Potential Games and the Inefficiency of Equilibrium

Similar documents
Network Congestion Games

Equil. Properties of Reacting Gas Mixtures. So far have looked at Statistical Mechanics results for a single (pure) perfect gas

Supplementary Materials

Where k is either given or determined from the data and c is an arbitrary constant.

( ) Differential Equations. Unit-7. Exact Differential Equations: M d x + N d y = 0. Verify the condition

VTU NOTES QUESTION PAPERS NEWS RESULTS FORUMS

Economics 600: August, 2007 Dynamic Part: Problem Set 5. Problems on Differential Equations and Continuous Time Optimization

y = 2xe x + x 2 e x at (0, 3). solution: Since y is implicitly related to x we have to use implicit differentiation: 3 6y = 0 y = 1 2 x ln(b) ln(b)

[ ] 1+ lim G( s) 1+ s + s G s s G s Kacc SYSTEM PERFORMANCE. Since. Lecture 10: Steady-state Errors. Steady-state Errors. Then

cycle that does not cross any edges (including its own), then it has at least

Section 6.1. Question: 2. Let H be a subgroup of a group G. Then H operates on G by left multiplication. Describe the orbits for this operation.

Worksheet: Taylor Series, Lagrange Error Bound ilearnmath.net

An N-Component Series Repairable System with Repairman Doing Other Work and Priority in Repair

Week 3: Connected Subgraphs

Linked-List Implementation. Linked-lists for two sets. Multiple Operations. UNION Implementation. An Application of Disjoint-Set 1/9/2014

Junction Tree Algorithm 1. David Barber

16.512, Rocket Propulsion Prof. Manuel Martinez-Sanchez Lecture 6: Heat Conduction: Thermal Stresses

First derivative analysis

u x v x dx u x v x v x u x dx d u x v x u x v x dx u x v x dx Integration by Parts Formula

Final Exam : Solutions

0 t b r 6, 20 t l nf r nt f th l t th t v t f th th lv, ntr t n t th l l l nd d p rt nt th t f ttr t n th p nt t th r f l nd d tr b t n. R v n n th r

Integral Calculus What is integral calculus?

6.1 Integration by Parts and Present Value. Copyright Cengage Learning. All rights reserved.

Engineering Differential Equations Practice Final Exam Solutions Fall 2011

The Matrix Exponential

Note If the candidate believes that e x = 0 solves to x = 0 or gives an extra solution of x = 0, then withhold the final accuracy mark.

A Stackelberg Strategy for Routing Flow over Time

Random Process Part 1

22 t b r 2, 20 h r, th xp t d bl n nd t fr th b rd r t t. f r r z r t l n l th h r t rl T l t n b rd n n l h d, nd n nh rd f pp t t f r n. H v v d n f

u r du = ur+1 r + 1 du = ln u + C u sin u du = cos u + C cos u du = sin u + C sec u tan u du = sec u + C e u du = e u + C

Combinatorial Networks Week 1, March 11-12

Economics 201b Spring 2010 Solutions to Problem Set 3 John Zhu

Exchange rates in the long run (Purchasing Power Parity: PPP)

Solutions to Supplementary Problems

The Matrix Exponential

n r t d n :4 T P bl D n, l d t z d th tr t. r pd l

Optimization. Nuno Vasconcelos ECE Department, UCSD

MECH321 Dynamics of Engineering System Week 4 (Chapter 6)


Th n nt T p n n th V ll f x Th r h l l r r h nd xpl r t n rr d nt ff t b Pr f r ll N v n d r n th r 8 l t p t, n z n l n n th n rth t rn p rt n f th v

10. EXTENDING TRACTABILITY

Chapter 13 GMM for Linear Factor Models in Discount Factor form. GMM on the pricing errors gives a crosssectional

Colby College Catalogue


Colby College Catalogue

Lecture 14. Relic neutrinos Temperature at neutrino decoupling and today Effective degeneracy factor Neutrino mass limits Saha equation

Calculus II (MAC )

ELEC 372 LECTURE NOTES, WEEK 11 Dr. Amir G. Aghdam Concordia University

Differential Equations

Colby College Catalogue

Consider a system of 2 simultaneous first order linear equations

Basic Polyhedral theory

Outline. Clustering: Similarity-Based Clustering. Supervised Learning vs. Unsupervised Learning. Clustering. Applications of Clustering

CMPS 2200 Fall Graphs. Carola Wenk. Slides courtesy of Charles Leiserson with changes and additions by Carola Wenk

Summary: Solving a Homogeneous System of Two Linear First Order Equations in Two Unknowns

Background: We have discussed the PIB, HO, and the energy of the RR model. In this chapter, the H-atom, and atomic orbitals.

