0, =e Te ep (-3Ae] eis =O, otherwise -3/so> &G = g fiw (.wrdw = sh e ae ug. CS Try the transformation Cactually the inverse) x= zt bw ye tg+dw Then X4x9+7> = (age Buy?+ ¢ 25+ due + Cage bun (é3+dw We want the cress terms to vanish ; this will allow few = fx fw. The cress terms are (24B + odo 48d tbl) gw =0 choose @z2, $2-d, Then x=az+bw p43 -bwacy 2pxy = 3873" Bw | Hence Qa = + {( eal +laPal) a ee goes AE hows Ge With G2 7/8 = 8 BO/ad Hence I 21abt (JL = mag 12 i er i i ew = arg e 7 T a $a fw ra Let FONEECP hope te-7. The reots of gKN-vee hinyy-weo are Kw ne ew alow a ew RW = ow he Ww te The magnitude of J is mm 2 [+a is +h tall four Nag al 200 agement = avo Now observe: x°ryPz av every where Kaw =%GY, Rye VE = ayHence GD ot fat? ESTER ew PB’ 22 eae = ‘I -( 22S steer +H * TERE € HF cash 4). eS Introduce K#X-¥ , BA X*¥. Then Fw tw =o [da fap fy ($A write fiepy 2 fi, SY. Then dp? Fry) = t San” ap feap» ae ey Now, define Fp) [perpde ~ > te, the anti- Fiapre f fidprda derivatives Then Fymwrs 3 [" (Fie) Fea oalae a (iw Bw 4 1” Fawr de = AZo Room} tte. 4 Foam But Pow? £. (2% . 25%). Hence 3 Fvw Drew = + fey TJ vew yew 2 eho,(3.32) 2+ Keno + Yen0 w (wD we X sind -% cos® Multiply (0 by 25 0 5 (2) by sin @ add X* 2 cos + wind Now, multiply (D by sin 8 5 (2) by cos d subtract PY 3 Y «Z sin O-We0s 6 t cos 8 sind T*| one wcosol © 74 med only root is X= 2 cCos8+we—sinb HF Zaind - w ws6 t 1 -E CIPO aS sinhO + wero Bind (3,0) = FE O° aside eudees* " aw a apisind +ucos*o - agw cose sin? 1 w-drgew7d sar 2” ne change! Pom supp fay ¥? x “fi Lhe or dxdy=1>A Serfidx cep et 3 A=3/4 S yay yh Hence fy pr gorPp) for osxsd vyied i) For xt, rd Fry = 4iii) For oexed , yrd 3.33 contd) MU Bey = BS 'dy (*exey) dx’ a > 4 ; sh(4e+ tex] ) ° 22(3+4--F-4) 2 PL $-0+ $x] = £0x*+x) iv) For o1 ,-4¢9 <1 = Z flan’ (Pay carey) 2a (C4) + OP en]a) The Voge of “LT, Ry are subsets ef te cal line Rip Yas a real nv BA ventoneble probability space ta X is P= (2,4 Per) ohere Q Ty Ho samp spr of He Unddabyiy engerinent, foie th oe piel op evade dapied om avd PO) is a set furdden thd otiys to eurg set E (ou eat) € 4% a nul PE). A veasonelle — probobilih spa fa Yow (R,, B, BEd) where Ry, He vorge of X is te induad somph space Unda He mee pplng, Xx, B is te Borel field OF events over Rama gt) ts a seh funckow thet asstgrs to every subset B €B a nubs BO. dD Th wet te Yiysay under He mopping YL is te set (-oo,y]o R d) Th th T/o viewpoint, te mverse image is conpoled as follows: $Y sah = $2X 43 sat = < Ee ae 1X < 93) > Co 331.Gas) let AS max (1,5) b. Hence with aro second term in (i) is ero and 4p * gag the Cup b- ucy- 2] enn- [yh = zee L99 - SS al Et) =z (+b) Sas «4 4 ot 2 p72 ot “ad ze ZA + be" Zar ——— iS Now observe that =, at wos “A at 4 = 24 av = ZF hoe) Ca? a3) goat nah 97 Law = « ot! oe tt 2 oD = (d+ on ra Hence El) = € "Cl a] eX +b = of +d+b Also we observe from ZR oe add thet Ey = dee PD&D ews [ox fodx s fox f£odxl= | EQ) ” . € [Cixficoldy =f" iri fcoda since Reps = EMIX) Hence Efx)< IEQUI< EIX| oo Tt Elzu0] exits then EE ER) = [" F7.0 tarde =F gvotoodx - £90) ED a % +h 2K: = #2 =/6 b) na. of households with children 213 Xs (piven thot children as in household) é = 3 = 246 St ° xb 6 Hence eripy =f xtinrdx /PiB) PCB) = aa Lept-trl dx = erfcay-erfieny = erfiar+erfar) 2.82 -x?, -$ - EIB) = ~ey ae [xe “ax (27-2 /2 06 022 RoW aMAt3Mp Mp t2Me —- MAND N*M 2 6M + AMtN FN man Pecos Va PcBy2 2P(e) = 24 4 PLA)= SPB) 6/7 E00 = EIA) P(A) + EXIBYPCB) + ECXIC) PLC) = Sb xe an) + Flak hte Ada + 4Gb fox eax) = 6.9 years op EM={* etyie fie) de E(Y1e)=@ by inspection EW =hi"ede = 7. rs 4D a EG?) = 1? p t0°C1-P)* P EM) =p 0 = EXXI-CEP = P- PPP PR. b) For binomial rv. E(X)enp £002 BF) Pag 2 ? 