Let X be a poisson distributed variable with parameter $\lambda$, what is $E[2^{-x}]$?

I know it should equal to

$\sum_{i=0}^\infty p(\frac{1}{2^x}=i)i$

but how do I calculate this sum?

If I isolate X in $1/2^x$, I get $p(x=log_{\frac{1}{2}}i)$ and then

$E[2^{-x}]=\sum_{i=0}^\infty i * \frac{e^{-\lambda}\lambda^{log_{1/2}i}}{(log_{1/2}i)!} = e^{-\lambda}\sum_{i=0}^\infty i * \frac{\lambda^{log_{1/2}i}}{(log_{1/2}i)!}$

But because of the 'i' in the sum I don't know how to solve this either.

Thanks in advance, and sorry if there were English mistakes.


Note for any given $\lambda>0$, using the property of mpf of Poisson distribution, we have $$\sum_{i=0}^\infty \frac{\lambda^i}{i!}e^{-\lambda}=1.$$

Thus $$E[2^{-X}]=\sum_{i=0}^\infty 2^{-i}\frac{\lambda^i}{i!}e^{-\lambda}=e^{-\lambda/2}\sum_{i=0}^\infty \frac{(\lambda/2)^i}{i!}e^{-\lambda/2}=e^{-\lambda/2}.$$

  • $\begingroup$ Thank you, but how did you reach the first sum? is it correct to just replace the i with the function of X, and if so, why? Also, why did the e^lambda get out of the sum with e^-lambda/2? $\endgroup$
    – hrsidkpi
    Feb 4 '17 at 16:54
  • $\begingroup$ Nevermind, I just realized that you split the e^-lambda to two and only ejected 1 of them. Still wondering about the first question though. $\endgroup$
    – hrsidkpi
    Feb 4 '17 at 17:12
  • $\begingroup$ @hrsidkpi For the first question, by definition, $E[2^{-X}]=\sum_{i=0}^\infty 2^{-i}P(X=i)$. $\endgroup$
    – John
    Feb 4 '17 at 18:53
  • $\begingroup$ thank you, the identity I was looking for was E[g(x)]. $\endgroup$
    – hrsidkpi
    Feb 6 '17 at 8:45

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.