1
$\begingroup$

The Statement of the Problem:

For the random variable $X$ with pmf $p_X(k)=(1/2)^k, k = 1,2,...,$

(a) Calculate $E[X]$ and $E[X(X-1)]$.

(b) Use part (a) to compute Var$[X]$.

Where I Am:

Ok, so I plugged this problem into WolframAlpha to see what it had to say, and it said that the expected value is $\approx 1.27$. But then, when I put just the explicit formula for $E[X]$ in there, it returned the following:

$$ \sum_{k=1}^{\infty} \frac{k}{2^k}=2$$

What gives?

EDIT: I figured out what I did wrong with W|A. It interpreted my variable as a normally-distributed r.v. instead of just a plain old natural number.

$\endgroup$
1
$\begingroup$

Hints: $$ E(X)=\sum_{k=1}^\infty k\frac{1}{2^k}=2\sum_{k=1}^\infty\frac{k}{2^{k+1}}\cdot $$ To handle the RHS above, consider the function $f(x)=\sum_{k=1}^\infty\frac{1}{x^k}=\frac{1}{x-1}$ which is well-defined for $x>1$. Then $$ \frac{1}{(x-1)^2}=-f'(x)=\sum_{k=1}^\infty\frac{k}{x^{k+1}} $$ and so $E(X)=2\times\frac{1}{(2-1)^2}=2$. For $E[X(X-1)]$, use a similar trick: $$ E[X(X-1)]=\sum_{k=1}^\infty k(k-1)\frac{1}{2^k}=4\sum_{k=1}^\infty\frac{k(k-1)}{2^{k+2}}=4\left(\sum_{k=1}^\infty\frac{k(k+1)}{2^{k+2}}-\frac{1}{2}\sum_{k=1}^\infty\frac{k}{2^k}\right) $$ and consider $f''(x)$. Once you know $E[X(X-1)]$ and $E(X)$ then $$ E(X^2)=E[X(X-1)]+E(X),\quad \text{Var}(X)=E(X^2)-E(X)^2. $$ As for the discrepancy with Wolfram Alpha, you need to check whether you have interacted with WA correctly as $\sum_{k=1}^\infty\frac{k}{2^k}=2$ is correct.

$\endgroup$
1
$\begingroup$

Note that

\begin{align} \sum_{k} \frac{k}{2^k} &= \frac{1}{2} + \frac{1}{4} + \frac{1}{8} + \cdots = 1\\ & \qquad \quad \frac{1}{4} + \frac{1}{8} + \cdots = \frac{1}{2} \\ & \qquad \qquad \;\;\; \frac{1}{8} + \cdots = \frac{1}{4}\\ &\qquad \qquad \qquad \qquad\vdots \end{align}

So $\sum_{k} \frac{k}{2^k} = 2$

$\endgroup$
0
$\begingroup$

For your question $X$ is basically a geometric random variable which denotes the number of trials needed to get the first success in an infinite sequence of Bernoulli trials with $P(success) = p = q = P(failure) = 0.5$.

Now, $$ \begin{eqnarray} E(X) &=& \sum_{k=1}^{\infty} kp^k; \quad \text{actually }p = 1/2\\ &=& p \sum_{k=1}^{\infty} kp^{k-1}\\ &=& p \sum_{k=1}^{\infty} \frac{d}{dp} p^k\\ &=& p \frac{d}{dp}\sum_{k=1}^{\infty} p^k\\ &=& p \frac{d}{dp} \left(\frac{p}{1-p}\right)\\ &=& \frac{p}{(1-p)^2} \end{eqnarray} $$ Now put $p = \frac{1}{2}$.

Proceed in a similar way to compute $Var(X)$ and $E[X(X-1)]$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.