You are on page 1of 13
IX, ..., Xn are independent random variables and that X; ~ Gamma(q;, 8) fori=1,...,n. Let Y = $0, Xi. Then Y ~ Gamma( SE 8). Solution Sketch: n My(t) = [[G- St)" = (1-8), t< 1/8. i=l We say that X has a normal distribution if it has the pdf _ 2 fle) = een {Sh 00 < 2 < 00, where —oo < jt < oo and a? > 0 are two parameters. Write X ~ N(u,0?). Properties: » Expectation: j1. We call j: the location parameter » Variance 0”. We call c the scale parameter. > Moment generating function: Eexp(tX) = Eexp(t(oZ + 1) = exp (ut) Eexp (toZ) loo 1 =o tt) © aot | =ex ib + exp (jut) exp (50 ) exp (« + 5 Arandom variable Z is said to follow a standard normal distribution if it has pdf fe)= een {-F}, re Write Z ~ N(0, 1). Properties of Stand: mal Distribution Moment generating function: Mr(t) = RexpltZ) = [ . exp(ts) ex (-32) de vi) fend) coll) [aol eens ~ow (57), -~ Var(Z) =1. Let Xj, ..., X;, be independent random variables such that Y; follows N(1;,07). Then, for constants aj, ..., an, Y= Soak; follows N (= 1 Sah) . i=l i=1 i=l Proof. n 1 My(t) = [exp {ta + pater} i=l n n 1 = exp {! S fli, +50 s ceth. i=l Recall that the mgf of N(j, 0”) is exp(it + 5074?) Let Xj,..., X,, be i.i.d. with common distribution N(j,0?). Then, ~ 1< a Y= 530% follows x(u.7) i=l Let X,, Xo, and X3 be i.i.d. random variables with common mgf exp{t + 207}. Compute the probability P(X, < 3). Derive the mgf of Y =X; + 2X» — 2X3. Compute the probability P(Y > 7). One Special [’: Exponential Distribu on] » X has exponential distribution X ~ exp(3) if X isT(a = 1,8). > pdf and cdf: 4.-2/8 2 50 S(@)= 0 otherwise. F(x@) =1—e7*/8 x > 0. » Expectation: 3. » Variance: 62. aeeeeed ate eee The exponential distribution is frequently used to model the following: Life time of electronic components The time until a radioactive particle decays. The time until default (on payment to company debt holders) in reduced form credit risk modeling. The time it takes for a bank teller to serve a customer. WX. Let Y = on, Xe Then ¥ ~ 2 ( 0, » A Chi-square distribution with parameter r, X ~ y?(r), is defined to be Gamma(a = r/2, 8 = 2). > pdf Reppert tea? r>0 Sle) = 0 otherwise. > mgf: Mx(t) =(1—20)7"?, t< 1/2. > Expectation: r. > Variance: 2r. -. X,, are independent random variables and that Xe ~x2(r) for’ = Suppose Z ~ N(0, 1), then W = Z? follows x2(1). Proof. 1. Since W = Z?, then Z = VW when Z > 0 and Z = —/W if Z<0. 2. The pdf of Z is fz(z) = seow {-3}. 3. The pdf of W is Fw) 1 (mY | 1}, vm? y| 1 ~ mee {- 2 Vice t mee {- 2 } Vw ye exp {-$} recall that [(0.5) = V7 a 1 os w “Toss exp {- 3} which is the paf of P(a = 0.5, 3 = 2), which is x?(1). o Beta Function B(a, 6) Another definite integral which is related to the T-funetion is the Beta function B(a, b) which is defined as B(a,b) = f et (1-4)? dt, a>0, b>O (1.72) ‘The relationship between the B-finetion and the P-function ean be demonstrated easily. By means of the new variable Therefore Eq. 1.72 becomes B(a,b) = [ le a>o, b>o (73) Now it can be shown that [re mate = FO) (74) which is obtained from the definition of the P-function with the change of variable s = pt. Setting p= 1+ wand == a+b, we get 1 1 sue gost 5) (1+ uj rary L . “ a7) and substituting this result into the Beta funetion in Eq. 1.73 gives = a [ee we dn BOD = FH u fre ut du _ Te) [Pay sero ort _ Pa) + T) Tarhy (1.76) Betaly,.5] 45 10 | By.z) ° -10 Figure 1.4: Plot of Beta Function All the properties of the Beta fumetion can be derived from the relationships linking the T-finetion and the Beta function. Other forms of the beta function are obtained by changes of varinbles. ‘Thus uel du Bed) =f arm af Blab) = 2 [ sint™#0 coset dd by t= sin? (173) [The Beta bution Motivating example: Suppose that X has a Gamma(a, 1) distribution, that X) has a Gamma(/3, 1), and that X;, X» are independent, where a > Oand 9 > 0. Let Yj = X; + X» and Yo = Ni/(X1 + Xo). Show that ¥; and Yo are independent. Solution: One-to-one transformation: «ry = yiyp and «2 = y1(1 — yp), 0 < 21,72 < cogives 0< y <0, 0< yy <1. J, Joint pdf of X; and X»: 1 fxy,x (01,22) = Tora we. O» Mean: at > Variance: oB (a+ 6+ 1(a +)?” » Beta distribution is not a special case of Gamma distribution. PDF | The Weibull Distribution The Weibull distribution is used similarly to the exponential distribution to model times to an event, but with an extra parameter included for flexibility. Definition Arandom variable X is said to have a Weibull distribution with parameters a and 6 («> 0, 6 > 0) if the pdf of Xis Fs af) = aprtx1e B x20 What is this distribution if alpha = 1? Both «and f can be varied to obtain a number of different- looking density curves, as illustrated below. \Wetbull density curves:

You might also like