Utility Design for Distributed Engineering Systems

Size: px
Start display at page:

Download "Utility Design for Distributed Engineering Systems"

Transcription

1 Utility Design for Distributed Engineering Systems Jason R. Marden University of Colorado at Boulder (joint work with Na Li, Caltech) University of Oxford June 7, 2011

2 Cooperative control Goal: Derive desirable collective behavior through local control algorithms Sensor coverage Appeal Local processing (manageable) Reduces communication Robustness Network Coding Wind farms Challenges Characterization Coordination Efficiency Trend: Game theory popular tool for cooperative control

3 Game theory social system Descriptive Agenda: Modeling model as game Decision Makers game theory Global Behavior Metrics: Reasonable description of sociocultural phenomena? Matches available experimental/observational data?

4 Game theory social system Prescriptive Agenda: Distributed robust optimization engineering system model as game Decision Makers game theory Global Behavior desired global behavior Metrics: distributed control Design parameters: Asymptotic global behavior? Communication/Information requirement? Computation requirement? Convergence rates? Decision makers Objective/Utility functions Decision/Learning rule

5 Why game theory? Reason #1: Constrained distributed optimization Global objective: Agent control policies: G( ) π i ( ) Goal: Design admissible control policies Wind farms Conventional distributed optimization: G( ) π 1 ( )... π n ( ) Structure of policies a byproduct of structure of global objective

6 Why game theory? Reason #1: Constrained distributed optimization Global objective: Agent control policies: G( ) π i ( ) Goal: Design admissible control policies Wind farms Game theoretic distributed optimization: U 1 ( ) π 1 ( ) Policies now depend on utility functions G( ) heterogeneity admissibility U n ( ) π n ( ) heterogeneity = game theory

7 Why game theory? Reason #1: Constrained distributed optimization Global objective: Agent control policies: G( ) π i ( ) Goal: Design admissible control policies Wind farms Reason #2: Hierarchical decomposition between game design and learning design Game design game structure Learning design Ex: Potential games Modularization of design Wide array of existing learning algorithms Robustness to decision rules [Young, 2005] [Gopalakrishnan, JRM, and Wierman, 2010]

8 Game theoretic control Setup Dynamics Model interactions as game decision makers / players possible choices local objective functions Potential games Local agent decision rules informational dependencies processing requirements Architecture common to many designs [Zhu and Martinez, 2009] [Kaumann et al. 2007] [Marden et al. 2007, 2008] [Mhatre et al. 2007] [Komali and MacKenzie 2007] [Zou and Chakrabarty 2004] [Campos-Nanez 2008] and many others

9 Game theoretic control Setup Model interactions as game decision makers / players possible choices local objective functions Desirable Properties Existence of (pure) NE Efficiency of NE Locality of information Potential game? Potential games GOAL Emergent global behavior desirable Dynamics Local agent decision rules informational dependencies processing requirements Desirable Properties Asymptotic behavior Limited information Fast convergence Equilibrium selection...

10 Game theoretic control Setup Model interactions as game decision makers / players possible choices local objective functions Desirable Properties Existence of (pure) NE Efficiency of NE Locality of information Potential game? Potential games GOAL Emergent global behavior desirable Dynamics Local agent decision rules informational dependencies processing requirements Learning in games [Young, ] [Marden et al., ] [Leslie et al., ] [Shah and Shin,2009] [Montanari and Saberi, 2009] and many others

11 Game theoretic control Setup Model interactions as game decision makers / players possible choices local objective functions Limited work [Wolpert and Tumer, 1999] [Arslan et al., 2007] [Marden and Wierman, 2009,10] many unresolved questions Potential games GOAL Emergent global behavior desirable Dynamics Local agent decision rules informational dependencies processing requirements Learning in games [Young, ] [Marden et al., ] [Leslie et al., ] [Shah and Shin,2009] [Montanari and Saberi, 2009] and many others

12 Game design game structure Learning design What is the viability of normal form games as a mediating layer for this decomposition? Goal: Develop underlying theory for game design to meet specifications: Constrained utility functions Efficiency of NE with respect to system level objective Game possesses desirable structure

13 Game design game structure Learning design What is the viability of normal form games as a mediating layer for this decomposition? Goal: Develop underlying theory for game design to meet specifications: Constrained utility functions Efficiency of NE with respect to system level objective Game possesses desirable structure Limitations: Normal form games not rich enough to meet objectives Coupled constraints in system level objective [N. Li and JRM, 2011] NE + Local + BB = computationally prohibitive [JRM and A. Wierman, 2011] NE + Local + BB = price of stability < 1 [JRM and A. Wierman, 2011]

14 Game design game structure Learning design What is the viability of normal form games as a mediating layer for this decomposition? Goal: Develop underlying theory for game design to meet specifications: Constrained utility functions Efficiency of NE with respect to system level objective Game possesses desirable structure Limitations: Normal form games not rich enough to meet objectives Coupled constraints in system level objective [N. Li and JRM, 2011] NE + Local + BB = computationally prohibitive [JRM and A. Wierman, 2011] NE + Local + BB = price of stability < 1 [JRM and A. Wierman, 2011]

15 Example: Consensus Global objective: Reach consensus on average of initial values using admissible controllers Setup: Player set: Information set: Value set: Initial values: N N i N V i v i (0) [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

16 Example: Consensus System level objective v i v j i N min v V Admissible controllers v i (t) =Π i Info about j at time t s.t. i N v i = i N v i (0) [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

17 Example: Consensus System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t coupled constraint [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

18 Example: Consensus System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t coupled constraint locality constraint [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

