Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

$X|z$ has a multinomial distribution $MN(m, \mathbf{q}(z))$ where $z$ is either 0 or 1 with probability $1/2$. I need to find: $E_X[\max\{\Pr(z=1|X), \Pr(z=0|X\}]$. Is there an analytical form to this?


My attempt so far, but it's not really promising:

By Bayes' rule: $\Pr(z=1|X)=\frac{\Pr(X|z=1)}{\Pr(X|z=1)+\Pr(X|z=0)}$

$\Pr(X=X_0|z=1)=\exp(lgamma(m+1)+X^'_0 \log(q(z))-lgamma(X_0+1)))$ where $lgamma$ is log of the gamma function.

It can be simplified to: $\Pr(z=1|X)=\frac{1}{1+\exp(X'Q)}$ where $Q=\log(q(0))-\log(q(1))$.

I thought of doing a simulation from here but even then it's tricky. How do I simulate X when I only know the distribution of $X|z$? Also, ideally there should be an analytical solution to this?

share|improve this question
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.