Scribd
Upload
88 views11 pages

Advanced Game Theory Slides

1) This document discusses game theory concepts like best response curves, Nash equilibriums, mixed strategies, and zero-sum games. 2) For the prisoner's dilemma game presented, there are three Nash equilibriums: (0,0), (2/3, 1/3), and (1,1). 3) In a zero-sum kicking game, the Nash equilibrium strategies are for the row player to kick left with probability 0.7 and the column player to jump left with probability 0.6, resulting in the row player scoring 62% of the time.

Uploaded by

Braden Muhl
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
88 views11 pages

Advanced Game Theory Slides

1) This document discusses game theory concepts like best response curves, Nash equilibriums, mixed strategies, and zero-sum games. 2) For the prisoner's dilemma game presented, there are three Nash equilibriums: (0,0), (2/3, 1/3), and (1,1). 3) In a zero-sum kicking game, the Nash equilibrium strategies are for the row player to kick left with probability 0.7 and the column player to jump left with probability 0.6, resulting in the row player scoring 62% of the time.

Uploaded by

Braden Muhl
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 11

Advanced game theory

Additional
notes
Best Response curve
Column
Left Right
Top 1 0
row 2 0
Bottom 0 2
0 1
Column’s choice Left Right
Row’s best choice Top Bottom
Row’s choice Top Bottom
Column’s best choice Left Right
Row choices r1…..rR / columns choices c1…..cC
bc(r) = best response for column
br(c) = best response for row
Nash equilibrium (c*,r*): c*=bc(r*)
r*=br(c*)
Example : row expects column to play left
therefore row will choose to play top
Mixed strategy
Column
Left Right
Top 1 0
row 2 0
Bottom 0 2
0 1
r = probability that row plays top
(1-r) = probability that row plays bottom
c= probability that column plays left
(1-c) probability that column plays right
Pure strategy : r & c = 0 or 1
Expected payoff of row ( row’s probability r of playing top
and columns chooses probability c of playing left )
Combination Probability Payoff to Row

Top, Left rc 2
Bottom, left (1-r)c 0
Top , right r(1-c) 0
Bottom , right (1-r)(1-c) 1
Therefore row’s payoff = 2rc+(1-r)(1-c)
Therefore =2rc+1-r-c+rc
Suppose increasing r by Δr
Therefore Δ payoff to row = 2Δrc – Δr+c Δr
= (3c-1) Δr
Therefore expression will be positive when 3c>1 : negative when 3c<1
therefore row will want to increase r whenever c>1/3, decrease r when c<1/3 and be happy
with any value of 0≤ r ≤ 1 when c=1/3
columns pay off : cr +2 (1-c)(1-r)
suppose c changes by Δc
therefore Δpayoff to column= rΔc+2rΔc-2Δc
=(3r-2)Δc
coulum will want to increase c when ever r>2/3:
decrease c whenever r<2/3 and be happy with any value of 0≤ c≤1 when r=2/3
C 1

Rows best response

1/3

columns best response

0 2/3 r

If column chooses c=0 row will want to make r as small as possible ,


so r =0 is the best response to c=0. this choice will
continue to be the best response up until c=1/3 ,
at which point any value of r between 0 and 1 is a best response
For all c>1/3 , the best response row can make is r=1
Therefore Nash equilibrium (0;0) (2/3,1/3) and(1,1)
Therefore 2 pure strategies and one mix strategy
GAMES OF
COMPETITION
Assume a zero sum game = diametrically opposed goals.

Column
Defend Left Defend right
Kick left -50 -80
row 50 80
Kick -90 -20
right 90 20

Row= more successful if column jumps the wrong way


Row follow mixed strategy : tries to keep column guessing
If row kicks left with probability p, he will get an expected payoff of
50p+90(1-p)when column jumps left and 80p+20(1-p) when
column jumps right .NB!! row wants to make this expected payoff
as big as possible and column wants to make it as small as possible
suppose column wants to kick left half the time
therefore if column jumps left row will have a
expected payoff of 50(1/2)+90(1-1/2)=70,
and if column jumps right, row will have an
expected payoff of 80(1/2)20(1-1/2)=50
Rows percent success

100

90

80

62

50

20

0 .7 1

Rows probability p of kicking left


Two equations {50p=90(1-p) and 80p+20(1-p)} are linear functions of p
therefore lines = straight. NB!! Row recognises that column will always try to
minimize his expected payoff .Thus for any probability of p the best payoff he
can hope for = the minimum payoff given by the two strategies=red
lines .maximum of these two payoffs = peak of red lines (=point of
intersection). Solve therefore p=0.7
Conclusion if row kicks to the left 70 percent of the time and column
responds optimally row will have a expected payoff of 50(0.7)+90(0.7)=62%
Columns choice : suppose column decides to jump to the left with probability
of q and jumps to the right with probability of (1-q) . rows expected payoff
will be 50q+80(1-q)if row kicks to the left and 90q+20(1-q) if row kicks to the
right .NB!! For each q column will want to minimize rows payoff but column
recognises that row wants to maximise this same payoff. Therefore if column
chooses to jump to the left with a probability of ½ she realises that row will
get a expected pay off of 50(1/2)+80(1-1/2)=65 if row kicks left
and 90(1/2)+20(1-1/2)=55 if row kicks right .
Therefore choose to kick left
Rows percent success

100
90

80

62

50
20

0 .6 1
Columns probability q of jumping left

from columns viewpoint = the maximum of the two lines that is relevant
since this reflects rows optimal choice for each choice of q.
therefore the red line .the point where rows maximum payoff is minimised
occurs at q=0.6 ( simply equate 50q+80(1-q)to 90q+20(1-q) and solve for q)
Conclusion:
Equilibrium strategies for each row should kick to the left with probability 0.7
and column should jump to the left with probability 0.6

NB!! Values such that rows payoff and columns payoff will be the same,
whatever the one player does. So ,when row chooses 0.7, column is indifferent
Between jumping left and jumping right or for that matter jumping left with a
Probability of 0.6

Also if column jumps left with a probability of 0.6, then row is indifferent
between kicking left and kicking right in particular he is happy to kick left
with a probability of 0.7

Therefore these choices are points of Nash equilibrium .


In equilibrium row scores 62% of the time and fails to score 38% of the time

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy