Escolar Documentos
Profissional Documentos
Cultura Documentos
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 AFTha 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 AGay 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, atGS = 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 neueWe 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 RouGD 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 wi 0, =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 -bwacy 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 $a fw ra Let FONEECP hope te-7. The reots of gKN-vee hinyy-weo are Kw ne ew alow a ew RW = ow he Ww te The magnitude of J is mm 2 [+a is +h tall four Nag al 200 agement = avo Now observe: x°ryPz av every where Kaw =%GY, Rye VE = ayHence GD ot fat? ESTER ew PB’ 22 eae = ‘I -( 22S steer +H * TERE € HF cash 4). eS Introduce K#X-¥ , BA X*¥. Then Fw tw =o [da fap fy ($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 x t, rd Fry = 4iii) 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 o 1 ,-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 aMAt3Mp 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? oSfined 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} PSsfay = [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. eSWrite 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 eaG4l® 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, rTHog? - [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 PopWrite 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- BAYpul 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 clGB) 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) alkhe 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 aanby 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. ae4.360 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. Ro43) 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 thencontd. 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. AaWe 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]=° AyGad 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 AfFor 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 QEDise 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] AlCHAPTER 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 CoSGD 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 mSGH 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 TSBAD 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. eetb) 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 linercombination 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, OtGB 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 $y ve [tenis ax, Ba Sapa begs a Co 4)= Fang) Lifeute X= Xx. * aft xe ed take W/36 Also X= X= X= 3g (conden Bt Vow X= Vox X,= Var X% = Le # [& Gerd dx -(&Y= 008 Kn = Ksskuek 4) ¢ = - 0.003. e® f& Hewee 1 -0.04 ~0.04 Ks 0.08 |-oot | 0.0% & -0.0+ -0.04 =| Sime Pas P= Ps %Oy Xi%ey% ave almost uncer aGz) Th The first step is to find ergenvelen envedors. tfeuco (K-AE) ore > det CK-AL)<0 det ( K-2E) = [* n\ Lae .9S=0 LS ad Hence d= 35 *h, = 05 For d= 35 Com pole d orth ud! (za feats ra (', “iy Fe Qy=0s (KREIS bays (ui) 7 - iL r “Ae [e } Thus @ fea [ lie os| Hence = X will produce YYF0 To ay let Be oe The 22"= A" YytA* ake / x SA NAS ra dasely patrix) =a Hence Z-A 8 £ & \s cola lemimg Prasat? orana Box23) oxtd) Smee tt ts easy bo generate sampler row. NCO, Ha Simp st Wo to obtain a seater diva. grou for Correlated vordon vantabler is te following : le YSaX+bK, xX: Ned YF eX+ aX, x NCo) Var (%)= (aX bX} = a+ b'=2 Var (%) =(CX\4 AX = erd?=2 Cwl(y Je Ks octbd=-15 Since. there ove 3 cautions & 4 unkinang Ae e=0 then d= +2? b- #35 ast - Signs of ad avbbitrary, Thos “JE x “/B'X, Ye-VEX, or Y= 0.9854 Xt 1067 xX \ Yi = what & AbG22) €owtd) & when calling normalscatter(n} follow by semicolumn to avoid Ysceing all the numbers on the screen i.e.,normalscatter (n); function [y,z, yc, 2c] =normalscatter (n)% function Definition yerandn(1,nj;Sgenerates Normal rv's from N(0,i)in a lrow, ncolumn array zerandn(1yn}; yeszeros(1,n};$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= Yocoud 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 © AA“ . 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 nerFor 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; teGB) 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 Otherwiseoa 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 t2EXoaes 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 [yiG9) 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 TeSm = 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 haveFe (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! toComplete 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.