19 Example: Consensus System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t coupled constraint locality constraint Game theoretic goal Define local and scalable cost functions such that all NE minimize system level objective while satisfying the coupled constraint irrespective of initial setup. U i (v i,v i )=F {v j,v j (0)} j Ni [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

20 Example: Consensus System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t coupled constraint locality constraint Game theoretic goal Define local and scalable cost functions such that all NE minimize system level objective while satisfying the coupled constraint irrespective of initial setup. U i (v i,v i )=F {v j,v j (0)} j Ni locality + scalability constraint embedded into cost functions [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

21 Example: Consensus System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t coupled constraint locality constraint locality + scalability constraint embedded into cost functions Game theoretic goal Define local and scalable cost functions such that all NE minimize system level objective while satisfying the coupled constraint irrespective of initial setup. U i (v i,v i )=F {v j,v j (0)} j Ni IMPOSSIBLE [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

22 Impossibility Setting #1: Setting #2: v i (0) = 0 v i (0) = 1 v i (0) = 0 v i (0) = 1 [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

23 Impossibility Setting #1: Setting #2: v i (0) = 0 v i (0) = 1 v i (0) = 0 v i (0) = 1 U i (v i,v i )=F FIXED {v j,v j (0)} j Ni [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

24 Impossibility Setting #1: Setting #2: v i (0) = 0 v i (0) = 1 v i (0) = 0 v i (0) = 1 U i (v i,v i )=F FIXED {v j,v j (0)} j Ni Problem: (v, v, v, v) (v, v, v, v, v) is a NE setting #1 is a NE setting #2 [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

25 Impossibility Setting #1: Setting #2: v i (0) = 0 v i (0) = 1 v i (0) = 0 v i (0) = 1 U i (v i,v i )=F FIXED {v j,v j (0)} j Ni Problem: (v, v, v, v) (v, v, v, v, v) is a NE setting #1 is a NE setting #2 [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

26 Goal: Develop underlying theory for utility design Goal: Develop underlying theory for utility design This talk Social systems: Normal form games is natural choice Engineering systems: Normal form games is one possible design choice Limitations: Provide analytical justification for moving beyond normal form games Game design game structure This talk: introduces state into game environment additional design freedom Learning design introduces state into learning environment [Young, 2010] [Pradelski & Young, 2011] [Arslan & Shamma, 2007] [JRM, State based potential games, 2011]

27 Goal: Develop underlying theory for utility design Goal: Develop underlying theory for utility design This talk Part 1: Formally introduce state base games as a new mediating layer Normal form game {1,...,n} A i U i : A R Players Actions Utilities [JRM, State based potential games, 2011]

28 Goal: Develop underlying theory for utility design Goal: Develop underlying theory for utility design This talk Part 1: Formally introduce state base games as a new mediating layer Normal form game {1,...,n} A i U i : A R Players Actions Utilities States State Transition State based game {1,...,n} A i U i : X A R X P : X A X [JRM, State based potential games, 2011]

29 Goal: Develop underlying theory for utility design Goal: Develop underlying theory for utility design This talk Part 1: Formally introduce state base games as a new mediating layer Myopic players Static equilibrium concepts Potential game extension Players Actions Utilities States State Transition State based game {1,...,n} A i U i : X A R X P : X A X [JRM, State based potential games, 2011]

30 Goal: Develop underlying theory for utility design Goal: Develop underlying theory for utility design This talk Part 1: Formally introduce state base games as a new mediating layer Myopic players Static equilibrium concepts Potential game extension Players Actions Utilities States State Transition State based game {1,...,n} A i U i : X A R X P : X A X Part 2: Develop theory for utility design in state based games Local utility design System level objectives with coupled constraints [JRM, State based potential games, 2011]

31 State based games Two simplifications: Myopic players Static equilibrium concepts Repeated play: State at time t: x(t) Each player myopically updates action: a i (t) =F i x(t), {x(τ),a(τ)} τ=0,1,...,t 1 State based game {1,...,n} A i U i : X A R X P : X A X a i (t) arg max a i A i U i (x(t),a i,a i (t 1)) myopic Cournot adjustment process One-shot payoff: State at time t+1: U i (x(t),a(t)) x(t + 1) = f(x(t),a(t)) [JRM, State based potential games, 2011]

32 Recurrent state equilibrium Action invariant state trajectory: X(x 0,a 0 )={x 1,x 2,x 3,...} x k+1 = P (x k,a 0 ) Definition: A state action pair [x 0,a 0 ] is a recurrent state equilibrium if U i (x, a 0 ) = max U i (x, a i,a 0 a i A i i) for all x X(x 0,a 0 ) x 0 X(x, a 0 ) for all x X(x 0,a 0 ) Recurrent state equilibrium fixed point of myopic Cournot adjustment process [JRM, State based potential games, 2011]

33 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] [JRM, State based potential games, 2011]

34 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] USUAL CONDITION alignment with potential function for unilateral deviations [JRM, State based potential games, 2011]

35 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] NEW CONDITION potential function nondecreasing along action invariant state trajectory USUAL CONDITION alignment with potential function for unilateral deviations [JRM, State based potential games, 2011]

36 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] NEW CONDITION potential function nondecreasing along action invariant state trajectory USUAL CONDITION alignment with potential function for unilateral deviations Fact: A recurrent state equilibrium exists in any state based potential game [JRM, State based potential games, 2011]

37 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] NEW CONDITION potential function nondecreasing along action invariant state trajectory USUAL CONDITION alignment with potential function for unilateral deviations Fact: A recurrent state equilibrium exists in any state based potential game Fact: Many learning algorithms for potential games extend to SBPG Gradient play Log-linear learning Finite memory better reply process extend to state based potential games [JRM, State based potential games, 2011]