Partial Derivatives: Suppose that z = f(x, y) is a function of two variables.

Section 11.6: Directional Derivatives and the Gradient Vector

3) Use the average steady-state equation to determine the dose. Note that only 100 mg tablets of aminophylline are available here.

Colby College Catalogue

Theorem 1. An undirected graph is a tree if and only if there is a unique simple path between any two of its vertices.

Epistemic Foundations of Game Theory. Lecture 1

EE750 Advanced Engineering Electromagnetics Lecture 17

828.^ 2 F r, Br n, nd t h. n, v n lth h th n l nd h d n r d t n v l l n th f v r x t p th l ft. n ll n n n f lt ll th t p n nt r f d pp nt nt nd, th t

Exercises for lectures 7 Steady state, tracking and disturbance rejection

1973 AP Calculus AB: Section I

Introduction to Arithmetic Geometry Fall 2013 Lecture #20 11/14/2013

4 4 N v b r t, 20 xpr n f th ll f th p p l t n p pr d. H ndr d nd th nd f t v L th n n f th pr v n f V ln, r dn nd l r thr n nt pr n, h r th ff r d nd

Strongly Connected Components

MA1506 Tutorial 2 Solutions. Question 1. (1a) 1 ) y x. e x. 1 exp (in general, Integrating factor is. ye dx. So ) (1b) e e. e c.

TOPIC 5: INTEGRATION

2. SIMPLE SOIL PROPETIES

Problem Set 6 Solutions

Relate p and T at equilibrium between two phases. An open system where a new phase may form or a new component can be added

,. *â â > V>V. â ND * 828.

Exam 2 Solutions. Jonathan Turner 4/2/2012. CS 542 Advanced Data Structures and Algorithms

Higher order derivatives

Some Results on Interval Valued Fuzzy Neutrosophic Soft Sets ISSN

4 8 N v btr 20, 20 th r l f ff nt f l t. r t pl n f r th n tr t n f h h v lr d b n r d t, rd n t h h th t b t f l rd n t f th rld ll b n tr t d n R th

Exercise 1. Sketch the graph of the following function. (x 2

+ f. e f. Ch. 8 Inflation, Interest Rates & FX Rates. Purchasing Power Parity. Purchasing Power Parity

arxiv: v3 [cs.gt] 1 Jan 2019

Propositional Logic. Combinatorial Problem Solving (CPS) Albert Oliveras Enric Rodríguez-Carbonell. May 17, 2018

46 D b r 4, 20 : p t n f r n b P l h tr p, pl t z r f r n. nd n th t n t d f t n th tr ht r t b f l n t, nd th ff r n b ttl t th r p rf l pp n nt n th

Topics covered. ECON6021 Microeconomic Analysis. Price Effects. Price effect I

2. Laser physics - basics

AP Calculus BC Problem Drill 16: Indeterminate Forms, L Hopital s Rule, & Improper Intergals

An Application of Hardy-Littlewood Conjecture. JinHua Fei. ChangLing Company of Electronic Technology Baoji Shannxi P.R.China

where: u: input y: output x: state vector A, B, C, D are const matrices

n

Poisson process Markov process

GRAPHS IN SCIENCE. drawn correctly, the. other is not. Which. Best Fit Line # one is which?

Multi-linear Systems and Invariant Theory. in the Context of Computer Vision and Graphics. Class 4: Mutli-View 3D-from-2D. CS329 Stanford University

CPSC 665 : An Algorithmist s Toolkit Lecture 4 : 21 Jan Linear Programming

Probability and Stochastic Processes: A Friendly Introduction for Electrical and Computer Engineers Roy D. Yates and David J.

COHORT MBA. Exponential function. MATH review (part2) by Lucian Mitroiu. The LOG and EXP functions. Properties: e e. lim.

