Show equality involving probability generating function.

30 Views Asked by At

I have gotten stuck on the following problem. We wish to show the following equality holds:

We want to show that the following equality holds:

$$ \frac{g_N(t) - 1}{t - 1} = \sum_{n \geq 0} P(N > n) t^n $$

where (g_N(t)) is the probability generating function for some unknown random variable (N). N assumes non-negative integer values and $|t|<1$

I have developed (g_N(t)) into

$$ \sum_{n \geq 0} t^n P(N = n) $$

giving us on the left-hand side:

$$ \frac{\sum_{n \geq 0} t^n P(N = n) - 1}{t - 1} $$

and on the right-hand side:

$$ \sum_{n \geq 0} (1 - P(N = n)) t^n $$

equating both sides gives us:

$$ \frac{\sum_{n \geq 0} t^n P(N = n) - 1}{t - 1} = \sum_{n \geq 0} (1 - P(N = n)) t^n $$

From here on I have tried multiplying over (t-1), I have tried to write $t^nP(N=n)$ as a geometric series in closed form $\frac{P(N=n)}{1-t}$ since $n>= 0$ and |t|<1. But I have just gotten more confused in the process. Would appreciate some help in showing this equality. Thanks.

1

There are 1 best solutions below

1
On BEST ANSWER

$$\sum_{n}P(N>n)t^{n}=\sum_{n=0}^{\infty}\sum_{m=n+1}^{\infty}P(N=m)t^{n}$$

Change the order of summation(Valid by Tonelli's Theorem as summands are positive):

$$\sum_{m=1}^{\infty}\sum_{n=0}^{m-1}P(N=m)t^{n}$$

Use Geometric sum formula,

$$\sum_{m=1}^{\infty}P(N=m)\bigg(\frac{1-t^{m}}{1-t}\bigg)$$

Rearrange (start the summation at $0$) now that the numerator is :-

$$1-P(N=0)+P(N=0)-\sum_{m=0}^{\infty}P(N=m)t^{m}=1-g_{N}(t)$$

And so you get the result.