138
Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Embed Size (px)

Citation preview

Page 1: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Algorithmic Game Theoryand Internet Computing

Vijay V. Vazirani

New Market Models

and Algorithms

Page 2: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Markets

Page 3: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Stock Markets

Page 4: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Internet

Page 5: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Revolution in definition of markets

Page 6: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Revolution in definition of markets

New markets defined byGoogle AmazonYahoo!Ebay

Page 7: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Revolution in definition of markets

Massive computational power available

for running these markets in a

centralized or distributed manner

Page 8: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Revolution in definition of markets

Massive computational power available

for running these markets in a

centralized or distributed manner

Important to find good models and

algorithms for these markets

Page 9: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Theory of Algorithms

Powerful tools and techniques

developed over last 4 decades.

Page 10: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Theory of Algorithms

Powerful tools and techniques

developed over last 4 decades.

Recent study of markets has contributed

handsomely to this theory as well!

Page 11: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Adwords Market

Created by search engine companiesGoogleYahoo!MSN

Multi-billion dollar market

Totally revolutionized advertising, especially

by small companies.

Page 12: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 13: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 14: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 15: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

New algorithmic and game-theoretic questions

Monika Henzinger, 2004: Find an on-line

algorithm that maximizes Google’s revenue.

Page 16: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

The Adwords Problem:

N advertisers; Daily Budgets B1, B2, …, BN Each advertiser provides bids for keywords he is interested

in.

Search Engine

Page 17: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

The Adwords Problem:

N advertisers; Daily Budgets B1, B2, …, BN Each advertiser provides bids for keywords he is interested

in.

Search Enginequeries (online)

Page 18: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

The Adwords Problem:

N advertisers; Daily Budgets B1, B2, …, BN Each advertiser provides bids for keywords he is interested

in.

Search EngineSelect one Ad

Advertiser pays his bid

queries (online)

Page 19: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

The Adwords Problem:

N advertisers; Daily Budgets B1, B2, …, BN Each advertiser provides bids for keywords he is interested

in.

Search EngineSelect one Ad

Advertiser pays his bid

queries (online)

Maximize total revenue

Online competitive analysis - compare with best offline allocation

Page 20: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

The Adwords Problem:

N advertisers; Daily Budgets B1, B2, …, BN Each advertiser provides bids for keywords he is interested

in.

Search EngineSelect one Ad

Advertiser pays his bid

queries (online)

Maximize total revenue

Example – Assign to highest bidder: only ½ the offline revenue

Page 21: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Example:

$1 $0.99

$1 $0

Book

CD

Bidder1 Bidder 2

B1 = B2 = $100

Queries: 100 Books then 100 CDs

Bidder 1 Bidder 2

Algorithm Greedy

LOST

Revenue100$

Page 22: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Example:

$1 $0.99

$1 $0

Book

CD

Bidder1 Bidder 2

B1 = B2 = $100

Queries: 100 Books then 100 CDs

Bidder 1 Bidder 2

Optimal Allocation

Revenue199$

Page 23: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Generalizes online bipartite matching

Each daily budget is $1, and

each bid is $0/1.

Page 24: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 25: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 26: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 27: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 28: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 29: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 30: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

advertisers queries

Page 31: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

Karp, Vazirani & Vazirani, 1990:

1-1/e factor randomized algorithm.

Page 32: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

Karp, Vazirani & Vazirani, 1990:

1-1/e factor randomized algorithm. Optimal!

Page 33: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Online bipartite matching

Karp, Vazirani & Vazirani, 1990:

1-1/e factor randomized algorithm. Optimal!

Kalyanasundaram & Pruhs, 1996:

1-1/e factor algorithm for b-matching:

Daily budgets $b, bids $0/1, b>>1

Page 34: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Adwords Problem

Mehta, Saberi, Vazirani & Vazirani, 2005:

1-1/e algorithm, assuming budgets>>bids.

Page 35: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Adwords Problem

Mehta, Saberi, Vazirani & Vazirani, 2005:

1-1/e algorithm, assuming budgets>>bids.

Optimal!