38 State based games Definition: A state based game is a state based potential game if there exists a potential function such that for any state action pair φ : X A R U i (x, a i,a i ) U i (x, a i,a i )=φ(x, a i,a i ) φ(x, a i,a i ) φ(x,a) φ(x, a) x = f(x, a) for [x, a] NEW CONDITION potential function nondecreasing along action invariant state trajectory USUAL CONDITION alignment with potential function for unilateral deviations Fact: A recurrent state equilibrium exists in any state based potential game Fact: Many learning algorithms for potential games extend to SBPG Gradient play Log-linear learning Finite memory better reply process ROBUSTNESS extend to state based potential games [JRM, State based potential games, 2011]

39 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions [Na Li and JRM, Designing games for distributed optimization, 2011]

40 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions flavor of design introduce state as estimation parameter State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

41 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions flavor of design introduce state as estimation parameter State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

42 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

43 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions x =(x 1,...,x n ) x i =(v i,e 1 i,e 2 i,...,e n i ) local state variables State based game {1,...,n} X v i e k i actual value agent i agent i s estimate of agent k s value A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

44 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

45 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions a i = ˆv i ê k i j any player ˆv i, ê k i j,k N change in value any neighbor change in estimate of agent k s value State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

46 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

47 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions x =(v, e) a =(ˆv, ê) local value update ṽ i = v i +ˆv i local estimate update ẽ k i = e k i ê k i j + passed to neighbors ê k j i passed from neighbors k = i State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

48 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions x =(v, e) a =(ˆv, ê) local value update ṽ i = v i +ˆv i local estimate update ẽ k i = e k i ê k i j + ẽ i i = e i i + nˆv i ê k j i ê i i j + k = i ê i j i State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

49 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions x =(v, e) a =(ˆv, ê) local value update ṽ i = v i +ˆv i local estimate update ẽ k i = e k i ê k i j + ẽ i i = e i i + nˆv i ê k j i ê i i j + k = i ê i j i conservation Initial estimation terms satisfy e k i (0) = n v k (0) i N State dynamics preserve e k i (t) =n v k (t) i N [Na Li and JRM, Designing games for distributed optimization, 2011]

50 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions x =(v, e) a =(ˆv, ê) local value update ṽ i = v i +ˆv i local estimate update ẽ k i = e k i ê k i j + ẽ i i = e i i + nˆv i ê k j i ê i i j + k = i ê i j i conservation Initial estimation terms satisfy e k i (0) = n v k (0) i N State dynamics preserve e k i (t) =n v k (t) i N [Na Li and JRM, Designing games for distributed optimization, 2011]

51 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

52 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions J i (x, a) =J G i (x, a)+j e i (x, a) J G i (x, a) = G J e i (x, a) = k N ẽ1 j,...,ẽ n j ẽk i ẽ k j ] 2 State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Designing games for distributed optimization, 2011]

53 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions Theorem: The game is a state based potential game with potential function φ(x, a) = G ẽ 1 i,...,ẽ n ẽk i + i ẽ k 2 j i N where x =(ṽ, ẽ) =P (x, a) i N k N [Na Li and JRM, Designing games for distributed optimization, 2011]

54 Given: System level objective Desired interaction graph min G(v) v V N i N Goal: Design local cost functions J i : Recurrent state equilibrium (convex, continuously differentiable) (undirected, connected) X j A j R such that Optimal solutions Theorem: An action state pair [x,a] is a recurrent state equilibrium if and only if Value profile v =(v 1,...,v n ) is optimal Estimation is accurate e k i = v k Change in value profile satisfies Change in estimation profile satisfies ˆv i =0 êk i j ê k j i =0 [Na Li and JRM, Designing games for distributed optimization, 2011]

55 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

56 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions flavor of design use penalty functions to relax constraints min v V G(v)+µ α(v) α(v) = k M max 0, i N µ>0 tradeoff parameter A k i v i C k State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

57 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions x =(x 1,...,x n ) v i e k i local state variables x i =(v i,e 1 i,...,e n i,c 1 i,...,c m i ) actual value agent i agent i s estimate of agent k s value c i C Av agent i s estimate of constraints State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

58 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

59 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions J i (x, a) =J G i (x, a)+j e i (x, a)+µ J c i (x, a) J c i (x, a) = i N same as before max 0, c k 2 i k M State based game {1,...,n} X A i P : X A X J i : X A R [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

60 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions Theorem: The game is a state based potential game with potential function max 0, c k i 2 X φ(x, a) =φ prev (x, a)+µ i N where (ṽ, ẽ, c) =P (x, a) k M A i [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

61 Given: System level objective Desired interaction graph m linear coupled constraint Goal: Design local cost functions Recurrent state equilibrium min G(v) v V N i N Av C (convex, continuously differentiable) (undirected, connected) J i : X j A j R such that Optimal solutions Theorem: An action state pair [x,a] is a recurrent state equilibrium if and only if Value profile Estimation is accurate v =(v 1,...,v n ) Change in value profile satisfies Change in estimation profile satisfies optimizes ˆv i =0 G(v)+µ α(v) e k i = v k, c i = max(0, Av C) êk i j ê k j i =0 [Na Li and JRM, Decoupling coupled constraints through utility design, 2011]

62 Example: Illustration Global objective: Reach consensus on average of initial values using admissible controllers Setup: Player set: Information set: Value set: Initial values: N N i N V i v i (0) System level objective v i v j i N min v V s.t. i N v i = i N v i (0) Admissible controllers v i (t) =Π i Info about j at time t

63 Global objective: Example: Illustration Reach consensus on average of initial values using admissible controllers Approach: Model as state based game Use algorithm gradient play cost value Φ (l) φ(x, Φ a) (b) Φ (x) value optimization constraint estimation total time step, t max 0, ẽ k i 2 φ(x, a) =φ prev (x, a)+µ i N k M

64 Conclusion Setup Model interactions as game decision makers / players possible choices local objective functions State based potential? games localizing utility functions localizing coupled constraints improving efficiency? improving scalability? exploit decomposition for MAS Dynamics Local agent decision rules informational dependencies processing requirements Finite memory better response log-linear learning gradient play improving convergence rates?

