Worked examples — Multiple Random Variables Example 1 Let X’ and Y be random variables that take on values from the set {—1,0,1}. (a) Find a joint probability mass assignment for which X and Y are independent, and confirm that X? and Y? are then also independent. (b) Find a joint pmf assignment for which X and Y are not independent, but for which X? and Y® are independent Solution (a) We assign a joint probability mass function for X and Y as shown in the table below. The values are designed to observe the relations: Pxy (rg, 4;) = Px (#e)Py (yj) for all k,j. Hence, the independenee property of X and Y is enforced in the assignment. Pxy (nus) a2=0| #3=1] Py(yy) T T wat z ; , t t t w=0 7 3 5 nol t t t t ne 46 1p i G Px (xe) t 3 3 3iven the above assignment for X and Y, the corresponding joint probability mass function for the pair X? and Y? is seen to be Note that Px2,y2(Fe.ij) = Pxa(#x)Py2(Vj) for all k and j, so X? and Y? are also independent. (b) Suppose we take the same joint pmif assignment for X? and Y? as in the second table, but modify the joint pmf for X and ¥ as shown in the following table. Pry(eeys)| ti ==1 | 2 =0| ta=1| Pry) yl I T T T — I rT It 4 5 + t t 1 9 3 T T T v i 6 Pra we t xe) | a5 5‘This new joint pmf assignment for X and ¥ can be seen to give rise to the same joint pmf assignment for X® and Y? in the cond table. However, in this new assignment, we observe that 1 7 $= Pav ors) # Pees) Pr (un) = - 1 182” 36 and the inequality of values can be observed also for Pxy (1. ys). Pxy (#3.y1) and Pxy (rs, ys), ete. Hence, X and Y are not independent. Remark 1. Since —1 and 1 are the two positive square roots of 1, we have Px(1) + Px(-1) = Pxa(l) and Py(1) + Py(-1) = Pya(), therefore Px2(W)Py2( [Px() + Px(-DLPy() + Py(—D) = Px (1)Py() + Px(-DPy()) + Px) Py(-1) + Px(-DPy(-D- On the other hand, Pxay2(1,1) = Pxy(1.)) + Pxy(-11) + Pxy(1,-1) + Pry (-1,-1). Given that X? and ¥? are independent, we have Pysy2(1,1) = Pxa(1) Pya(l), that is, Pxy (1,1) + Pxy(—1,1) + Pxy (1,1) + Pxy(-1,-1) = Px(Q)Py (1) + Px(-I)Py (0) + Px (Py (1) + Px(-DPy(-D). However, there is no guarantee that Pyy (1,1) = Px(1)Py(1),Pxy(1,-1) = Px(1) Py(-1), ete., though their sums are equal 2. Suppose X* and Y® are considered instead of X? and Y?. Can we construct a pmf assignment where X* and Y* are independent but X and Y are not? 3. If the set of values assumed by X and Y is {0,1,2} instead of {—1,0,1}, can we construct a pmf assignment for which X? and Y? are independent but X and Y are not? Example 2 Suppose the random variables X and Y have the joint density function defined by How = {2 ty) 2<2<6, Osoc= eS 10 (b) ‘The marginal cdf for X and Y are given by Fx(a) =P <2)= ff sea) dude 0 2 r<2 * Plenty _ Qn? +5m-18 _ [fl a9 lr=——ap 2S <6. ° Party dyd: [ f zo oy Fy) =P 6 0 Sporty y+ ley D dyd, = OY Qcycs = Ll 2io tos (OS SS, Sf ort y dyda = 1 y25 Lf 210 tos (c) Marginal edf for X: fx (x) = Lry(ay = (48 2< 7 <6 dx. 0 other si = fy =f Mt <9 <5 Marginal cdf for Y: fy(y) = 7 Fv(v) = {i o3Y >2)= 6 ps 3 0 [ (20+ y) dyde = 5 1 pe 23 rooaase [ [erro yd = / tots > pacrveda=aie [forty andy —(c) Joint distribution function Suppose (2, y) is located in {(2,y) : 2 > 6,0 3l5 Example 5 Let X and Y be independent exponential random variables with parameter o and f, respectively. Consider the square with corners (0,0), (0,4), (a,) and (a,0), that is, the length of each side is a. eal ° oa Bor ~ (a) Find the value of a for which the probability that (X,Y) falls inside a square of side ais 1/2. (b) Find the