Page 36: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

New Algorithmic Technique

Idea: Use both bid and

fraction of left-over budget

Page 37: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

New Algorithmic Technique

Idea: Use both bid and

fraction of left-over budget

Correct tradeoff given by

tradeoff-revealing family of LP’s

Page 38: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Historically, the study of markets

has been of central importance,

especially in the West

Page 39: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

A Capitalistic Economy

depends crucially on pricing mechanisms,

with very little intervention, to ensure:

Stability Efficiency Fairness

Page 40: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Do markets even have inherentlystable operating points?

Page 41: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

General Equilibrium TheoryOccupied center stage in Mathematical

Economics for over a century

Do markets even have inherentlystable operating points?

Page 42: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Leon Walras, 1874

Pioneered general

equilibrium theory

Page 43: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Supply-demand curves

Page 44: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Irving Fisher, 1891

Fundamental

market model

Page 45: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Fisher’s Model, 1891

milkcheese

winebread

¢¢

$$$$$$$$$$$$$$$$$$

$$

$$$$$$$$

People want to maximize happiness – assume

linear utilities.Find prices s.t. market clears

Page 46: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Fisher’s Model n buyers, with specified money, m(i) for buyer i k goods (unit amount of each good) Linear utilities: is utility derived by i

on obtaining one unit of j Total utility of i,

i ij ijj

U u xiju

]1,0[

x

xuuij

ijj iji

Page 47: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Fisher’s Model n buyers, with specified money, m(i) k goods (each unit amount, w.l.o.g.) Linear utilities: is utility derived by i

on obtaining one unit of j Total utility of i,

Find prices s.t. market clears, i.e.,

all goods sold, all money spent.

i ij ijj

U u xiju

xuu ijj iji

Page 48: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 49: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Arrow-Debreu Theorem, 1954

Celebrated theorem in Mathematical Economics

Established existence of market equilibrium under very general conditions using a deep theorem from topology - Kakutani fixed point theorem.

Page 50: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Kenneth Arrow

Nobel Prize, 1972

Page 51: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Gerard Debreu

Nobel Prize, 1983

Page 52: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Arrow-Debreu Theorem, 1954

.

Highly non-constructive

Page 53: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Adam Smith

The Wealth of Nations

2 volumes, 1776.

‘invisible hand’ of the market

Page 54: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

What is needed today?

An inherently algorithmic theory of

market equilibrium

New models that capture new markets

Page 55: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Beginnings of such a theory, within

Algorithmic Game Theory

Started with combinatorial algorithms

for traditional market models

New market models emerging

Page 56: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Combinatorial Algorithm for Fisher’s Model

Devanur, Papadimitriou, Saberi & V., 2002

Using primal-dual schema

Page 57: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Primal-Dual Schema

Highly successful algorithm design

technique from exact and

approximation algorithms

Page 58: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Exact Algorithms for Cornerstone Problems in P:

Matching (general graph) Network flow Shortest paths Minimum spanning tree Minimum branching

Page 59: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Approximation Algorithms

set cover facility location

Steiner tree k-median

Steiner network multicut

k-MST feedback vertex set

scheduling . . .

Page 60: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

No LP’s known for capturing equilibrium allocations for Fisher’s model

Eisenberg-Gale convex program, 1959

DPSV: Extended primal-dual schema to

solving nonlinear convex programs

Page 61: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

2s

1s

2t

1t

A combinatorial market

Page 62: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

2s

1s

2t

1t

A combinatorial market

)(ec

Page 63: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

2s

1s

2t

1t

A combinatorial market

)1(m

)2(m

)(ec

Page 64: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

A combinatorial market

Given: Network G = (V,E) (directed or undirected)Capacities on edges c(e)Agents: source-sink pairs

with money m(1), … m(k)

Find: equilibrium flows and edge prices

1 1( , ),...( , )k ks t s t

Page 65: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Flows and edge prices

f(i): flow of agent i p(e): price/unit flow of edge e

Satisfying: p(e)>0 only if e is saturated flows go on cheapest paths money of each agent is fully spent

Equilibrium

Page 66: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Kelly’s resource allocation model, 1997

Mathematical framework for understanding

TCP congestion control

Highly successful theory

Page 67: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

TCP Congestion Control

f(i): source rate prob. of packet loss (in TCP Reno)

queueing delay (in TCP Vegas) p(e):

Page 68: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

TCP Congestion Control

f(i): source rate prob. of packet loss (in TCP Reno)

queueing delay (in TCP Vegas)

Kelly: Equilibrium flows are proportionally fair:

only way of adding 5% flow to someone’s

dollar is to decrease 5% flow from

someone else’s dollar.

p(e):

Page 69: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

primal process: packet rates at sources

dual process: packet drop at links

AIMD + RED converges to equilibrium

in the limit

TCP Congestion Control

Page 70: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Kelly & V., 2002: Kelly’s model is a

generalization of Fisher’s model.

