6 If X is symmetric about 0, then X 4 —X and hence for any x € RB, Fy(x) = F_x(x) and hence Fx(z) = P(X -2) fix(-2), Ve €R. Assume that the moments in question exist. Then EX" = Fx(—2). This implies f(x) = 2.2" fx(x)dex = 0, since the function ++ 2" f(z) is odd. o Remark 1.226. Let X be a continuous RV, which is symmetric about p € R. As argued in the above proposition, we have for all x € R Fy(u+ 2) = P(X Sp ta) =1-Fy(u—2) and hence fx(j+2) = fx(u—z), Ve. Conversely, given a continuous RV X such that fx(u+2) = fix (u— 2), Ve for some € R, we have Fy_,, = K,-x and hence X is symmetric about p. We now look at some special examples of discrete RVs. Example 1.227 (Degenerate RV). We have already mentioned this example earlier in Exam- ple 1.179, Fix ¢ € R. Say that X is degenerate at c if its distribution is given by the p.m.f 1, if Ix (z) = P(X = 2) = 0, otherwise. © ‘This is a discrete RV with support Sx = {c}. As computed earlier, EX = c. We also have EX" =c",Vn > Land My(t) = e'*, Vt €R. Note that Var(X) = EX? — (EX)? = 0. Remark 1.228 (Bernoulli Trial). Suppose that a random experiment has exactly two outcomes, identified as a ‘success’ and a ‘failure’. For example, while tossing a coin, we may think of obtaining ahead as a success and a tail as « failure. Here, the sample space is 2 = {Success, Failure}. A sin- ale trial of such an experiment is referred to as a Bernoulli trial. In this case, Probability({Success}) = 1—Probability({Failure}). If we define an RV X : 0+ R by X(Suecess) = 1 and X (Failure) =66 0, then X is a discrete RV with p.m-f. 1 ~ Probability({Success}), if x = 0, Sxl) = P(X = £) = } Probability({Success}), if x 1, 0, otherwise. If Probability({Success}) = 0 or Probability({Failure}) = 0, then X is degenerate at 0 or 1, respectively, The ease when Probability({Suecess}) € (0, 1) is therefore of interest. Example 1.229 (Bemoulli(p) RV). Let p ¢ (0,1). An RV X is said to follow Bernoulli(p) distribution or equivalently, X is a Benoulli(p) RV if its distribution is given by the p.m. 1p, ife fx) = 9p, ife=1, 0, otherwise. In relation with the Bernoulli trial described above, p may be treated as the probability of success. Here, EX = p, EX? = p, Var(X) = EX? (EX)? = p—p? = p(1—p), Mx(#) = 1—p+pe',teR By standard arguments, we can establish the existence of these moments. Notation 1.230. We may write X ~ Bernoulli(p) to mean that X is a Bernoulli(p) RV. Similar notations shall be used for other RVs and their distributions. Example 1.231 (Binomial(n,p) RV). Pix a positive integer n and let p € (0,1). By the Binomial theorem, we have Ls peas =3(t)ta-24 & and hence the function f : — [0,1] given by (D)p"Q py, ifr © {0,1 n}, 0, otherwise. f(@)=or isapm.f., An RV X issaid to follow Binomial(n, p) distribution or equivalently, X isa Binomial(n, p) RV if its distribution is given by the above pm... Here " nl n fn-1 BX Ste ett yt (io eM =p) = np p+ (= a)l"! = np, and EX(X-1) = SMELT 4 Any nnadp? $3 (Ba 2-2) = mln ye ‘Then EX? = BX(X —1)+EX = np — n2p? = np(1— p). Also ~ lp? + np and Var(d EX? — (EX)? = n(n — 1)p? + Mx(t) = Be = Ye (Ja — py = (1— pt pet)" vee R. & By standard arguments, we can establish the existence of these moments. Note 1.232. Observe that Binomial(J,p) distribution is the same as Bernoulli(p) distribution. ‘We shall explore the connection between Binomial and Bernoulli distributions later in the course. Remark 1.233 (Factorial moments). In the computation for EX? for X ~ Binomial(n,p), we first computed EX(X — 1), which is easy to compute. It turns out that expectations of the form EX(X ~ 1), EX(X —1)(X —2) ete. are often easy