ANALYSIS IN THE FREQUENCY DOMAIN

EEO 401 Digital Signal Processing Prof. Mark Fowler

Convex Optimization. Optimality conditions. (EE227BT: UC Berkeley) Lecture 9 (Optimality; Conic duality) 9/25/14. Laurent El Ghaoui.

The University of Sydney MATH 2009

Transcription:

Optmzaton and Control o Ntork Potntal Gam and th Inny o Equlbrum Ljun Chn 3/8/216

Outln Potntal gam q Rv on tratg gam q Potntal gam atom and nonatom Inny o qulbrum q Th pr o anarhy and lh routng q Rour alloaton q Ntork dgn gam and th pr o tablty 2

Stratg gam q D: a gam n tratg orm a trpl q Notaton q q q N G { N S N u N } q th t o playr agnt q S th playr tratgy pa q u : S R th playr payo unton S S : th t o all prol o playr tratg 1 S2! S N 1 2! N : prol o tratg 1 2! 1 +1! N : th prol o tratg othr than playr 3

Potntal gam atom Dnton: A unton a potntal unton or gam or Whn t th gam alld a potntal gam. Dnton: A unton an ordnal potntal unton or gam or Whn t th gam alld an ordnal potntal gam. 4 S R Φ : G S S. u u Φ Φ. > Φ Φ > u u S R Φ : G S S Φ Φ

Eampl Gam Potntal 11 3 2 1 3 22 1 5

Equlbrum a pur tratgy Nah qulbrum or ordnal potntal gam Proo: I a potntal unton a pur tratgy Nah qulbrum 6 G. S Φ Φ Φ. Φ Φ u u u u. Φ Φ

I Φ ha a mamum at thn a pur tratgy Nah qulbrum o th ordnal gam. Evry nt ordnal potntal gam ha a pur tratgy Nah qulbrum. Contnuou ordnal potntal gam ha a pur tratgy Nah qulbrum th tratgy pa ompat and potntal ontnuou. 7

Congton gam Dnton: A ongton modl dnd a ollo q q N th t o playr M th t o alt or rour S { N M S N j M } q th t o th rour that playr an u q j k j th ot to ur ho u th rour j hn ur ar ung t Dnton: A ongton gam aoatd th a ongton modl a gam th ot k j j j { N S N N } k j 8

Evry ongton gam a potntal gam th potntal Congton gam hav many applaton q Ntork dgn Φ j j k 1 k j k. 9

Potntal gam nonatom Nonatom gam: th ur numbr nnt q N r la o nntmal playr q th ma o la playr q th raton o la playr that hoo tratgy q u ; th payo or a playr o la th Dnton: an qulbrum or all S > u Dnton: A nonatom gam a potntal gam thr t potntal unton Φ uh that u ; ; Φ. u ;. 1

Eampl: lh routng Condr a multommodty lo ntork q N our-dtnaton par ommodt V E q Eah ommodty ha a total rat and an u a t o path P q Th aggrgat tra among lnk : q lnk ot a nonngatv ontnuou nondrang unton o tra q Th ot ; r 11

Wardrop qulbrum: th ot o all th path atually ud ar qual and l than tho hh ould b prnd by a ngl ur on any unud path. { V E; r } a potntal gam th potntal Φ d. 12

Inny o qulbra Equlbra o tratg gam ar typally nnt Eampl: Pronr Dlmma D C 33 4 D C 4 11 13

Pgou ampl 1 t On ommodty th rat 1 q A unqu Wardrop qulbrum th all tra routd on th lor dg q A bttr lo: rout hal o th tra on ah o th to dg 14

Quton ar Th tn o qulbrum Quanty th nny o qulbrum Th onvrgn o qulbrum 15

Quanty th nny Pr o anarhy: quanty nny th rpt to om objtv unton pr o anarhy obj n valu o th ort qulbrum optmal obj n valu <1 or mamzaton; >1 or mnmzaton Intrtd n tuaton n hh an bound th pr o anarhy 16

Slh Routng At Wardrop qulbrum Th abov th KKT optmalty ondton or 17 mn ; ; r Φ r d :.t. mn

A lo or a Wardrop qulbrum and only t a global mnmum o th potntal unton Dn th objtv unton.. th ot o lo a Dnton: An optmal lo or th lo that mnmz. 18 } ; { r E V. Φ d C } ; { r E V C