Decoupling Coupled Constraints Through Utility Design

Decoupling Coupled Constraints Through Utility Design 1 Decoupling Coupled Constraints Through Utility Design Na Li and Jason R. Marden Abstract The central goal in multiagent systems is to design local control laws for the individual agents to ensure that

More information

Designing Games to Handle Coupled Constraints

Designing Games to Handle Coupled Constraints Designing Games to Handle Coupled Constraints Na Li and Jason R. Marden Abstract The central goal in multiagent systems is to design local control laws for the individual agents to ensure that the emergent

More information

Designing Games for Distributed Optimization

Designing Games for Distributed Optimization Designing Games for Distributed Optimization Na Li and Jason R. Marden Abstract The central goal in multiagent systems is to design local control laws for the individual agents to ensure that the emergent

More information

Game Theoretic Learning in Distributed Control

Game Theoretic Learning in Distributed Control Game Theoretic Learning in Distributed Control Jason R. Marden Jeff S. Shamma November 1, 2016 May 11, 2017 (revised) Abstract In distributed architecture control problems, there is a collection of interconnected

More information

Achieving Pareto Optimality Through Distributed Learning

Achieving Pareto Optimality Through Distributed Learning 1 Achieving Pareto Optimality Through Distributed Learning Jason R. Marden, H. Peyton Young, and Lucy Y. Pao Abstract We propose a simple payoff-based learning rule that is completely decentralized, and

More information

Achieving Pareto Optimality Through Distributed Learning

Achieving Pareto Optimality Through Distributed Learning 1 Achieving Pareto Optimality Through Distributed Learning Jason R. Marden, H. Peyton Young, and Lucy Y. Pao Abstract We propose a simple payoff-based learning rule that is completely decentralized, and

More information

Overcoming the Limitations of Utility Design for Multiagent Systems

Overcoming the Limitations of Utility Design for Multiagent Systems 1 Overcoming the Limitations of Utility Design for Multiagent Systems Jason R. Marden Adam Wierman Abstract Cooperative control focuses on deriving desirable collective behavior in multiagent systems through

More information

Achieving Pareto Optimality Through Distributed Learning

Achieving Pareto Optimality Through Distributed Learning 1 Achieving Pareto Optimality Through Distributed Learning Jason R. Marden, H. Peyton Young, and Lucy Y. Pao Abstract We propose a simple payoff-based learning rule that is completely decentralized, and

More information

Connections Between Cooperative Control and Potential Games Illustrated on the Consensus Problem

Connections Between Cooperative Control and Potential Games Illustrated on the Consensus Problem Proceedings of the European Control Conference 2007 Kos, Greece, July 2-5, 2007 Connections Between Cooperative Control and Potential Games Illustrated on the Consensus Problem Jason R. Marden, Gürdal

More information

Understanding the Influence of Adversaries in Distributed Systems

Understanding the Influence of Adversaries in Distributed Systems Understanding the Influence of Adversaries in Distributed Systems Holly P. Borowski and Jason R. Marden Abstract Transitioning from a centralized to a distributed decision-making strategy can create vulnerability

More information

Understanding the Influence of Adversaries in Distributed Systems

Understanding the Influence of Adversaries in Distributed Systems Understanding the Influence of Adversaries in Distributed Systems Holly P. Borowski and Jason R. Marden Abstract Transitioning from a centralized to a distributed decision-making strategy can create vulnerability

More information

Coevolutionary Modeling in Networks 1/39

Coevolutionary Modeling in Networks 1/39 Coevolutionary Modeling in Networks Jeff S. Shamma joint work with Ibrahim Al-Shyoukh & Georgios Chasparis & IMA Workshop on Analysis and Control of Network Dynamics October 19 23, 2015 Jeff S. Shamma

More information

Achieving Pareto Optimality Through Distributed Learning

Achieving Pareto Optimality Through Distributed Learning 1 Achieving Pareto Optimality Through Distributed Learning Jason R. Marden, H. Peyton Young, and Lucy Y. Pao Abstract We propose a simple payoff-based learning rule that is completely decentralized, and

More information

Achieving Pareto Optimality Through Distributed Learning

Achieving Pareto Optimality Through Distributed Learning 1 Achieving Pareto Optimality Through Distributed Learning Jason R. Marden, H. Peyton Young, and Lucy Y. Pao Abstract We propose a simple payoff-based learning rule that is completely decentralized, and

More information

Generalized Efficiency Bounds In Distributed Resource Allocation

Generalized Efficiency Bounds In Distributed Resource Allocation 1 Generalized Efficiency Bounds In Distributed Resource Allocation Jason R. Marden Tim Roughgarden Abstract Game theory is emerging as a popular tool for distributed control of multiagent systems. To take

More information

Near-Potential Games: Geometry and Dynamics

Near-Potential Games: Geometry and Dynamics Near-Potential Games: Geometry and Dynamics Ozan Candogan, Asuman Ozdaglar and Pablo A. Parrilo January 29, 2012 Abstract Potential games are a special class of games for which many adaptive user dynamics