Find combinatorial polynomial time

algorithms!

Page 71: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Jain & V., 2005:

Strongly polynomial combinatorial algorithm

for single-source multiple-sink market

Page 72: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Single-source multiple-sink market

Given: Network G = (V,E), s: sourceCapacities on edges c(e)Agents: sinks

with money m(1), … m(k)

Find: equilibrium flows and edge prices

1,..., kt t

Page 73: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Flows and edge prices

f(i): flow of agent i p(e): price/unit flow of edge e

Satisfying: p(e)>0 only if e is saturated flows go on cheapest paths money of each agent is fully spent

Equilibrium

Page 74: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

2

2

110$

10$

Page 75: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

2

2

1 10$

10$

$5

$5

Page 76: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

2

2

1 10$

10$

120$

Page 77: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

2

2

1 120$

10$

$10

$40

$30

Page 78: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Jain & V., 2005:

Strongly polynomial combinatorial algorithm

for single-source multiple-sink market

Ascending price auctionBuyers: sinks (fixed budgets, maximize flow)Sellers: edges (maximize price)

Page 79: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Auction of k identical goods

p = 0; while there are >k buyers:

raise p; end; sell to remaining k buyers at price p;

Page 80: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

Find equilibrium prices and flows

Page 81: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

Find equilibrium prices and flows

m(1)

m(2)

m(3)m(4)

cap(e)

Page 82: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

min-cut separating from all the sinkss

60

Page 83: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

p

60

Page 84: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

p

60

Page 85: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Throughout the algorithm:

s itc(i): cost of cheapest path from to

sink demands flow ( )

( )( )

m if i

c iit

Page 86: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

p

: ( )i c i p

60

sink demands flow ( )

( )m i

f ip

it

Page 87: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Auction of edges in cut

p = 0; while the cut is over-saturated:

raise p; end; assign price p to all edges in the cut;

Page 88: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

pp 0

0(2)c p

60 50 (2) 10f

Page 89: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

p0

p

0(2)c p

60 50 0(1) (3) (4)c c c p p

Page 90: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

p0

p1

60 50 20

0(2)c p

0 1(1) (3)c c p p

(1) (3) 30f f

Page 91: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

p0

p1

p

60 50 20

Page 92: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

p0

p1 p

2

60 50 200 1 2(4)c p p p

(4) 20f

Page 93: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

p0

p1 p

2 nested cuts

60 50 20

Page 94: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Flow and prices will:

Saturate all red cutsUse up sinks’ moneySend flow on cheapest paths

Page 95: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

Implementation

Page 96: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

Page 97: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

Capacity of edge =tt i

( )( )

( )

m if i

c i

Page 98: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

min s-t cut

60

Page 99: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

p

60

Page 100: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

p

60

Page 101: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2

t3

t4

t

p tt i

Capacity of edge =

( )( )

m if i

p

: ( )i c i p

Page 102: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

t

pp 0

0(2)c p

60 50

f(2)=10

Page 103: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

t

p0

p

60 50

Page 104: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

t

p0

p1

0(2)c p

0 1(1) (3) (4)c c c p p

60 50 20

Page 105: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

t

p0

p1

p

Page 106: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

s

t1

t2 t3

t4

t

p0

p1 p

2

0 1 2(4)c p p p

Page 107: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Eisenberg-Gale Program, 1959

max ( ) log

. .

:

: 1

: 0

ii

i ij ijj

iji

ij

m i u

s t

i u

j

ij

u xx

x

Page 108: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Lagrangian variables: prices of goods

Using KKT conditions:

optimal primal and dual solutions

are in equilibrium

Page 109: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Convex Program for Kelly’s Model

max ( ) log ( )

. .