to compute for integer valued RVs X. We refer to such expectations as factorial moments of X. Remark 1.234 (Symmetry of Binomial(n, 3) distribution). Let X ~ Binomial(n, p) and let ¥ = n—X. Since Mx(t) = (1 — p+ pe')?, Wt € I, we have My(t) = Bel¥ = Bet My(=t) = e"(1 =p + pe“t)" = (+ (1 phe)" Since MGFs determine the distribution, we conchide that Y ~ Binomial(n,1—p). In particular, ifp = 3, then Y =n—X 4 X ~ Binomial(n, }). Rewriting the relation, we get $—X 4X — 8. Therefore, X ~ Binomial(n, }) is symmetric about $ We now look at more examples of discrete RVs. Later in the course, we shall discuss their motivation through various random experiments8 Example 1.235 (Uniform RVs support Sx = {21,22,-'+ a} and pan. fx : R — [0,1] given by with support on a finite set). Consider a discrete RV X with 3, ifr © Sx, fx(a) = 0, otherwise. We had considered the case Sy = {1,2,+++ 6} in Example 1.180 and computed the expectation In the general setting, we have 1 a1 EX=* ox, EX*=2 263 a8 , Mx(t) = Ee = L ¥ et weeR nS and hence Var(X) can be computed by the formula EX? — (EX)?. By standard arguments, we can establish the existence of these moments. Example 1.236 (Poisson (x) RV). Fix A> 0, Note that e* = oj, 4 and hence the function f:R— 00,1] given by A, if € {0,1,2,---} F(a) = 0, otherwise. isa pam... An RV X is said to follow Poisson(.) distribution or equivalently, X is a Poisson() RV if its distribution is given by the above p.m.f.. Recall that we have already computed the following EX = A,Var(X) = and Mx(t) = eM),vi € IR in Example 1.216. As done for the case of Binomial(n, p) RVs, we can compute factorial moments. For example, BX 1) = Sale er a 7 Gai In fact, EX(X ~ 1) ---(X — (n= 1)) =" for alln > 1 yen Example 1.237 (Geometric (p) RV). Fix p € (0,1). Note that function f : R > [0,1] given by ©, p(1 —p)* = 1 and hence the sle)= p(L =p)", if € {0,1,2,---}, 0, otherwise.6 isap.m.f.. An RV X is said to follow Geometrie(p) distribution or equivalently, X is a Geometrie(p) RV if its distribution is given by the above pm. Let us compute the MGF. Here, —? _ 1— (ple Mx(t) = Ee™ = Ye*p(1 — p)* ca for all ¢ such that 0 O and p € R. An RV X is said to follow Cauchy(j1, 6) distribution if its distribution is given by the p.df 6 1 5 Fee x(x) =0 ‘The fact that fx is a pL. is easy to check. Set y = = and observe that oo lps od 2/7 1 Qe ee [listeners 2 [ge =P aap ee Few We have already considered the case 1 = 0,0 = 1 in Example 1.186 and Example 1.218, where we have seen that EX and the MGF do not exist for this distribution. In the general setting, note that 4S# ~ Cauchy(0,1) and by a similar argument, we can show that EX and MGF do not exist. Moreover, fx(¢ +2) = fir(u — 2), Vx € R and using Remark 1.226, we conclude that X is symmetric about Example 1.240 (Exponential(\) RV). Let A > 0. Note that Js" exp(~4) de = \ and hence the fimetion f B+ [0,00) given by texp(—£), if >0, fy => P(—$), 0, otherwise is a pdf. An RV _X is said to follow Exponential(A) distribution or equivalently, X is an Exponential(A) RV if its distribution is given by the above p.d.f.. We have already considered the case \ = 1 in Example 1.217, where we computed the moments and the MGF. Following similar arguments, in the general setting we have EX*= "nl, Var(X) =, Mx(@) =(1—at)"1,ve 0, On (0,00), consider the function a +> i 2*te~* dr, It is called the Gamma function and the value at any o > 0 is denoted by F(a) Remark 1.242. We recall some important properties of the Gamma function. (a) For a > 0, we have P(a) > 0. (b) Ta) = (@- P(e = 1), ifa > 1 (c) P(U) = fg? e* de = 1 and hence using (b), P(n) = (» — 