228 arian why" : t hk 7z4 atin hg” Zon) gostivar O° = no-np trp O? =p np +npl- (npr = npct-P) = OPE nek€) From prob. 3.6 Elx*)=d*+K if * yt Porch) s 2 ey eet nw gt EM= 24 Gr = ae" =z a! Hence 7? = da - k= d) By direct intepration ind i “Ve Fe FW)? Fae [xe 7) dK using completion of 5 wares ¢ the fut that > ae Let Fn 24 that Eine E(X*) 24? +7? var (x)= 7? €) The mean valie is given by Fee tp tae, Net Ya tt. Obtain | Kedkaw [ree ‘de But [A etut = V/A. Hence X=kVA . The variance can be computed as gee Re oe Set Yam St. Obtain . eae freee. Gnsider Leet dt=n! for nzIl , To obtain this form , set =u in X*. Obtain 220? [we “dw = 20. Thus sad?- 0 = ¢?(a-Z) w 04s of? oSfined ROD Fyn = ff oe ROP aa dy ose ey let x=rcose,yersind , dxdy=rdrdo n= Oe PP ards =L1- € 37] aia -ip fare ge 7" ucpr F(z) = [73 $963)d3 = [ope thd, Ee EW@)= [4326 23d = 2 VarlZ) = E(@)-(E@Y* = 2-20.43. 77 a“ For any gt) Etjen) =f. Jin foudx @ Write gon) = gun 4 plan a) + 7g SE a Let gen = hex. Then if hyo xo for n>? Obtain hegy = hyo + hina th’ ewe EChew} = [Shu fade +" Kyra fendx +( hy oe fiwdx . shut Nan °° where 0>= Var (X) br Let gin hx). Then Gs gurgm ee . = Pun tlahyo Wyo + atk yy} SS” ECR) = [7 (Ryo Hahyoh'yn +aCh yoy ) oa) tindy = Ryn + Chyy h’yo + Chor} PSsfay = [2 hylan dx Lt pect el SEE ax areurp © SOC? ae? -%) Vis Gaussian random variable with distribution No,0) , se ELY]=O w Btn « fkrw fer uy ri (41%) = fr Se hepitn at L rE +92 P4Y ame ol Aare. L vane oxp [- oo I op [- (arpar } ° Tat ose 7 SOCK PD Eeviae<] =f" 9 typ cin de 2 leet onpl- aati? "(X 4 This result shows that altheugh E(Y)=0, the expected valve of “YY gum K = pK. The bert predictor Y, op Y is YF ex. Of course thir vecwlt only rmakes sere if we com observe X. eSWrite on Coys feix Cyl9) £2 @ From exanple wF-B _¢ ~pxy L 2 O-pyo™ (gl) =§ ——— 7,€ Fre (30) NCAT -+ com fe (a= varom € . recall lin ae” = Se Avo There fore let of (ar (core) oy" and. * 1 Ta’ = Zoe Using, this notation + 2. —Te (4-p”) fey (gl) = ae bee faye id= Sep Cor po a) Hence. He. face (m= = e sera eaG4l® a) pers) = PUG edd: = PCA) = POSED PS AE P(xed, Ved) * POPE es 2 pixe-ayp(yea) © oF tag # PC Ws) Hence Land Y are dependent rvs. b) Etz)= 2% P(X) s > EEPG = gCi-S tora) 2 oO EIv]= 29?) = EKPY = FUrdrorgey) eg-2-4¢ ECEY] = ZX PCB SEPP gre teat 20 ECQVECT)]= 0-3 70 + ECETI Hence X and ¥ are uncovrelated rv’s, rTHog? - [2 peo ta poodx wfxpodem af "xpardtenso Lo pended Write the aymented function apie [2 {peoapeoo +A PUFA, X PIN ASX proldlx YD oo om tnpeny = --Ay- MXM > or piaya Ke BAUM kee From oo Kf exp (-Aa= Ax?) dx © 1 From t Kf Xexp(-AsX= Asx) dx = From (22 K fx exp (-AsX-AyX) dX = UP + 7? Consider exp (-AaX-Ag20*) = op Fe) opal) let As4 ahs ut Ee . Then exp (=a X= Aax?) = K’ exp [- ge (xPPP where K’& Kexp (P/aa*) . Thus ot ‘ 5 K fers (xB ged > k's war “ie wine 2 ' at Te [0 dy oy ow pee ' Ta CHOMP yy nue a Otte p> dew L a Tae Sve Hence pdf that maximizes HOD is NYO, Pp 3 % pdf thet minimizes pdf is pouzo PopWrite pig [rare OF ax If n is odd, integrand is odd, rayien of int is even and fro. If n is even, proceed as follow: Let 80°? , then c= 07%, 1 pen wee [reat dx at or a dt 2 - Lo eB dx = (anys a differentiate n times with respect to of , obtain [ee chy oP ax ntimes bas = (am (dy) dO Tl fimes or ry a an Elo? pg” Ware Lag Pe BIS 2-3-5. n- 7 let mgan , Obtain 1 ~t(s FE 0 eo FT 2 eae omne™ where m aan is even. CS GD FR oR -(v-FI? ~ 2m mm, Tn, Tae + wes - Tm, = ni mi Wet M027 I me, But mip = ee, , Hence 7o- “* =o For ang 2 E(2*)20 unless 20. Honce With ong 2 2 (%-¥)-(F-¥) , we obtain Y= mX+(VFER) ts HL be Y- BAYpul the text: 5 “pe FT. eek x = E(% ~%X-p) te ¥I-p eer zy ~ a ES eee Ep Gay = Oy (s- iz) When p=l, Lis atineae forction of X > no error im —— &23 meon= E(t) x B= 4 ZEB nN Vosiance: N Be E(fd=4.€[2 2] f [EZR EL 2X] = [NOM nonndnger et Ver [ole ELE oe ee a G24) By Chebyshev epee PLip- B|> ale] < ont aol gee area Set clGB) 9 ewe fe “en +(e] = cin t/a) a: aa Coy deg o¢ sve) mob), » eld Affe Yl = t[eet get]. fet ef 74 |, =O dr too ef siett €Pa rit Obtain +ptL- deg [ee gt-i+dt] $-b0 wy amean if SRhO » £oH,.