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.

Suppose that

$V= max_i${$V_i+\epsilon_i $}

where $i=1...N$ and each $epsilon_i$ has a a type 1 extreme value distribution with $F(x)=e^{e^{-x}}$

I need to show that

$E[V]=u+ln(\sum_je^{V_j})$

where u is eulers constant, i.e. $E[\epsilon_i]$

I have so far that P(option i is picked)=$\frac{e^{V_i}}{\sum_je^{V_j}}$ (This seems to be a standard result, so I'm fairly sure of it)

Using this formula, I get that:

$E[V] = u +\sum_i\frac{e^{V_i}}{\sum_je^{V_j}}V_i $

and then I dont know how to proceed.

EDIT: Ok I realize that my formula for E[V] is incorrect, it should be

$E[V] = \sum_i\frac{e^{V_i}}{\sum_je^{V_j}}(V_i+E[\epsilon_i|V_i+\epsilon_i>V_j+\epsilon_j ] $ for all $j\neq i$

share|improve this question
add comment

1 Answer

up vote 1 down vote accepted

Let us first correct the value of $F$ given in the post and use $F(x)=\exp(-\mathrm e^{-x})$, for every $x$, as the common CDF of the random variables $\epsilon_i$.

Then, $V=\max\limits_iv_i+\epsilon_i$ hence, for every $x$, $[V\leqslant x]=\bigcap\limits_i[\epsilon_i\leqslant x-v_i]$. By independence, $$ \mathbb P(V\leqslant x)=\prod\limits_iF(x-v_i)=\prod\limits_i\exp(-\mathrm e^{-x+v_i})=\exp(-\mathrm e^{-x}\mathrm e^w)=F(x-w), $$ with $\mathrm e^w=\sum\limits_i\mathrm e^{v_i}$. Thus, $\mathbb P(V\leqslant x)=\mathbb P(\epsilon_1\leqslant x-w)$ for every $x$, that is, $V$ is distributed as $\epsilon_1+w$. In particular, $$ \mathbb E(V)=\mathbb E(\epsilon_1)+w=\gamma+\log\left(\sum_i\mathrm e^{v_i}\right), $$ where $\gamma=\mathbb E(\epsilon_1)$ is indeed Euler's constant.

share|improve this answer
add comment

Your Answer

 
discard

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

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