1)! for alll positive integers n.a (d) T() = J eer dx = Ye. Putting 2 = ¥, this relation may be rewritten as 1 2 y G)- on ( 5) (e) Fix § > 0. Putting 2 — 4, in the integral for P(a) = [@°x*te~* dr, we get Ta) = Joe y13-* exp(— 8) dy. Va. Example 1.243 (Gamma(a, 3) RV). Fix a > 0,4 > 0. By the properties of the Gamma function described above, the function f : I — [0, 90) defined by Tay 21 3-*exp(—4), if x >0, f(z) = 0, otherwise. isa pdf. An RV X issaid to follow Gamma(a, 8) distribution or equivalently, X isa Gamma(a, 3) RV if its distribution is given by the above p.d.f. Note that for a = 1, we get back the pdf for an Exponential(9) RV (see Example 1.240), i.e. Gamma(1, 8) distribution is the same as Exponential(3) distribution. For general a > 0, > 0, we have EX =08, Var(X)=a8?, Mx(t) = (1- Bt) ed By standard arguments, we can establish the existence of these moments Example 1.244 (Normal(js, 02) RV). Fix 4 € R,o > 0. Note that P(3) = Jo° Hertdt = VF (see Remark 1.242). Putting t 1. Putting y = 2(x — 2) (equivalently, x = oy + y:), we have se lleo (SF) de Therefore, the function f : R — (0,00) defined by (z=? on (- sai) Wee and after suitable manipulation, we have 7. /% exp (—¥) dyn isapd.f. An RV X is said to follow Normal(j1, 02) distribution or equivalently, X is a Normal(j, 0?) RV, denoted by X ~ N(u1,0?) if its distribution is given by the above pdf. If X ~ N(u,0?), from our above discussion we conclude that Y = 4=# ~ N(0,1). Now, My(t) = Ee¥ a/v (-4) dy (Sel) In particular, yy(t) = In My(t) = $,Vt © R with W(t) = t,.u"@) = 1,ve © R. Evaluating at t = 0, by Proposition 1.215 we conclude that EY = 0 and Var(Y) = 1. But X = oY + and hence EX = 1, Var(X) = 0. This yields the interpretation of the parameters yz and o in the distribution of X. Further, My(t) = Ee'® = Zell¥ +) = o My (ot) = exp(ut + 407t?), Wt € R. Definition 1.245 (Standard Normal RV). We say X is a Standard Normal RV if X ~ N(0,1), ie, EX =0 and Var(X) = 1 Notation 1.246. Normal RVs are also referred to as Gaussian RVs and Normal distribution as Gaussian distribution. Remark 1.247 (Symmetry of Gaussian Distribution). If X ~ N(s,02), note that far(u +2) = fx(u—2),We € Rand using Remark 1.226, we conclude that X is symmetric about its mean y. Remark 1.248 (Moments of a Standard Normal RV). Let X ~ N(0,1). Then X is symmetric about 0 and using Proposition 1.225, we conclude EX" = 0 for all odd positive integers n. If n is an even positive integer, then n = 2m for some positive integer m and n 1 eon 2 ext Fe [een (9) & 1 pmep (2 =a (s)«mB 4 [emo (-2) a gm peo “3 a "4 exp (—v) dy, (putting y = o 1 Tim +5) -# (0) = (2m~=1) x where we have used the properties of the Gamma fimetion, In particular, EX* = 3. Definition 1.249 (Beta function). Recall that the integral /2 x°~!(1—)*! de exists if and only if @ > O and 8 > 0. On (0,00) x (0, 00), consider the function (a, 8) + ff 2°-"(1~ 2)" dr. It is called the Beta function and the value at any («, (2) is denoted by B(a, 8) Remark 1.250. Note that for a > 0,8 > 0, we have B(a, 8) > 0 and B(a, 8) = B(B,a). Moreover, To) (8) Ba.) = Tara Example 1.251 (Beta(o, 3) RV). Fix a > 0,8 > 0, By the properties of the Beta function described above, the function f : R - (0,00) defined by A fle) = oar 1, fx (0,1) 0, otherwise. isa pdf. An RV X is said to follow Beta(a, 8) distribution or equivalently, X is a Beta(a, 8) RV if its distribution is given by the above p.d.f.. If @ = 8, then f(1— x) = f(x), ¥x © R and hence X £1—X. Then, xX-} 42 X, ie, X is symmetric about 4. For all «, 8,r > 0, we have Bla+r,8) Bla, 8) . 1 Tatra, eV dp = EX’ men (1-2)? de =