= a(act) alkhe eosiert way Yo de thie problen te + Pree to look up th expantion of [Q-Pe®]™ (e.g: Diserele Dielributions, by NAL. Tehnsan oS. Kota, John Wiley a Sons publisher, 1969) Theg. que [a-retf= F (NE Y(EEY eB! 2S CH A Phreays (MANE By" Therepore Bt N [g-retf"= 2 P [xed] eo Up by depenetion ! 6. &) y ep aanby at [ fee 5 AP em ede = afve oO, (will converge ig a = oe P[x24] < se Shy Far a Ainimun "ve =0, ah = = cae aos e Ft oy %Q =0 > ae DR-k& (apler some trivial algebro) Hewe pixeqd a t -" th often [g- Pe] ©) Pew+ [a-Pe'] Lye =o oe sare _ Ae 7 8 ypeRe ¥ A Nek da Now veel thet @ = Pixsal< eo [AMES q- ope Bt eyss ap fle Pp " I = [Ase Fan we ce 2 [-9ef [9] 28 Oree since GQ Ped avd Qe da. ae4.360 rey oO yer S e dx. it ar Consider the Se Ce a cenplex variable): Tw $e & ede “Pet fee ae where A= ae A*= ae by parkal groctans. Consider tl fetlouiey covtoore: Alig & A, Along GQ: Dene 2% (a evel for 20 Alor C: Zug = omy GC ee” yelid for WO flewce ry = xq e w>e f wage ON das aa yen wt f t e da= -= e wo 4 ee or oo | sor _ re wo Saree rm Lene ~exfes| —26 Ll ence i - #ke =e alla.Reeatt Ht Efe = EN yp Bix Cahy with X=1. Now Consider Ye 2x: They & Efe ~ T E[x+ ¥]- ek ety] St+ At OOo, Since the convolution of tooo reck Ponclions if a fviangle fonction of twice the width with center at the mean we obtain, S za G a gh, (ae a-pa. Equivalently, Ee = UNwS | Ea sussoa eweed Pa Frage eh} welch te reerely_ a shizt to Ga. Ro43) E [Xn Xneg] = E UC Bn 8 Bas Prog > A Zneaee] 2 El Enna) QZ Prg ~ A Par Brae t@ Fa Pegs Rel) = Rath) ~ QRy (het) - & Ryld-t) + ARy th) = (Qe) Reth)- Q Rel RH) -AReth-) R=0 h40 ; RAE? Reo Hence Pctoy> (4) 0 cateryo> Rew = -Ao* Reet = - Ao Ru(th)? 0 Aet9,33,-- wa ER)=K E(XnXn-t3* (bXn-1? Fa) Xn) 2BK EDXnXnr)= (Xn t Pn) Aner) #1 Cb Xn-a + B14 Zn} Mya) t 2BK E(XnXnew] = BK Ibi<4 Pes Ga) z-axtby W=cxtdy Erem?tym] = Efe stent bv? twicerxtdy)} BS BCX camsesr.¢ Cb dm] =Efe Fy, (thy = Bey Camtews , buyeds) est In goneral Preem =F POrk) PCY= MA) 2 Ao aenye oo 4 lt) 52 ,¢ Plz=n) +E PAW ZL Plemare a let men-& 5 thencontd. 7 . % O 2 Rie 2 Rede = rt) - < ‘ 6 a afl ey & W= 2 © ie a"=e Thes ip X: Poisson (a); Y: Poissew 5) then bee é1 oh ev & ¢ 5 O89 fi This pmb tho} Cas) Risson (a+b) 42> tL C5" n=912,.. for Q=2 and b=3. AaWe use the central limit theorem C CLT) Let KX: izt.-, 2000 denote the state of the Cth toaster FAG i 1 if teaster dented with prb. p © To if foaster ok with preb. Fed-p Jet W2 BX; , what is PI ues we 2000} Elwy= ECE) = ZE%I > P Vertw) = We - (WP = (ERS - CE eee # BFE ey - (2%) isp = nptm pnp Pp =npp Therefore EFIWJ= p= 2000 x0.08 = /oo Var [Ww] = npR= 2000 x 0.05 x0.95 = 9S 1 devo bp Xz foe» Mos ws 2 — 3G Plies WE 2000 J aar lie 2 at ! ~~ oby Tew Si, @ * ax 1 io ~erf 4.03) = OIs CYT€® E[z ra = E(w] oy a 3 EXP Va’ S Pg. * =O since E[x] =P. b) The Foiscou approx wiohon is obtained by. sething ast “Pp d ustw ~o ba OM % PC &= @ S Tl eur the Pollowing.. Leb X be th wien 4 Peceenct Tex with We Xow Yopg. Elwl-o Ver Wat and bUk«; np] © Pla ‘Stety aueaeer(at)})," Jerg_nunaste fpnorna) [rete feeb 08 1, iaeteni Oy eke, mumee tutes! 14, mumzete(pormy) Saispiayipnormay: PiEsbet ibunat Propabiisey of 1", munzote (Lifts «ket, rnumzate tel (j))et}="rnumdate (pe) Sdiselaytbeadog’)Sample run for Problem 4.39 » eltvsnorm(1024,32,20,1) ff has 1024 points; Uniform PMF of 32 points; sum of 20; no shift ans = (location of mean of sum) 310 ‘Area of pof of sum is1 Sener eeoe ras Exact Nonnal Probability of {255 dy = -|+\ =0 Thos rb ic wt necessary +e have ie = fcr to ogt E[x]=° AyGad Smee the [Xp ace itd we wvite, & (Z|- ECZX’> Paar Bh iy X NC), E[XtI=1 ond E(Ako Vor [2.= Efe ERI" Now £[Z.]