Th pr o anarhy Th pr o anarhy ρ C C. Pgou ampl Suppo that γ y dy thn ρ γ. Pgou ampl th dgr-d polynomal ot q q ρ d d +1 + 1 ρ 4 / y dy 3 19

Rour alloaton Condr a mpl ntork: th our ur har a lnk and th ntork lnk managr ant to alloat lnk rat uh that Sytm: ma. t. tlty unton ar not knon to th lnk managr 2

Markt-larng mhanm Eah ur ubmt a bd or llngn to pay th managr k to alloat th ntr lnk apaty and t a pr p uh that q A th ur ha a dmand unton D p p p q Th lnk managr hoo a pr to lar th markt D p / 21

Pr takng ur and ompttv qulbrum Th ur a pr takr: do not antpat th t o h paymnt on th pr It ratonal or th ur to mamz th ollong payo Klly 98 A par a ompttv qulbrum 22 p p u p p p u p u / or any

Thorm Klly 98: thr t a unqu ompttv qulbrum uh that olv th problm ytm. Proo: ondr th Lagrangan q At prmal-dual optmal 23 p p / p p D ' ' > p p p p

q Sn > at lat on potv. So p >. q Thu. q Lt p thn p a ompttv qulbrum and olv th problm Sytm. q In th a th unqun o ollo rom th unqun o p. / p 24

Pr antpatng ur and Nah qulbrum Pr antpatng ur ralz that th pr t aordng to p and ll adjut thr bd / aordngly. Th mak th modl a gam hr ur payo Johar 4 > u Condr Nah qulbrum uh that u u or all or all 25

Thorm Hajk t al: thr t a unqu Nah qulbrum. Morovr th rat ar unqu oluton o th ollong problm 26 t.. ˆ ma Gam: hr. 1 1 ˆ + dz z

Proo: q I a Nah qulbrum at lat to playr hav nonzro bd. u q Thn trtly onav and ontnuouly drntabl n. q Thn at qulbrum ' ' t t 1 t q Th abov ondton alo unt. t t t t t > 27

q Th problm Gam ha a unqu optmal thr t a p uh that ' 1 p > ' p p p. Morovr q Lt / p and p / t t. Thn p at th abov optmalty ondton. 28

Th pr o anarhy Aum hav Thn Sn hav Thn 29 1 dz z ˆ z z z z + 1 2 dz z 2 1 ˆ

Lt and ar th optma o problm Sytm and Gam hav Th pr o anarhy 3 ˆ ˆ 2 1 2 1/ ρ

Tght bound Dn th JT bound by For any thr a rour alloaton gam th th pr o anarhy at mot. q Proo: rt not that an aum. q Dn a gam th q At optmal th ny q At qulbrum 31 ˆ n n n ʹ + β β > ε β + ε & < 2 ˆ 1 1 1 ʹ 1 / 1 ˆ ˆ ˆ ˆ ' 1 1 ' 1 C ʹ ʹ

q Thn a th playr numbr go to nnty. q Thu th ny at qulbrum approahng In vry rour alloaton gam th pr o anarhy at lat. q Proo: lt and ar th optmal and qulbrum 32 1. ˆ ˆ ˆ ˆ 1 C ʹ + ʹ + β + 1 ˆ ' β β

Th bound. q Proo: ttng ho th bound at mot 3/4. q Aum hav 33 4 3/ β 1 2 & 1/ & <. 4 3 / 1 / / 1 / / 1 / 1 ˆ 2 + + + ʹ + ʹ +

Ntork dgn gam Condr a ntork or ah dg E q N our-dtnaton par playr q Eah playr an hoo a path q Th total ot V E th a nonngatv ot P q Lt n dnot th numbr o playr ho path ar ung dg. Eah o tho playr pay a har π o th ot q Th ot or ah playr ; / n / n 34

{ V E; } a potntal gam th potntal unton Φ n j 1 Evry ntork dgn gam ha at lat on Nah qulbrum. j. 35

k t k 1+a playr and a > arbtrarly mall q To Nah qulbra: all hoo th uppr dg or all hoo th lor dg 36

Pr o tablty Pr o tablty pr o tablty objn valuo th bt qulbrum optmal objn valu Sn Φ 1 + 1/ 2 +! tablty at mot 1 + 1/ 2 +! + 1/ k. C + 1/ k C th pr o 37