More information

Near-Potential Games: Geometry and Dynamics

Near-Potential Games: Geometry and Dynamics Near-Potential Games: Geometry and Dynamics Ozan Candogan, Asuman Ozdaglar and Pablo A. Parrilo September 6, 2011 Abstract Potential games are a special class of games for which many adaptive user dynamics

More information

1 Equilibrium Comparisons

1 Equilibrium Comparisons CS/SS 241a Assignment 3 Guru: Jason Marden Assigned: 1/31/08 Due: 2/14/08 2:30pm We encourage you to discuss these problems with others, but you need to write up the actual homework alone. At the top of

More information

Selecting Efficient Correlated Equilibria Through Distributed Learning. Jason R. Marden

Selecting Efficient Correlated Equilibria Through Distributed Learning. Jason R. Marden 1 Selecting Efficient Correlated Equilibria Through Distributed Learning Jason R. Marden Abstract A learning rule is completely uncoupled if each player s behavior is conditioned only on his own realized

More information

Learning Approaches to the Witsenhausen Counterexample From a View of Potential Games

Learning Approaches to the Witsenhausen Counterexample From a View of Potential Games Learning Approaches to the Witsenhausen Counterexample From a View of Potential Games Na Li, Jason R. Marden and Jeff S. Shamma Abstract Since Witsenhausen put forward his remarkable counterexample in

More information

Learning in Linear Games over Networks

Learning in Linear Games over Networks Fiftieth Annual Allerton Conference Allerton House, UIUC, Illinois, USA October -, 0 Learning in Linear Games over Networks Ceyhun Eksin, Pooya Molavi, Alejandro Ribeiro, and Ali Jadbabaie Dept. of Electrical

More information

Game Theory and Control

Game Theory and Control Game Theory and Control Lecture 4: Potential games Saverio Bolognani, Ashish Hota, Maryam Kamgarpour Automatic Control Laboratory ETH Zürich 1 / 40 Course Outline 1 Introduction 22.02 Lecture 1: Introduction

More information

Influencing Social Evolutionary Dynamics

Influencing Social Evolutionary Dynamics Influencing Social Evolutionary Dynamics Jeff S Shamma Georgia Institute of Technology MURI Kickoff February 13, 2013 Influence in social networks Motivating scenarios: Competing for customers Influencing

More information

Network Games: Learning and Dynamics

Network Games: Learning and Dynamics Network Games: Learning and Dynamics Asu Ozdaglar Conference on Decision and Control (CDC) December 2008 Department of Electrical Engineering & Computer Science Massachusetts Institute of Technology, USA

More information

Security Against Impersonation Attacks in Distributed Systems

Security Against Impersonation Attacks in Distributed Systems 1 Security Against Impersonation Attacks in Distributed Systems Philip N. Brown, Holly P. Borowski, and Jason R. Marden Abstract In a multi-agent system, transitioning from a centralized to a distributed

More information

Performance Tradeoffs in Distributed Control Systems

Performance Tradeoffs in Distributed Control Systems University of Colorado, Boulder CU Scholar Aerospace Engineering Sciences Graduate Theses & Dissertations Aerospace Engineering Sciences Spring 1-1-2016 Performance Tradeoffs in Distributed Control Systems

More information

Distributed Receding Horizon Control of Cost Coupled Systems

Distributed Receding Horizon Control of Cost Coupled Systems Distributed Receding Horizon Control of Cost Coupled Systems William B. Dunbar Abstract This paper considers the problem of distributed control of dynamically decoupled systems that are subject to decoupled

More information

Quantized Average Consensus on Gossip Digraphs

Quantized Average Consensus on Gossip Digraphs Quantized Average Consensus on Gossip Digraphs Hideaki Ishii Tokyo Institute of Technology Joint work with Kai Cai Workshop on Uncertain Dynamical Systems Udine, Italy August 25th, 2011 Multi-Agent Consensus

More information

Fast Convergence in Semi-Anonymous Potential Games

Fast Convergence in Semi-Anonymous Potential Games Fast Convergence in Semi-Anonymous Potential Games Holly Borowski, Jason R. Marden, and Eric W. Frew Abstract The log-linear learning algorithm has been extensively studied in both the game theoretic and

More information

MS&E 246: Lecture 4 Mixed strategies. Ramesh Johari January 18, 2007

MS&E 246: Lecture 4 Mixed strategies. Ramesh Johari January 18, 2007 MS&E 246: Lecture 4 Mixed strategies Ramesh Johari January 18, 2007 Outline Mixed strategies Mixed strategy Nash equilibrium Existence of Nash equilibrium Examples Discussion of Nash equilibrium Mixed

More information

Efficient Sensor Network Planning Method. Using Approximate Potential Game

Efficient Sensor Network Planning Method. Using Approximate Potential Game Efficient Sensor Network Planning Method 1 Using Approximate Potential Game Su-Jin Lee, Young-Jin Park, and Han-Lim Choi, Member, IEEE arxiv:1707.00796v1 [cs.gt] 4 Jul 2017 Abstract This paper addresses

More information

Bayes Correlated Equilibrium and Comparing Information Structures

Bayes Correlated Equilibrium and Comparing Information Structures Bayes Correlated Equilibrium and Comparing Information Structures Dirk Bergemann and Stephen Morris Spring 2013: 521 B Introduction game theoretic predictions are very sensitive to "information structure"

More information

Inertial Game Dynamics

Inertial Game Dynamics ... Inertial Game Dynamics R. Laraki P. Mertikopoulos CNRS LAMSADE laboratory CNRS LIG laboratory ADGO'13 Playa Blanca, October 15, 2013 ... Motivation Main Idea: use second order tools to derive efficient