- el2: 2x xf - €[2X]- Z ZEB ICM = E[2 Xi [+ oo) Using th resol ot. Problen 4.44 that ERE Z EpMle el we get (2x. “(ss 3 EXE |= 30 Hene Var A= ant ant i = 3nenw-n-“ = av AfFor nw odd, tle solutin is Cosy. Form eveny a (ttle work is Teavived. for ee Lz ora. ih 4 Elo-w es Pesyre eC = ohn ah = iar £ tS ¥ ay -0 oddk even bk) Let ve 2 => far fe Cort) al “ays (“Fety 2 92a ° 7 a wed 4 ~ FWey' - Te XENC ye), ten Godan Ere Bot gees Ze) aixom For u= ak, he a coe eS eal eb ka ao Hewce Ep {= ge os QEDise fotiowing prsblom caloslates, she Chi roweee Aeitntthe ieoaitpar wien weansand variances OF Midstartd ane Serv ava" and 3 eeapecesvety, enter t1-S0rataos Sasi Sa energize weton EinctLonfpdeit;ededlspaf3i mens quare (kL, A2,13,8, 4 feeperetion of ue culcomnee vactors pitt (do) fee GEL Eh) -(2/ (2° (2/2) | gamma (Rb/2))}*((£4deL21°( (4-21 /21) Aexptoitded 2/2); Fbtz-ocones tas 0 pata 2" 02/2982 /2) 190 Sede) U2 /2) Pampa /2) Boonen; Fpl 4 (24372) ranma 9/2099 (Ares) 109-2) /2) Pomp dole /2h¢ STinepace 0,R, 4 Wiese me plot the’ three Chi pas for 30,40yand 50 deg.of froedon Hsoretit ont pd 82 BIST pen etx paee re'ue,chack that" the’ areas’ ate Unity under each pat Sunes density for oc3D(solis), 40 {dotted S0(atars)") sevi2se) me eee} ‘compute the normalizing factors for the Normal pats i eare @tph) eigen) i feat lai) tania} SSeed/ (eee opis sagen) Sinieiall2e' ene somal azzayo pafeicoroneetist)s Patnaovoneo Ls: Patea-ovones (1) Where we compute the elenents of the Normal arrays for Let ‘pida(L}=cooft tox (0.54 (¢-mal} /atgehit) 2): Batna(|coeszronp (20,5: icmug) /atgchi2!"2)) atna(i}acoesSvexp (-0.5¢4 {Lomas} /atgenis}"2)s Bieape ee caren); itpeat of armel Foe arse tet nunztetet) (aves of noezal. for Ned0 ie! munzste(221) soteraps vs patasl Parca of noenal. for NoS0 is* ounzete(23i) CSSpltaSg Piea-bugmatkemotaignay for e130, Mzsround imug-stgeni) +cont hh uceetnataupvaigental SLacround nuboaagond3} 3 be snenye co be Saregeaced are constructed. tote Riklelpeall ig eho subtersey of ene enicaguee ea TEE AISo ote, oc the ettdeas pani ia’ the supearray MCE Reena "pat' tops" Ineegeated secncstpanstansaty, Mecoceaieuiaats}; SESescoceaes iss ua3}? Sencteeuns iinet: gcovpatna tiassial Saiocpaens (usa spdgiepast iteu12); Spatecparetiizeata)) Beassepass (iiss 3)) Bearicpacal (a1): sesraveaena (2i2"ui2) 2 weeorpaenn aisca « insapantene varia yepiiieiy: yeu Test fone signa about the babilics Prommiseraps(aylonpsat| ML Prob off" imdete (112) "k¢tynumzate 12) *) ['SoenelBesh off spnomdete (211), ELMS HP[Mi2q] AlCHAPTER 5 GD himeke 2% WG) ty) [ohne £2 Ler Madr, edn ty 2 Ke it ri rT GED We use te fect thet, hn {Xen} Q 2 7Xexp = GIXs x] [Bel -YIXee Bl Hence PIXE X= =P(X$%X, Bi pe ef F(z = BRK) PCB Condition Prob. aS CED We note that this multidimensional Gauerian is factorable so that x 2 : L 5 oF 1 > aa fin- aise - Eee With pons aro ept- 33) we write fens J.) Foe where [7% 4% 21, Ay marginal pet TER can be obtained by « FHP! fim *f, Fe hase Mon tga cos Xd AK oo ky 2 GG) [27m dk, Lapa Goh, LING Gord tyes LEY ns ard %m atwilary GGG Sta) bye, 5.54) involves integrating all fgcw a tf AY. The results are always Gaussian « a CoSGD a) let X= KH %IT , then Re KR x [Js 1 Me) { : Xn ahi Any oe KX Hence any row is obtained frm ang other row bp @ scalar multiplication Therepore there is at mest one linearly independent row. Thus thevowk is at most 4, KX fix In Xt ») The expectation operator tends 4o destroy the linear dependence ie. E(K%)) +> EX) EQX) - E(KnX) o> + E Xn). and we coamet usually write ang row as a linear combination of ang “other. Indeed if Md N> ERK) rE aes ae seen where XK” is the BY outcome of observing Xe , then prvided Non, the matrix can be full rank . twa GD We can write fine Elune - wcm- 4)) al % f= $ cuche2y- Ulm 49) 54 Then B,th0 ult) = fas 8) 8X, 4% are independent ™ Hence ElX%) = EXJEO) but ED)=0 . Thus EQX)=0 4X dX are orthagonal 5 mSGH zx = Xi Xa) Gt Xp) Xt Anke + BEX Ef xe} ns Ecole) + BF ety) SS dy. assumption & E [uXi"} ZT SD We proceed as in Prob.d.6: ° * 2 1 aor» Ee -a9” By) °& WXe Bere a7! fe MJ Lh) ECW E 0G) = BEX ae) + zz ERAN GA) 2 ® For @ 2B El wT) EG) Hence E (NE (ke-me I?) = &, E CUXe-aill?) | Ha Bee-wo Bp ay = BZ oe manga Ee eens EF any ay EG): We + BF ECM Ay Vale BE EEG @) aK oT From the Schau, inepaalty @ ELK API ELOY) LEC aa) 23-075 IK 6 b) GS = ELXs-ay 20 #-2 6) Ka = EI at] must be real # itp Kay = ELA) OGM) © ELS ay) Kt) = Ko But Kyyt3 Kyat 2 # Kis TSBAD a) det (Uy -A1)=0 det a B)eo* RerAtlo 20 ASSAD DAS Ke d,258, ow (Kye S47 leads to 2geRdro , at =)T [were ee 7 RRA WALD Ky A224, o (Ky-2T)G,20 leads to f B+EK20 gy 4:5 -~y" eta hoo gal "ea Te eat = 0" 3S A" LY PEAL" oI | a oe a = ¢ [2 alla a . [ . a check if CKeC#I [* ar “i *] =I This Yoox > ky=T since CT#C' is net unrtarg « b) aabb’ abtbe! ads albtblc van =[veave eee aa avec wee abtbe’= a’btbic if arc asc! abtliczabte'b ifarc a’zc i) AAI = ATAT=A'A since ATA’, AAT Hence i¢ A'A= AK ten [AAT= AK and the prodoct ts real cymmetnic. eetb) KAT I= a.a7ka ue a AT A co AKA CTA") hence Alo is aly “4 diy. dy. 7 <) All diagonal matrices of the same siae share same ge de C1, 0,2, i =[KX AN AB] ch ae Hence sr WWT me EK b) Consider the case when m=2,n=3 then with XE Hy 0%) %8 ha, Xa, Xa)” Xa Xi + Xan XP Thus there are three rous but only two independant Xap XT + Xa XW vectors Hence any rou con be written asa linercombination of the othar tue. Rank is 2. More grorely » Remi num =yuned = ZF ape S? FG Hence Zt and GeGr-s Gare which we are RGA GA tt Ane © ALCZ, zg) But A; is *ackeble as an upper beun and Gees? 2 ie, if uk» » Define ut A then nyied ond which has a max. when urd, Km KR x is ang real constant, OtGB ey) = CAX+B) » ATU+B =(2 +1 29(5) 232 7” 77 6 *(aTAT Aa = AT a ‘S$ Q -1y2 . rata sear ale Z a)fa)=>. rT SIO We find finn the expression Hat Greipel ~2ped le find. fron the opr eh Wie * KEK GEE OK aly IF] — um idly Now diagonal . ro Mee obtein eh awe F(') FL] Cuncrinalyed ) Hevee ve aty =() 1G) Thus the transformation At [i A] Hex t only solution to YoXitX ieee a ne Xia% is XM Ft) METH) fin = $y EA, hy 2 aie ep SARs FC CES) = aye PIL e+ FH i ¥, y 7 TTR) oT Pal ATI ye TOD = ate emp (9-2/7)GD [Zp x]- Z pew FZ pp OH ELE nOton ° 22 PePi (Xe- Be WR, “i= 22 wake, ei qo Cnvederite pouction of X- ic : et f.@)= Lxts z & we odo Efe ee wheve vS w/t wid cw. Hee Be) = Cbg enter Ae em or by table of inlegals), thes y = [B, eis & neue! 10, fer Se ho}t+ g* Let ca be poronebers of bivenial. ae Bye)= Zcelend (h %) e* art = (Peep qr gy” £9 =[B, wy] = (expr aE [pxeqorg]™ Meese oye (HX) pA gE deo Thos mean is rp. Vorionea Is nrpg.£0 oD G20) e@= e Ke [Kihice Kis Covent ae mashex , QL = (ty, a, 0, 002)” ne Jee tay Fe 4 4% +04 %4) (a): JUL € £, Orme) 2 dtd, Udy, - ‘i Es ep lane FPO) Tle problen corti Cf We recogeiye Hot OKA= 22 Kee een G Tes oe -@ 2 Kuen ak W,=0 H 2. eae -8[ Ke 22 Kaka ori] 90, Quy day fet 1 5 W)2Wr= PE. 3, 9u, 20h, +B[2 Kl Kika+ Ka z Ky wg + Ky 2 kas Bi - 2 2 Kg Ky re -z Keg Oe Fae <— ya PY anaes | = Ka Kea Ka Kia + Kar Kia Pa a=»Gay le first ovder pole ie conpoted from fy, tf forrrtnte (xr) 0 $ KSI a °, else Ba. Sypris A = fe @= £,® The second = order pap, ts computed Prov $y ve [tenis ax, Ba Sapa begs a Co 4)= Fang) Lifeute X= Xx. * aft xe ed take W/36 Also X= X= X= 3g (conden Bt Vow X= Vox X,= Var X% = Le # [& Gerd dx -(&Y= 008 Kn = Ksskuek 4) ¢ = - 0.003. e® f& Hewee 1 -0.04 ~0.04 Ks 0.08 |-oot | 0.0% & -0.0+ -0.04 =| Sime Pas P= Ps %Oy Xi%ey% ave almost uncer aGz) Th The first step is to find ergenvelen envedors. tfeuco (K-AE) ore > det CK-AL)<0 det ( K-2E) = [* n\ Lae .9S=0 LS ad Hence d= 35 *h, = 05 For d= 35 Com pole d orth ud! (za feats ra (', “iy Fe Qy=0s (KREIS bays (ui) 7 - iL r “Ae [e } Thus @ fea [ lie os| Hence = X will produce YYF0 To ay let Be oe The 22"= A" YytA* ake / x SA NAS ra dasely patrix) =a Hence Z-A 8 £ & \s cola lemimg Prasat? orana Box23) oxtd) Smee tt ts easy bo generate sampler row. NCO, Ha Simp st Wo to obtain a seater diva. grou for Correlated vordon vantabler is te following : le YSaX+bK, xX: Ned YF eX+ aX, x NCo) Var (%)= (aX bX} = a+ b'=2 Var (%) =(CX\4 AX = erd?