: ( )

: ( ) ( )

, : 0

i

pip

pi

m i f i

s t

i f i f

e flow e c e

i p f

Page 110: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

JV Algorithm

primal-dual alg. for nonlinear convex program

“primal” variables: flows

“dual” variables: prices of edges

algorithm: primal & dual improvements

Allocations Prices

Page 111: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Rational!!

Page 112: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Irrational for 2 sources & 3 sinks

s1 t1

1

s2

t2

1

t21 2

$1 $1

$1

Page 113: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Irrational for 2 sources & 3 sinks

s1 t1

1

s2

t2

1

t2

31

3

3

Equilibrium prices

Page 114: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Max-flow min-cut theorem!

Page 115: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Other resource allocation markets

2 source-sink pairs (directed/undirected) Branchings rooted at sources (agents)

Page 116: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Branching market (for broadcasting)

s1 s2

)1(m3s

)2(m)(ec

(3)m

Page 117: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Branching market (for broadcasting)

s1 s2

)1(m3s

)2(m)(ec

(3)m

Page 118: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Branching market (for broadcasting)

s1 s2

)1(m3s

)2(m)(ec

(3)m

Page 119: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Branching market (for broadcasting)

s1 s2

)1(m3s

)2(m)(ec

(3)m

Page 120: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Branching market (for broadcasting)

Given: Network G = (V, E), directed edge capacities sources, money of each source

Find: edge prices and a packing

of branchings rooted at sources s.t. p(e) > 0 => e is saturated each branching is cheapest possible money of each source fully used.

S V

Page 121: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Eisenberg-Gale-type program for branching market

max ( ) log ii Sm i b

s.t. packing of branchings

Page 122: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Other resource allocation markets

2 source-sink pairs (directed/undirected) Branchings rooted at sources (agents) Spanning trees Network coding

Page 123: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Eisenberg-Gale-Type Convex Program

max ( ) log iim i u

s.t. packing constraints

Page 124: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Eisenberg-Gale Market

A market whose equilibrium is captured

as an optimal solution to an

Eisenberg-Gale-type program

Page 125: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Theorem: Strongly polynomial algs for

following markets :2 source-sink pairs, undirected (Hu, 1963)spanning tree (Nash-William & Tutte, 1961)2 sources branching (Edmonds, 1967 + JV, 2005)

3 sources branching: irrational

Page 126: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Theorem: Strongly polynomial algs for

following markets :2 source-sink pairs, undirected (Hu, 1963)spanning tree (Nash-William & Tutte, 1961)2 sources branching (Edmonds, 1967 + JV, 2005)

3 sources branching: irrational

Open: (no max-min theorems):2 source-sink pairs, directed2 sources, network coding

Page 127: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

EG[2]: Eisenberg-Gale markets with 2 agents

Theorem: EG[2] markets are rational.

Chakrabarty, Devanur & V., 2006:

Page 128: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

EG[2]: Eisenberg-Gale markets with 2 agents

Theorem: EG[2] markets are rational.

Combinatorial EG[2] markets: polytope

of feasible utilities can be described via

combinatorial LP.

Theorem: Strongly poly alg for Comb EG[2].

Chakrabarty, Devanur & V., 2006:

Page 129: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

EG

Rational

Comb EG[2]

SUA

EG[2]

3-source branching

Fisher

2 s-s undir

2 s-s dir

Single-source

Page 130: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Efficiency of Markets

‘‘price of capitalism’’ Agents:

different abilities to control prices idiosyncratic ways of utilizing resources

Q: Overall output of market when forced

to operate at equilibrium?

Page 131: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Efficiency

( )( ) min

max ( )I

equilibrium utility Ieff M

utility I

Page 132: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Efficiency

Rich classification!

( )( ) min

max ( )I

equilibrium utility Ieff M

utility I

Page 133: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

1/(2 1)k

Market EfficiencySingle-source 1

3-source branching

k source-sink undirected

2 source-sink directed arbitrarily

small

1/ 2

. . 1/( 1)l b k

Page 134: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Other properties:

Fairness (max-min + min-max fair) Competition monotonicity

Page 135: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms

Open issues

Strongly poly algs for approximatingnonlinear convex programsequilibria

Insights into congestion control protocols?

Page 136: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 137: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms
Page 138: Algorithmic Game Theory and Internet Computing Vijay V. Vazirani New Market Models and Algorithms