More information

Distributed Learning based on Entropy-Driven Game Dynamics

Distributed Learning based on Entropy-Driven Game Dynamics Distributed Learning based on Entropy-Driven Game Dynamics Bruno Gaujal joint work with Pierre Coucheney and Panayotis Mertikopoulos Inria Aug., 2014 Model Shared resource systems (network, processors)

More information

Correlated Equilibrium in Games with Incomplete Information

Correlated Equilibrium in Games with Incomplete Information Correlated Equilibrium in Games with Incomplete Information Dirk Bergemann and Stephen Morris Econometric Society Summer Meeting June 2012 Robust Predictions Agenda game theoretic predictions are very

More information

Stable Utility Design for Distributed Resource Allocation*

Stable Utility Design for Distributed Resource Allocation* Stable Utility Design for Distributed Resource Allocation* Ragavendran Gopalakrishnan 1 and Sean D. Nixon 2 and Jason R. Marden 3 Abstract The framework of resource allocation games is becoming an increasingly

More information

Game Theory with Information: Introducing the Witsenhausen Intrinsic Model

Game Theory with Information: Introducing the Witsenhausen Intrinsic Model Game Theory with Information: Introducing the Witsenhausen Intrinsic Model Michel De Lara and Benjamin Heymann Cermics, École des Ponts ParisTech France École des Ponts ParisTech March 15, 2017 Information

More information

Cyber-Awareness and Games of Incomplete Information

Cyber-Awareness and Games of Incomplete Information Cyber-Awareness and Games of Incomplete Information Jeff S Shamma Georgia Institute of Technology ARO/MURI Annual Review August 23 24, 2010 Preview Game theoretic modeling formalisms Main issue: Information

More information

First Prev Next Last Go Back Full Screen Close Quit. Game Theory. Giorgio Fagiolo

First Prev Next Last Go Back Full Screen Close Quit. Game Theory. Giorgio Fagiolo Game Theory Giorgio Fagiolo giorgio.fagiolo@univr.it https://mail.sssup.it/ fagiolo/welcome.html Academic Year 2005-2006 University of Verona Summary 1. Why Game Theory? 2. Cooperative vs. Noncooperative

More information

6.254 : Game Theory with Engineering Applications Lecture 8: Supermodular and Potential Games

6.254 : Game Theory with Engineering Applications Lecture 8: Supermodular and Potential Games 6.254 : Game Theory with Engineering Applications Lecture 8: Supermodular and Asu Ozdaglar MIT March 2, 2010 1 Introduction Outline Review of Supermodular Games Reading: Fudenberg and Tirole, Section 12.3.

More information

Privacy and Fault-Tolerance in Distributed Optimization. Nitin Vaidya University of Illinois at Urbana-Champaign

Privacy and Fault-Tolerance in Distributed Optimization. Nitin Vaidya University of Illinois at Urbana-Champaign Privacy and Fault-Tolerance in Distributed Optimization Nitin Vaidya University of Illinois at Urbana-Champaign Acknowledgements Shripad Gade Lili Su argmin x2x SX i=1 i f i (x) Applications g f i (x)

More information

Computing Equilibria of Repeated And Dynamic Games

Computing Equilibria of Repeated And Dynamic Games Computing Equilibria of Repeated And Dynamic Games Şevin Yeltekin Carnegie Mellon University ICE 2012 July 2012 1 / 44 Introduction Repeated and dynamic games have been used to model dynamic interactions

More information

Static (or Simultaneous- Move) Games of Complete Information

Static (or Simultaneous- Move) Games of Complete Information Static (or Simultaneous- Move) Games of Complete Information Introduction to Games Normal (or Strategic) Form Representation Teoria dos Jogos - Filomena Garcia 1 Outline of Static Games of Complete Information

More information

Minimizing Price of Anarchy in Resource Allocation Games

Minimizing Price of Anarchy in Resource Allocation Games University of Colorado, Boulder CU Scholar Electrical, Computer & Energy Engineering Graduate Theses & Dissertations Electrical, Computer & Energy Engineering Spring 4-1-2014 Minimizing Price of Anarchy

More information

Learning Equilibrium as a Generalization of Learning to Optimize

Learning Equilibrium as a Generalization of Learning to Optimize Learning Equilibrium as a Generalization of Learning to Optimize Dov Monderer and Moshe Tennenholtz Faculty of Industrial Engineering and Management Technion Israel Institute of Technology Haifa 32000,

More information

Dynamic Games with Asymmetric Information: Common Information Based Perfect Bayesian Equilibria and Sequential Decomposition

Dynamic Games with Asymmetric Information: Common Information Based Perfect Bayesian Equilibria and Sequential Decomposition Dynamic Games with Asymmetric Information: Common Information Based Perfect Bayesian Equilibria and Sequential Decomposition 1 arxiv:1510.07001v1 [cs.gt] 23 Oct 2015 Yi Ouyang, Hamidreza Tavafoghi and

More information

Mathematical Optimization Models and Applications

Mathematical Optimization Models and Applications Mathematical Optimization Models and Applications Yinyu Ye Department of Management Science and Engineering Stanford University Stanford, CA 94305, U.S.A. http://www.stanford.edu/ yyye Chapters 1, 2.1-2,

More information

The Role of Population Games and Evolutionary Dynamics in Distributed Control Systems

The Role of Population Games and Evolutionary Dynamics in Distributed Control Systems The Role of Population Games and Evolutionary Dynamics in Distributed Control Systems Nicanor Quijano, Carlos Ocampo-Martinez, Julian Barreiro-Gomez, German Obando, Andres Pantoja, Eduardo Mojica-Nava