=2 Cwl(y Je Ks octbd=-15 Since. there ove 3 cautions & 4 unkinang Ae e=0 then d= +2? b- #35 ast - Signs of ad avbbitrary, Thos “JE x “/B'X, Ye-VEX, or Y= 0.9854 Xt 1067 xX \ Yi = what & AbG22) €owtd) & when calling normalscatter(n} follow by semicolumn to avoid Ysceing all the numbers on the screen i.e.,normalscatter (n); function [y,z, yc, 2c] =normalscatter (n)% function Definition yerandn(1,nj;Sgenerates Normal rv's from N(0,i)in a lrow, ncolumn array zerandn(1yn}; yeszeros(1,n};$initializes ye vector zx=zeros(1)n);$initializes 2c vector for kel: yo(K)=0. 93544y (Kk) +1-067%2 (Kk): zo(k)e-1.414*2(K) 7 end tthis loop generates two vectors of correlated rv's rax_value = max (max (abs (y} ),max (abs (2) ) ,max (abs (yc) } ,max (abs (zc) }]) ‘the above instruction picks the largest magnitude entry from all 4 vectors 4 in order to set an identical set of axes for two figures. 4 max_value will be a number like 4. figure(1)% will allow Fig.1 to coesist with Fig.2 plot (y,2,'0")8 the litlle “o" means that unconnected dote will appear Bin the figures title({'scatter data for two uncorrelated Normal random variables*}) xlabel (‘realizations of X1')% gives a title to the x-axis, ylabel (‘realizations of X2')8 gives a title to the y-axes axis('square')¥instructs graphing routine that x and y axis will Scover the same displacements axis({-max_value max_value -max_value max_value]} ‘the horizontal and vertical axes will go from Scmax_value to tmax_value on both figures Figure (2) plot (yc, 2¢, to") title((‘scatter data for two correlated Normal random variables with correlation coefficient of -0.75']) xlabel (‘realizations of x1") . ylabel (‘realizations of x2") axis("square') axis((-max_value max_value ~nax_value max_value))E23) Let Ae Ca) Fag. Rai. Ther He onty solution fo the set o ot. ° cfious " ft a [*\-e aco] LY x By tay (oe) Fans £4. (s.a-9) Le ON OD fy Gugge Teo wheve ate [T) = ma a | Be. Bae = \e" sal ~ael Buh wo B= A' dt (B]= op Al Thos by C4: 24) = (det{2}| $y Or “y Xe) PH(6.24) This problers wa specal Case of Prob. 5-23 with ay Castly Computable inverse. We canneh Solue He ovtginal system using Ye Aired omethod unless we use ouxiltary. Vanrobles N= Xe te X : Ye SG pose eee You Xeat « ouniot tc Nye Xn Now we eel Y equations In iw ur frmowns. 4s (quays Ax where Vber ut As Jollet ~ o0---0\ ig easdy. computed fromthe nm indivi- 4 Sidhe wus * = Aw Naa any xpos Bede Ee tao 4s XF= Yocoud 1 -\ 00... A’= oO 1-10-90 aw Oo 171-9 Oo a \ CReonk dat [A= ad[ A" | =! Hence fy (au “ytfa) = he Cyt gee ete 7a) TS ah Ray Corts) uttegele oot cath respec to Margen “24s © AA“ . has squaved — Te show theb otk error consistency. , ta anus EA pit. Thue (fam p= fae fates 4 BERK BM E (above) yields: El EX ee BEM DH = fe [oe ptns dened toa 2 = £ ry Hence fin has te pee thet A =O Aine E [ipo ph [= 28 nerFor reolizobons Xe, Xe kayo tha Lkelihood pouchone becomes = a -r\2% LQ)= NO acagucn) += aCe) Ae a Ey Bae = NTA ZK) so Tle ‘ao solubn is rejected becouse it gies a maininmunen, Thos for X,, X, x LL ZX; teGB) 1, Writ PCR; p= (Qetep* Thea Jp pO (a) Apt icp" — (A) Powe or 0= Ro-p)-p (vA) 4 Dipfecentiation doesnt work here Since the rmaxmuum of te deli hood fonction does uct occur where “the cevivotite is Zev. For 1 ordered realizations Mary Many = on Sch that xX) > Moro 2 Mom we wrk have OS Xgy ) D2 Ma: lo mex. L Ca, = (b-a)", we Ain (b-a)" sobgect to above vestrictions. Hence ov noain (X,y Xa) ond b*= max CXyey %). Let Ze we Cy) Plas Feud Fb al R@= + Feeds + t (i- R@] where R,@= 0 a Pel=i- 2eIPea Py 2) ent a LED iH By the formule EG] = ELeEcxyl] = Elem] =Efeleim =m] | EL ee) M =m] = eiwndote => choracteris tres function, Bw Blo = ELEM] = [ einm be" DEM] dg mebont ye, Bam - “te Femi pejin = ee > - »~CR — ze) et ob ELX]=ELELXI My) = ETM] = Ex] = E[M] ( the meau of X ts equal to the mean fe)oy = E[X'] - Ag : =E[S*+ M] - Ax 2 Ags + ED) - Ay 2 = Asst Oy a SD a Define dummy Yurdem variable ext Y = S-SR by the formula: Fes (1 = fing TOIT! fay ognsl