Namespaces
Variants
Actions

Difference between revisions of "Saddle point in game theory"

From Encyclopedia of Mathematics
Jump to: navigation, search
(Importing text file)
 
m (tex encoded by computer)
 
Line 1: Line 1:
A point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830601.png" /> of a function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830602.png" /> defined on the Cartesian product <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830603.png" /> of two sets <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830604.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830605.png" /> such that
+
<!--
 +
s0830601.png
 +
$#A+1 = 29 n = 0
 +
$#C+1 = 29 : ~/encyclopedia/old_files/data/S083/S.0803060 Saddle point in game theory
 +
Automatically converted into TeX, above some diagnostics.
 +
Please remove this comment and the {{TEX|auto}} line below,
 +
if TeX found to be correct.
 +
-->
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830606.png" /></td> <td valign="top" style="width:5%;text-align:right;">(*)</td></tr></table>
+
{{TEX|auto}}
 +
{{TEX|done}}
  
For a function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830607.png" /> the presence of a saddle point is equivalent to the existence of optimal strategies (cf. [[Strategy (in game theory)|Strategy (in game theory)]]) for the players in the [[Two-person zero-sum game|two-person zero-sum game]] <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830608.png" />.
+
A point ( x  ^ {*} , y  ^ {*} ) \in X \times Y $
 +
of a function  $  F $
 +
defined on the Cartesian product  $  X \times Y $
 +
of two sets  $  X $
 +
and  $  Y $
 +
such that
  
 +
$$ \tag{* }
 +
F ( x  ^ {*} , y  ^ {*} )  = \
 +
\max _ {x \in X } \
 +
F ( x, y  ^ {*} )  = \
 +
\min _ {y \in Y } \
 +
F ( x  ^ {*} , y).
 +
$$
  
 +
For a function  $  F $
 +
the presence of a saddle point is equivalent to the existence of optimal strategies (cf. [[Strategy (in game theory)|Strategy (in game theory)]]) for the players in the [[Two-person zero-sum game|two-person zero-sum game]]  $  \Gamma = ( X, Y, F  ) $.
  
 
====Comments====
 
====Comments====
A point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s0830609.png" /> satisfying the condition (*) is called a saddle point of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306010.png" /> in general. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306011.png" /> is a differentiable function on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306012.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306013.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306014.png" />, while the Hessian matrix <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306015.png" /> is non-singular and neither positive definite nor negative definite, then locally near <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306016.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306017.png" /> is a saddle point. The corresponding splitting of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306018.png" /> near <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306019.png" /> is determined by the negative and positive eigenspaces of the Hessian at <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306020.png" />.
+
A point $  ( x  ^ {*} , y  ^ {*} ) \in X \times Y $
 +
satisfying the condition (*) is called a saddle point of $  F $
 +
in general. If $  F $
 +
is a differentiable function on $  \mathbf R  ^ {n} $
 +
and  $  ( \partial  F / \partial  x _ {i} ) ( x  ^ {*} ) = 0 $,  
 +
$  i = 1 \dots n $,  
 +
while the Hessian matrix $  ( \partial  ^ {2} F / \partial  x _ {i} \partial  x _ {j} ) ( x  ^ {*} ) $
 +
is non-singular and neither positive definite nor negative definite, then locally near $  x  ^ {*} $,  
 +
$  x  ^ {*} $
 +
is a saddle point. The corresponding splitting of $  \mathbf R  ^ {n} $
 +
near $  x  ^ {*} $
 +
is determined by the negative and positive eigenspaces of the Hessian at $  x  ^ {*} $.
  
Indeed, by the [[Morse lemma|Morse lemma]] there are coordinates <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306021.png" /> near <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306022.png" /> such that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306023.png" /> has the form
+
Indeed, by the [[Morse lemma|Morse lemma]] there are coordinates $  y _ {1} \dots y _ {n} $
 +
near $  x  ^ {*} $
 +
such that $  F $
 +
has the form
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306024.png" /></td> </tr></table>
+
$$
 +
F( y)  = F( x  ^ {*} ) - y _ {1}  ^ {2} - \dots - y _ {r}  ^ {2}
 +
+ y _ {r+} 1  ^ {2} + \dots + y _ {n}  ^ {2} ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306025.png" /> is the index of the quadratic form determined by the symmetric matrix <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306026.png" />. (The index of a quadratic form is the dimension of the largest subspace on which it is negative definite; this is also called the negative index of inertia (cf. also [[Quadratic form|Quadratic form]] and [[Morse index|Morse index]]).)
+
where $  r $
 +
is the index of the quadratic form determined by the symmetric matrix $  ( \partial  ^ {2} F / \partial  x _ {i} \partial  x _ {j} ) ( x  ^ {*} ) $.  
 +
(The index of a quadratic form is the dimension of the largest subspace on which it is negative definite; this is also called the negative index of inertia (cf. also [[Quadratic form|Quadratic form]] and [[Morse index|Morse index]]).)
  
Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306027.png" /> be the spaces of strategies of two players in a zero-sum game and let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306028.png" /> be (the first component of) the pay-off function (cf. [[Games, theory of|Games, theory of]]). Then a saddle point is also called an equilibrium point. This notion generalizes to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/s/s083/s083060/s08306029.png" />-player non-cooperative games, cf. [[#References|[a2]]], Chapt. 2; [[Games, theory of|Games, theory of]]; [[Nash theorem (in game theory)|Nash theorem (in game theory)]]; [[Non-cooperative game|Non-cooperative game]].
+
Let $  X, Y $
 +
be the spaces of strategies of two players in a zero-sum game and let $  F: X \times Y \rightarrow \mathbf R $
 +
be (the first component of) the pay-off function (cf. [[Games, theory of|Games, theory of]]). Then a saddle point is also called an equilibrium point. This notion generalizes to $  n $-
 +
player non-cooperative games, cf. [[#References|[a2]]], Chapt. 2; [[Games, theory of|Games, theory of]]; [[Nash theorem (in game theory)|Nash theorem (in game theory)]]; [[Non-cooperative game|Non-cooperative game]].
  
 
====References====
 
====References====
 
<table><TR><TD valign="top">[a1]</TD> <TD valign="top">  M.W. Hirsch,  "Differential topology" , Springer  (1976)  pp. Chapt. 6</TD></TR><TR><TD valign="top">[a2]</TD> <TD valign="top">  J. Szép,  F. Forgó,  "Introduction to the theory of games" , Reidel  (1985)  pp. 171; 199</TD></TR></table>
 
<table><TR><TD valign="top">[a1]</TD> <TD valign="top">  M.W. Hirsch,  "Differential topology" , Springer  (1976)  pp. Chapt. 6</TD></TR><TR><TD valign="top">[a2]</TD> <TD valign="top">  J. Szép,  F. Forgó,  "Introduction to the theory of games" , Reidel  (1985)  pp. 171; 199</TD></TR></table>

Latest revision as of 08:12, 6 June 2020


A point $ ( x ^ {*} , y ^ {*} ) \in X \times Y $ of a function $ F $ defined on the Cartesian product $ X \times Y $ of two sets $ X $ and $ Y $ such that

$$ \tag{* } F ( x ^ {*} , y ^ {*} ) = \ \max _ {x \in X } \ F ( x, y ^ {*} ) = \ \min _ {y \in Y } \ F ( x ^ {*} , y). $$

For a function $ F $ the presence of a saddle point is equivalent to the existence of optimal strategies (cf. Strategy (in game theory)) for the players in the two-person zero-sum game $ \Gamma = ( X, Y, F ) $.

Comments

A point $ ( x ^ {*} , y ^ {*} ) \in X \times Y $ satisfying the condition (*) is called a saddle point of $ F $ in general. If $ F $ is a differentiable function on $ \mathbf R ^ {n} $ and $ ( \partial F / \partial x _ {i} ) ( x ^ {*} ) = 0 $, $ i = 1 \dots n $, while the Hessian matrix $ ( \partial ^ {2} F / \partial x _ {i} \partial x _ {j} ) ( x ^ {*} ) $ is non-singular and neither positive definite nor negative definite, then locally near $ x ^ {*} $, $ x ^ {*} $ is a saddle point. The corresponding splitting of $ \mathbf R ^ {n} $ near $ x ^ {*} $ is determined by the negative and positive eigenspaces of the Hessian at $ x ^ {*} $.

Indeed, by the Morse lemma there are coordinates $ y _ {1} \dots y _ {n} $ near $ x ^ {*} $ such that $ F $ has the form

$$ F( y) = F( x ^ {*} ) - y _ {1} ^ {2} - \dots - y _ {r} ^ {2} + y _ {r+} 1 ^ {2} + \dots + y _ {n} ^ {2} , $$

where $ r $ is the index of the quadratic form determined by the symmetric matrix $ ( \partial ^ {2} F / \partial x _ {i} \partial x _ {j} ) ( x ^ {*} ) $. (The index of a quadratic form is the dimension of the largest subspace on which it is negative definite; this is also called the negative index of inertia (cf. also Quadratic form and Morse index).)

Let $ X, Y $ be the spaces of strategies of two players in a zero-sum game and let $ F: X \times Y \rightarrow \mathbf R $ be (the first component of) the pay-off function (cf. Games, theory of). Then a saddle point is also called an equilibrium point. This notion generalizes to $ n $- player non-cooperative games, cf. [a2], Chapt. 2; Games, theory of; Nash theorem (in game theory); Non-cooperative game.

References

[a1] M.W. Hirsch, "Differential topology" , Springer (1976) pp. Chapt. 6
[a2] J. Szép, F. Forgó, "Introduction to the theory of games" , Reidel (1985) pp. 171; 199
How to Cite This Entry:
Saddle point in game theory. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Saddle_point_in_game_theory&oldid=18503
This article was adapted from an original article by V.L. Kreps (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article