Help with p.m.f and p.g.f question.

54 Views Asked by At

So we have the following question.

$Y$ is a discrete r.v with the positive integers as values. The p.m.f. of $Y$ is of the form:

$P(Y=k) = c \cdot \frac{1}{k!}, k = 1,2...$

a) what is the value of c?

$U_1, U_2, \dots$ are I.I.D r.v.s with $U_i \epsilon U(0,1)$ $U_1, U_2,...$are also independent of Y.

b) Set

$M = max(U_1, U_2, ..., U_Y)$

Show that for any $t$ in the interval $[0,1]$

$P(M\le t) = g_y(t)$

where $g_y(t)$ is the p.g.f. of $Y$.


So what I've done is calculated the c using the sum of p.m.f = 1 i.e.

$\sum_{k=1}^\infty p_x(k) = 1$

so

$\sum_{k=1}^\infty c \cdot \frac{1}{k!} = 1$

$c \sum_{k=1}^\infty \frac{1}{k!}$ = $c ([\sum_{k=0}^\infty \frac{1}{k!}] -1)$

this gives

$c = \frac{1}{e-1}$

Now for the b) part I first calculate the p.g.f:

$g_y(t) = \sum_{k=1}^\infty t^k \cdot \frac{1}{e-1} \frac{1}{k!}$

$= \frac{1}{e-1} ([\sum_{k=0}^\infty \frac{t^k}{k!}] - 1)$

$= \frac{e^t - 1}{e -1}$

Now this is where i get uncertain. I then compare to $P(M \le t)$ which is due to independence:

$P(M \le t) = P(Max(U_1, U_2, ..., U_Y) \le t) = \prod_{i=1}^Y P(U_i \le t)$

As $U_i \epsilon U(0,1)$ we get $P(U_i \le t) = t and Therefore:

$\prod_{i=1}^Y P(U_i \le t) = t^Y$

Which is not the same as $g_y(t)$.

Can anyone explain where my reasoning is wrong? Thanks!