ex5m3_4.doc

advertisement

Random Signals for Engineers using MATLAB and Mathcad

Copyright

1999 Springer Verlag NY

Example 3.4 Moment Generating Functions



The Moment generating function is used to derive the first and second moments of the binomial and geometric distributions. The moment generating function for a binomial is given syms t k n p q kf=sym('k!');

M=exp(t*k)*subs(kf,k,n)*p^k*q^(n-k)/(subs(kf,k,k)*subs(kf,k,n-k)); pretty(M)

k (n - k)

exp(t k) n! p q

-----------------------

k! (n - k)!

M=simplify(symsum(M,k,0,n)); pretty(M)

n

(q + exp(t) p)

Matlab performed the simplification of the summation. We now use Equation 3.3-7 to find the moments

MP=diff(M,t);

M1=limit(MP,t,0);

M1=simplify(M1); pretty(M1)

(n - 1)

(q + p) n p

Now we substitute t = 0 into the expression to find as example 3.2 that when we use the fact p + q = 1 we have E[T] = n p

The second moment is found by differentiating again

MPP=diff(MP,t)

MPP =

(q+exp(t)*p)^n*n^2*exp(t)^2*p^2/(q+exp(t)*p)^2+(q+exp(t)*p)^n*n*exp(t)* p/(q+exp(t)*p)-(q+exp(t)*p)^n*n*exp(t)^2*p^2/(q+exp(t)*p)^2

Substitution for t = 0 and simplifying

M2=limit(MPP,t,0); pretty(simplify(M2))

(n - 2)

(q + p) n p (n p + q)

Using the fact that (p + q ) = 1 we have

E[T2] = pretty(subs(M2,q,1-p))

n p (n p + 1 - p)

The variance is computed in the usual manner VAR[X] = E[T2] - E[T]2 =

VAR=subs(M2,q,1-p) - subs(M1,q,1-p)^2; pretty(simplify(VAR))

2

n p - n p

The process is repeated for the geometric distribution. First we obtain a closed form for the moment generating function using Matlab

ML=simplify(symsum(exp(t*k)*p^k*q,k,0,inf)) pretty(ML)

ML =

-q/(-1+exp(t)*p)

q

- -------------

-1 + exp(t) p

The first and second moments are generated by differentiation and substitution for t = 0 in the resultant expression

E[K] =

MLP=limit(diff(ML),t,0)

MLP =

1/(p-1)^2*q*p

We repeat the process for the second moment

E[K 2 ]=

MLPP=limit(diff(ML,2),t,0); pretty(MLPP)

q p (p + 1)

- -----------

3

(p - 1)

The variance is computed and simplified by noting q = 1 - p and by substitution VAR[K] = E[K2] - E[K]2

=

VARL=subs(MLPP,q,1-p) - subs(MLP,q,1-p)^2 pretty( simplify(VARL))

VARL =

-(1-p)*p*(p+1)/(p-1)^3-1/(p-1)^4*(1-p)^2*p^2

p

--------

2

(p - 1)



Download