More information

Distributed Optimization over Random Networks

Distributed Optimization over Random Networks Distributed Optimization over Random Networks Ilan Lobel and Asu Ozdaglar Allerton Conference September 2008 Operations Research Center and Electrical Engineering & Computer Science Massachusetts Institute

More information

Tipping Points of Diehards in Social Consensus on Large Random Networks

Tipping Points of Diehards in Social Consensus on Large Random Networks Tipping Points of Diehards in Social Consensus on Large Random Networks W. Zhang, C. Lim, B. Szymanski Abstract We introduce the homogeneous pair approximation to the Naming Game (NG) model, establish

More information

3.3.3 Illustration: Infinitely repeated Cournot duopoly.

3.3.3 Illustration: Infinitely repeated Cournot duopoly. will begin next period less effective in deterring a deviation this period. Nonetheless, players can do better than just repeat the Nash equilibrium of the constituent game. 3.3.3 Illustration: Infinitely

More information

A Generic Bound on Cycles in Two-Player Games

A Generic Bound on Cycles in Two-Player Games A Generic Bound on Cycles in Two-Player Games David S. Ahn February 006 Abstract We provide a bound on the size of simultaneous best response cycles for generic finite two-player games. The bound shows

More information

EE613 Machine Learning for Engineers. Kernel methods Support Vector Machines. jean-marc odobez 2015

EE613 Machine Learning for Engineers. Kernel methods Support Vector Machines. jean-marc odobez 2015 EE613 Machine Learning for Engineers Kernel methods Support Vector Machines jean-marc odobez 2015 overview Kernel methods introductions and main elements defining kernels Kernelization of k-nn, K-Means,

More information

Finding Optimal Strategies for Influencing Social Networks in Two Player Games. MAJ Nick Howard, USMA Dr. Steve Kolitz, Draper Labs Itai Ashlagi, MIT

Finding Optimal Strategies for Influencing Social Networks in Two Player Games. MAJ Nick Howard, USMA Dr. Steve Kolitz, Draper Labs Itai Ashlagi, MIT Finding Optimal Strategies for Influencing Social Networks in Two Player Games MAJ Nick Howard, USMA Dr. Steve Kolitz, Draper Labs Itai Ashlagi, MIT Problem Statement Given constrained resources for influencing

More information

4. Opponent Forecasting in Repeated Games

4. Opponent Forecasting in Repeated Games 4. Opponent Forecasting in Repeated Games Julian and Mohamed / 2 Learning in Games Literature examines limiting behavior of interacting players. One approach is to have players compute forecasts for opponents

More information

Competitive Scheduling in Wireless Collision Channels with Correlated Channel State

Competitive Scheduling in Wireless Collision Channels with Correlated Channel State Competitive Scheduling in Wireless Collision Channels with Correlated Channel State Utku Ozan Candogan, Ishai Menache, Asuman Ozdaglar and Pablo A. Parrilo Abstract We consider a wireless collision channel,

More information

SEVERAL social and engineering systems can be thought of as a collection of multiple

SEVERAL social and engineering systems can be thought of as a collection of multiple The Importance of System-Level Information in Multiagent Systems Design: Cardinality and Covering Problems Dario Paccagnan and Jason R. Marden 2 Abstract A fundamental challenge in multiagent systems is

More information

Research and Development

Research and Development Chapter 9. March 7, 2011 Firms spend substantial amounts on. For instance ( expenditure to output sales): aerospace (23%), o ce machines and computers (18%), electronics (10%) and drugs (9%). is classi

More information

COMPUTATION OF GENERALIZED NASH EQUILIBRIA WITH

COMPUTATION OF GENERALIZED NASH EQUILIBRIA WITH COMPUTATION OF GENERALIZED NASH EQUILIBRIA WITH THE GNE PACKAGE C. Dutang 1,2 1 ISFA - Lyon, 2 AXA GRM - Paris, 1/19 12/08/2011 user! 2011 OUTLINE 1 GAP FUNCTION MINIZATION APPROACH 2 FIXED-POINT APPROACH

More information

1 Maximizing a Submodular Function

1 Maximizing a Submodular Function 6.883 Learning with Combinatorial Structure Notes for Lecture 16 Author: Arpit Agarwal 1 Maximizing a Submodular Function In the last lecture we looked at maximization of a monotone submodular function,

More information

MS&E 246: Lecture 12 Static games of incomplete information. Ramesh Johari

MS&E 246: Lecture 12 Static games of incomplete information. Ramesh Johari MS&E 246: Lecture 12 Static games of incomplete information Ramesh Johari Incomplete information Complete information means the entire structure of the game is common knowledge Incomplete information means

More information

Supermodular Games. Ichiro Obara. February 6, 2012 UCLA. Obara (UCLA) Supermodular Games February 6, / 21

Supermodular Games. Ichiro Obara. February 6, 2012 UCLA. Obara (UCLA) Supermodular Games February 6, / 21 Supermodular Games Ichiro Obara UCLA February 6, 2012 Obara (UCLA) Supermodular Games February 6, 2012 1 / 21 We study a class of strategic games called supermodular game, which is useful in many applications

More information

Game Theory, Evolutionary Dynamics, and Multi-Agent Learning. Prof. Nicola Gatti

Game Theory, Evolutionary Dynamics, and Multi-Agent Learning. Prof. Nicola Gatti Game Theory, Evolutionary Dynamics, and Multi-Agent Learning Prof. Nicola Gatti (nicola.gatti@polimi.it) Game theory Game theory: basics Normal form Players Actions Outcomes Utilities Strategies Solutions

