# Moments and Moment-gen. function.

1. Oct 23, 2011

### SithV

Hi people!

Im having problems with my Prob.Theor. assignment=(
I was hoping that u might be able to help me...

I have 2 problems that i ve no idea how to solve!Oo

Heres 1st one
We re given rand. var. X, its mean value U, the stand deviation S (sigma).
We need to show that E(z)=0 and var(z)=1
if the relation between X and Z is this eq. Z=X-U/S

2nd
Show if a rand. var. has the prob. density
f(x)=1/2*Exp[-lxl] -inf<X<inf
lxl-abs value
then its moment gen func. is
Mx(t)=1/1-t^2

we re using the formula from the definition
and gettin this
1/2(Int[Exp[tx]*Exp[-lxl]) -inf<X<inf

but lxl=+-x

then we get 2 integrals
1/2(Int[Exp[tx]*Exp[-x])
and
1/2(Int[Exp[tx]*Exp[x])
both in -inf<X<inf

now if we integrate it we get
1/2Exp[x(t-1)]/t-1
and
1/2Exp[x(t+1)]/t+1
both in -inf<X<inf

whats next?=(

Would appreciate any help!=(

2. Oct 23, 2011

### mathman

For your first problem z=(x-U)/S. Direct calculation will give you the results.

For your second problem, the ranges of integration of the split integrals are (-∞,0) for the +x integral and (0,∞) for the -x integral. Plug in x=0 and the resulting terms (be careful of signs) add to get the answer.

3. Oct 23, 2011

### SithV

Ok, i think i got the second one.., but what do i do with 1/2 in front of 1/1+t and 1/1-t when i multiply them?

Now the to first one. You mean i need to integrate it? Like this: Int[Z*(X-U)/S], but what range do i pick???
And then var(Z) i can find from S^2=U'2-U^2
which is S=Sqrt[U'2-U^2]
where U^2 is E(Z)
and U'2=Int[(Z^2)*(X-U)/S]
but again what range???
Thank you!!!

4. Oct 24, 2011

### mathman

1/(1-t) + 1/(1+t) = 2/(1-t2). So multiply by 1/2 to get your answer.

The range for the integration is the entire real line.
A simpler approach is as follows:
E[(X-U)/S] = {E(X) - U}/S = 0
E[({X-U)/S}2] = {E(X2) -2UE(X) +U2}/S2 = 1