jorseo ITl=13r Sl =| S rls HX and Y are independéwt random yariables = fag (1 = $00 fy =X ENEMY =Ave = fas(noernser = fala [PxseNas H for oere] fi ogest = fe Ye Otherwise ob PIXexJA Y=] Pirex X24) _ PIAJ —. Fe A Yor= LIXsxIA, Y7] aia yey SRS KER Or) loo Otherwiseoa Ce) Let XS minimum meanrSguare error A estimate ef X > XP ELXIA,L+f =f xh OAL Vdx % oh, Xe* wh - X=f x Ap aX t ys Te [peed x et oH) A nem ore oO Gx) Mee S(Xeyt Xan) +We CF2 49 Xe bX2t BW) Xe EXq tro X= = (Lar wits) +We ek Xo + bw + e%t We % ye ZW, t wr * BX X2= ew, FE Wet EX Ms Me EW HEME a Xy HE ws xy = 5 WG +E Wy 4 FW t a +t Tory W t2EXoaes Xq = Swit Fg + Tn closed form Drea tt Xe lo 2 = °eZ t + [t=k) ¢ kK. Ww 26e Wh A matix fe of Ne doom given sa Ne “hut 2 Hee = at th both. Wo ja Fem of X, wi Ys “te O We & \ ce L = i Wa , who oO l= = e=4 ro bol fon of Ty Pores oF _ Writ) (Ky - EDX) “Belge hy, Ze] OEE ang oSne [ee] ewl 9 fue)=It fueled ty wed ae independent «(Yep ( (Sz! “') _ Lell = (suid, ju, 7 [yiG9) To Prove Covrelary b These G.I-1, note That The sequence ofevedts is decreasing here , ve B, 2B. DB,... or 24 Live letly BS CBS cBe... Lf we duicth apply There G-I- y Te raereasing Sequence of events BS pee gut Bs = Op, ny o Bat RB (sie Bae (Tas)*) so lic PLB.) = PLBol with Bey 4 OB, ty This decreasing Seavece of everts. I mo yc set of outcomes Noe: UBS = (OB) = sof va every Bn cs e~ ~~ Set of o-teames ® sobs ostepeas set of outcomes in any Be wn every Bn tm» any BS - 6-10 Sa)Sm) = eo) a a fe ge & Poon TeSm = Kit Kato tXm Su - Sin = Xmas t Kmre to t Xn Xba independent youdem sequence => 3% S,-Sare Independent rondem variables Bs wf fg Sm) Se and Si- Sin Are Sum of aussian Yandom variables, therefe re, Saad oo are also ened her variables ELS ]-0 YAR S.J = E {S.-SnJ= 0 =n-m VAR (Su- i n- (Ses ee os) 201-m) = (Sy, Sind = ~ fia =. Stmmyb lémen Ww SIO No, they need not be Continuous frem the Jelt Fer extmple: Consider a discrete random variable X wth the Lollowing prebabil.ty Pix= ae ; Me 7 Let FO | be the POF cf X, then Fe (X=53= =, but for 6>0 we haveFe (X=S-e)=f the ect Therefore, Lim Fx, (X=S-€)% Fe (XS) From the examaple, we see that PDFS 1884 not be Continous from the left. eS eevee F G@® @ Mylo) *E [epi] = 2 PLS] XL 3,7 -t (3804 supe ge orm) CD Re Lenn) 2 Eb x*E) . =z PRAY XP 8) X Ly] = (aseasereaedee-d 2 ord, orm) a a! © Xb=) | fg wy Oe & value prob outcome 4 git xty-SO > [4 “b oO at %o" $0 ST xpo}-8, X0a-0]> P[Sal\- 3 PIxt-3]-+ ad Piemeol- $ Gad) So qomt pmt doer not Rube, ce. Xlo4XL} A Mott jadepondeat. ;oD a) My l= EVYGIM=EfZ wax crit 2D bey Cmd) 2 bux bi rh Ag i] ») RyyBed=Ef Yer vba) = 20M BLD Rage Park, tet] or (Rextovma+ Ronleiind + Rexboscd + Rye Lib n-d co) Keon Ef Yoa-ae by) [re apa = Reyy brn) - My Fa My Tn) = Ky bul+ Kal] _ \ 4 Kyg lant) + Kb) / 4) sa ys |[2&3 -(% ~ [Fea “= ») Th yx N( Hy, Ee)Wheee dr = (0 q (K Ys My ba/ . Ry fy, a {Ws aE," ep -#{tr-») Kr 0-4) | 20 6 Yn Vy < 400 G4) de ooh fa jos t pdt £ (tut 500, 5). Wow The] = 2 TC) = TEI+ SoH ee 2X By defintin, Then Tis] +X 20 rrileperen, smce TIS) ic asm of eacher TIRAS not rachel ya [1a sum Tat isX. Thos Bhi cee vo atnes $(, ty, /45) = Ft (te Iti) (ty =f lb- 55) f, (435) (Atte were") we, aed Ae re 7 atteenatwe)s Do a vector Tons tra ati. From TLL TE] (Z !) ( TDA via mate 7 | °) fu te _ zo. = © ekewhere “1 Witt detecmmant T= I.a3) C615) © xDm $12 SA (8) 0b -e] wher P[rq= ‘ n=o DeDd= Efe] ple) — Te “L b) Efe be) -A ED) = 5S peal phi] & [a4] 25" py Efae} +e Pll, a “Gs een) ra tite Lirylil leet a e+) TET DT Ts phdgh dd ye so O72 3Ne 20 Try and xi YM (,3%) c) Site TH Xs will be covelated, eee sper Te @rrelatin Coeff ice et Se! toComplete ow exprestitn fy Te jot pdt ove may E [Ke Xmes] =E] > fh Pe) 2 AS pli -51 | => D2 EAA) Pod pin] c J SM" ip \, 2h Lenya 2 Qn) (greytd [Aste ats Fte7 +ehetar $4] ~ dd, V4 3X 2 dy? = -% + r + ce at — 2 p= Cov Xn, Xner | EL Xe] - AX 2 MT eg FE ins Org Oxraery 7 So TM gout pdt becmes (xx) = _l