conditional pdf of (X,Y) given that X >a and ¥ > a. Solution (a) The density function of X and Y are given by : oy see= fer" 220 pq {hem vee Since X and Y are independent, so fxy(«,y) = fx (xx) fy (y). Next, we compute P< X a,¥ >a) = PIX <2,Y aY >a] Plax X<#,a aj _ Plas X al P[Y > a] since X and Y are independent [EEE Bee ™ dndy a2)(e-aP — ¢- Ay) suy> ate te OPO OY 8 (gay >a =) JS asemere 8 dady © eee a8 ’ 0 otherwise fav(niylX > a,¥ > 0)=—2 rey (eylX > a¥ >a) xy (2, = Daag EY HX > a = fafe-are-Au e-em“ for > ay >a, 0 otherwise Example 6 A point is chosen uniformly at random from the triangle that is formed by joining the three points (0,0), (0,1) and (1,0) (units measured in centimetre). Let X and Y be the co-ordinates of a randomly chosen point (i) What is the joint density of X and ¥? (ii) Calenlate the expected value of X and Y, i chosen point expected co-ordinates of a randomly Find the correlation between X and Y. Would the correlation change if the units are measured in inches? Solution () Sxy (ey) = 1 Arca A 0 1 Gi) fxm -f fx (x,y') av =f 2 dy =%1— 2). 2, («,y) lies in the triangle. = Ly fr@=f fxy(eu) w= [ 2 de =2(1—y) [ (1-2) de [5 “-3 2 Hence, E[X] = 6(iii) To find the correlation between X and Y, we consider 1 ploy [fen eety ln Io ‘ [ u(d—2y +92) dy 0 [F-er+4] =z [2a 2 3 0 COV(X,Y) = BIXY] ~ EIXEY a2 (V4 ~ 12 \3) ~ 36" EIXY] = by | dy ¢ 1 a! 2 1 wl ey] it I ( a}, 6 x? 1 VAR(X) = B[X?] — [2| Be i 1 Similarly, we obtain VAR(Y) = 75. COV(X,Y) & ORY ee /VAR(X) /VARY) ik COV(aX,bY) _ abCOV(X,Y) _ a(aX)a(bY) —aa(Xyba(¥) ples a and 6, ‘Therefore, the correlation would not change if the units are measured in inche Since p(aX,bY) = (X.Y), for any scalar multi« Example 7 Let X,Y, Z be independent. and uniformly distributed over (0,1). Compute PIX > YZ} Solution Since X,Y, Z are independent, we have Ixy2lt, 2) = Ix(o)fv(wfz2)=1, 05" <10 yz) I i / fxyi2(ss2) drdydz 2 LEL digas = f° [°0 ve) dydlz -[(-) az= 3, Example 8 The joint density of X and Y is given by S(e,y) {«c Cw V 0, Fxyy(a) =P [# < | Y co pay -/f eo &) drdy =f [ eo FY drdy 2/ysa oe canta (ony = [arene al a ° By differentiating Fyyy(a) with respect to a, the density function X/¥ is given by fxpy(a) =1/(a+1)?,0 4; now -1< y < Lis ne or satisfied so that fy a) = aso, —1 < z/y <1 is automatically satisfied so that 1 1 li 1 nd Ly= o dy=—* 1 i I, 6 f, by zal) 6 1 f. (iii) 1 < |z] <4; note that tuv(2 ») = g only for cy < v (ii) |2| <1; in this Ja(z) so that 1 al dy —=Inly = —[In4 — In|2| y= 4 ni glint tne 4 Ind = if|z|<1 In summary, fz(2) = 4 1) . : ¥: Sale) sllnd In|el] fl <|z| <4 oO otherwise oo a Remark Check that [ fale) de = / lind - info) dz o 4 ‘h [> 7 hee In|2{] de 4 -f, aha of melExample 10 Let X and Y be two independent Gaussian random variables with zero mean and unit varianee. Find the pdf of Z = |X — Y| Solution We try to find Fz(2) = P[Z < 2]. Note that z > 0 sinee Z is a non-negative random variable. Consider the two separate cases: « > y and 2 < y. When X = Y, Z is identically zero. (@) r>y,Z<2er-y0; that is, 2—2 x a“ aE, (e?He-2)"172 gy Lay [P14 wan Le 10 8) de = ee M4 dx dl .ney Z<2ey-2<2,2>0; thatisr Ty]: Pip > Ta and together with P[P, > Ty] + P[Ps > Ta] Ab dat As 1, we obtain P{Ls > Te] and P[Tp > Ts] Justification:- Since 7, and Ty are independent exponential random variables, their joint density fry,r, (ta, tu) is given by fry .rg(ta,ts) dtadtys = Pita Ts] = [ [ Aadpe Me AB? ipl bdo = [Ngee A8"4) dg o =| Nae ala as Age Ourmea dey 0 lb —, MM Ap Ma +A A+ AB ihate tot then imegrate wrt fom 0 XO 12