<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
 
 <title>----</title>
 <link href="http://blog.stompchicken.com/atom.xml" rel="self"/>
 <link href="http://blog.stompchicken.com/"/>
 <updated>2011-10-15T12:40:15+01:00</updated>
 <id>http://blog.stompchicken.com/</id>
 <author>
   <name>Stephen Spencer</name>
   <email>spam@stompchicken.com</email>
 </author>

 
 <entry>
   <title>Nash's Theorem</title>
   <link href="http://blog.stompchicken.com/blog/Nash%27s-Theorem"/>
   <updated>2011-10-07T00:00:00+01:00</updated>
   <id>http://tom.preston-werner.com/blog/Nash's-Theorem</id>
   <content type="html">&lt;p&gt;I recently encountered a proof of Nash's theorem that was so straightforward that it surprised me enough to want to share it.&lt;/p&gt;

&lt;h1&gt;Preamble&lt;/h1&gt;

&lt;p&gt;Let's assume we have a n-player game where each player can select from a finite number of deterministic strategies. This is the case in something like chess, or poker.&lt;/p&gt;

&lt;p&gt;A player plays a pure strategy by selecting a single deterministic strategy from their available set. A player plays a mixed strategy if they select some probability distribution over all their available strategies and select a strategy to play at the start of the game by sampling from their distribution.&lt;/p&gt;

&lt;p&gt;We need to define S, a set of all strategy profiles, which are assignment of players to mixed strategies. There is also a payoff to each player for playing the strategy profile $s_i$, which we call $u_i(s)$&lt;/p&gt;

&lt;h1&gt;Nash's theorem&lt;/h1&gt;

&lt;p&gt;Nash's theorem says that for each player there is some strategy profile &lt;code&gt;$s^*$&lt;/code&gt; such that for each i&lt;/p&gt;

&lt;p&gt;$$u_i(s) \leq u$$&lt;/p&gt;

&lt;p&gt;for all s. The profile &lt;code&gt;$s^*$&lt;/code&gt; is called a Nash equilibrium&lt;/p&gt;

&lt;p&gt;and has been used as a 'solution' concept in games. I have some issues with calling it a solution, but it is an equilibrium in the sense that it does not profit an individual player to deviate from playing the strategy profile.&lt;/p&gt;

&lt;h1&gt;A sketch of the proof&lt;/h1&gt;

&lt;p&gt;First of all, we need a&lt;/p&gt;
</content>
 </entry>
 
 
</feed>