er - XL, XEnt 1) To V1-Pe a 5 (at) | Oxe-a)' oe. te Whee Mars otk, ad P= %Alternate simple solani > GS) (GiS)osi) Solchos veri Chasachenthic Prati (CFs) 3G, te fe") . Efe Gh Bi ret ths) = 8, () Gn). Mow Fp (wv) = Efe" ] um Ax NOAA), so = exp (iva -hurd) = By (')= exe [12 ~via) 4 Fy (E)zenp GUAT TY a) ws By (v= exp (4A 7 Z uw gd) ad XH NOVA).bi) fy () -fh (x2 x) fe Gdn ) fig Ge) = {ae Oa tas) Lo (eos) diet Xa ade Cy Oy bing) 4 o guise fy (%a) = a oF Ue) (Ets) (@-)} pe fee (mn) = LNT Mm (a=)! e 4 [xa) eg at Pettis 4 ue CO bh etn fag (v0) = KE , aa diay) 4Os) DO ace net oo! e wo U (Xn) = Xe | eX apr © «e) “Gesell._ Xfn1 — Ytny EAD & 0 —_—-[hteal ———+0 Let the system function be represented by an operator & or YinI =2£ {xin3} from definition: hon nso LSM ISOI=[ 5 nav The input sequence, X01, Can be reprented by Xtmj= DX CK15 (9-Kp Then, yond = of tmp nck {2 xe S tr Kf = Z ACKIoL {STH > Z Xb kd The order of o and 2 are interchangeable, 'f o€ 1s & Continuous, operator , we. hind is Absolute summabler 2 ihWll SM <2. Therefore : te yes 2 xtdhin-«l = KOI * hoot wb ALAt _atee and 7 wn ata= 35 | Amie "dw a1 Loa )te -jum jie esa lh (a ame en "dw sa LT 7 te jw n-wr = a” Athidw +f Sate" dul If" NOt § wena > atmie? dw wan 1 MEF Bt +Uf the infinite Sum Converges in: form ly we have 1-3 T gtmyeTr du = 0 men TT Hence : atn3 = 39 Cr QAtwidw = at wo Ytri= z. Uxtrah (4-k9 Yres= 3 x [e, xUKI hon afer e * Lf the infinrte. Sum (14 square bracket J Cenverges wv un: formly » we have + Ytwi= 3 XEK) le. h oa é = g xteae” “| Hes -jun = wa [z. xe” OR: Yu = Hea « GID a Me Z transforms zit eZ YCZ1=XEz1 ee I eal 7 => ——t ts Yczi= GRcrS DU-Be) ~ (ray (2-8) PF YO Im sah YELLE a costo vshegal” “YOM needs Causalcty, and ie Peseta = ee aS vel wt! re = yon pe oa * aaa (2 let 2) = Ytn1= za (eto no 0 neo b> By part Ca K See Appi Ad (discitke Tne)rc = Xf) Yczis Trae" XUZ1> ar = yin ag EFI Pde , nse , *>P a) ap" ™ YO zy f (pra up-p> dp veeqe Ooh teat VOI Tee yap NEO = yore! 2. yo? ne yor [ Ao ~(-0)" < 1th ) ie Xind=Xcw-1+Winy, we cnc pe GIFS Ay from the equation abeve, we Can see that Xt os & J inear Combination © wen-415, & 70. However, Since Wn) are independent, therefore Xor-13 and Woe ave independent bp Ewe Efe | “XGh = /xXt-11+ Woe) Bxtna be = Ley iam \7w.o) but Xtn-13 and wind twe independent therefere: By) = By (9) E,W) see 'XLo] is stationary sO Sime wen iS gaussian with Zere mean, we hve: Dy, iw) =e 4 but, we Knows Bx) = Bel) Beyl) let w/w; Then EY = By VY) BV) and By Fu) ~ B20) BW By (ew) = Bb") Bw)el all the terms On hoth s: ides: ; EW amul teply Ex we Bm) (B.hor J Beh) > Bx Vow [e# +7 agence ewe] = B ™ lex POE, 2] Now, let 1 — ei then, 70, if AIK 1, then Ml By) —= Fe) =] Ww) 2 p= ae aes therefore : - t 3.0) 2,6 #4 ds from (ed, ! a SO E> Ytra= ht * (wont x cra] Uy tnt pov [ yea + 3] ry >? EL uytn Kats J 5 Kt at3l= IAP). ob, EL Ytwal = IMAI. UR, hEKICN Ch - 3037] MEF * hina Fas heck 3)” fe = * ptas Cd, ey = 9a!“ ytas/ tines ©) Ky Om na Bao hexrpcigs Cn-m-K+LI > Ky Cm-nlewes.s Se, Kyy Cm {% peer hela Som -K+ Cay : (2, ntca htk-w I Oy,pk pe mot (= e*) CU 7 Remax (gm) sSbrax(on) 2. max (Om) omg "Fe Cow = Ov P aD) so Kylma] = Ow PO '"YO-F*) CD ye cin shou E[ Xf) Mri] FO. mt ob f& ror. As Oe el 24 wh) Xml) = be nti-€ wh] so E[xDAXbel = F Zi ee & [ue Wil] n 2 et ze 20-1 -23 at Sa jro -« (4S) = ol: Ver (X) #0.622) xfol=0 , XIm] aie J4+WR) fe nai Od. 2) From (I), XP] = = e" “Wh , ne so E(Xt)] = else” “wrl) = ze "bee ze" 0 =0. b> Ky bon] 2 ELK) (KEI 3] = E[KI>I XI] sinct mere 1b 200. =E( Ze wan &, e“wri)/ Let m =E[ (Acoswlors)1 Bsinulomm)) (Acos un +8 cine) ] z : =O (Gsubwn) aren + Sia Worm) 51% wn) = O ws (w Gorm) -wn) ~ Orcos um = Rxlrl. wss. b) Consider E x0 2E[ (AeosenrBsnwr)| 5 E[A?] cs 70n + EAB] CD FE[As] (> +E LB)- situa Now EAB] = EUs ET) = oO =0 =E/AB) 3o E [#01] = 3s (ws?un + sitar) t F cmsht Ariel -order monet (#0 ) Hence X La] Cannot be 5 Atbmarg wm The Smit Sense.