More information

A Modified Q-Learning Algorithm for Potential Games

A Modified Q-Learning Algorithm for Potential Games Preprints of the 19th World Congress The International Federation of Automatic Control A Modified Q-Learning Algorithm for Potential Games Yatao Wang Lacra Pavel Edward S. Rogers Department of Electrical

More information

Mechanism Design: Basic Concepts

Mechanism Design: Basic Concepts Advanced Microeconomic Theory: Economics 521b Spring 2011 Juuso Välimäki Mechanism Design: Basic Concepts The setup is similar to that of a Bayesian game. The ingredients are: 1. Set of players, i {1,

More information

Multi-Agent Learning with Policy Prediction

Multi-Agent Learning with Policy Prediction Multi-Agent Learning with Policy Prediction Chongjie Zhang Computer Science Department University of Massachusetts Amherst, MA 3 USA chongjie@cs.umass.edu Victor Lesser Computer Science Department University

More information

Cyclic Equilibria in Markov Games

Cyclic Equilibria in Markov Games Cyclic Equilibria in Markov Games Martin Zinkevich and Amy Greenwald Department of Computer Science Brown University Providence, RI 02912 {maz,amy}@cs.brown.edu Michael L. Littman Department of Computer

More information

Retrospective Spectrum Access Protocol: A Completely Uncoupled Learning Algorithm for Cognitive Networks

Retrospective Spectrum Access Protocol: A Completely Uncoupled Learning Algorithm for Cognitive Networks Retrospective Spectrum Access Protocol: A Completely Uncoupled Learning Algorithm for Cognitive Networks Marceau Coupechoux, Stefano Iellamo, Lin Chen + TELECOM ParisTech (INFRES/RMS) and CNRS LTCI + University

More information

Dynamic and Adversarial Reachavoid Symbolic Planning

Dynamic and Adversarial Reachavoid Symbolic Planning Dynamic and Adversarial Reachavoid Symbolic Planning Laya Shamgah Advisor: Dr. Karimoddini July 21 st 2017 Thrust 1: Modeling, Analysis and Control of Large-scale Autonomous Vehicles (MACLAV) Sub-trust

More information

Learning, Games, and Networks

Learning, Games, and Networks Learning, Games, and Networks Abhishek Sinha Laboratory for Information and Decision Systems MIT ML Talk Series @CNRG December 12, 2016 1 / 44 Outline 1 Prediction With Experts Advice 2 Application to

More information

Game theory Lecture 19. Dynamic games. Game theory

Game theory Lecture 19. Dynamic games. Game theory Lecture 9. Dynamic games . Introduction Definition. A dynamic game is a game Γ =< N, x, {U i } n i=, {H i } n i= >, where N = {, 2,..., n} denotes the set of players, x (t) = f (x, u,..., u n, t), x(0)

More information

Multi-Robotic Systems

Multi-Robotic Systems CHAPTER 9 Multi-Robotic Systems The topic of multi-robotic systems is quite popular now. It is believed that such systems can have the following benefits: Improved performance ( winning by numbers ) Distributed

More information

REPEATED GAMES. Jörgen Weibull. April 13, 2010

REPEATED GAMES. Jörgen Weibull. April 13, 2010 REPEATED GAMES Jörgen Weibull April 13, 2010 Q1: Can repetition induce cooperation? Peace and war Oligopolistic collusion Cooperation in the tragedy of the commons Q2: Can a game be repeated? Game protocols

More information

Constrained Consensus and Optimization in Multi-Agent Networks

Constrained Consensus and Optimization in Multi-Agent Networks Constrained Consensus Optimization in Multi-Agent Networks The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters. Citation As Published Publisher

More information

Robust distributed linear programming

Robust distributed linear programming Robust distributed linear programming Dean Richert Jorge Cortés Abstract This paper presents a robust, distributed algorithm to solve general linear programs. The algorithm design builds on the characterization

More information

BELIEFS & EVOLUTIONARY GAME THEORY

BELIEFS & EVOLUTIONARY GAME THEORY 1 / 32 BELIEFS & EVOLUTIONARY GAME THEORY Heinrich H. Nax hnax@ethz.ch & Bary S. R. Pradelski bpradelski@ethz.ch May 15, 217: Lecture 1 2 / 32 Plan Normal form games Equilibrium invariance Equilibrium

More information

Multimarket Oligopolies with Restricted Market Access

Multimarket Oligopolies with Restricted Market Access Multimarket Oligopolies with Restricted Market Access Tobias Harks 1 and Max Klimm 2 1 Department of Quantitative Economics, Maastricht University, the Netherlands. t.harks@maastrichtuniversity.nl 2 Department

More information

Non-Collision Conditions in Multi-agent Robots Formation using Local Potential Functions

Non-Collision Conditions in Multi-agent Robots Formation using Local Potential Functions 2008 IEEE International Conference on Robotics and Automation Pasadena, CA, USA, May 19-23, 2008 Non-Collision Conditions in Multi-agent Robots Formation using Local Potential Functions E G Hernández-Martínez

More information

Game Theory and Control

Game Theory and Control Game Theory and Control Jason R. Marden 1 and Jeff S. Shamma 2 1 Department of Electrical and Computer Engineering, University of California, Santa Barbara, Santa Barbara, CA, USA, 93106; jrmarden@ece.ucsb.edu

More information

11. Learning graphical models

11. Learning graphical models Learning graphical models 11-1 11. Learning graphical models Maximum likelihood Parameter learning Structural learning Learning partially observed graphical models Learning graphical models 11-2 statistical

More information

A Hybrid Framework for Resource Allocation among Multiple Agents Moving on Discrete Environments

A Hybrid Framework for Resource Allocation among Multiple Agents Moving on Discrete Environments A Hybrid Framework for Resource Allocation among Multiple Agents Moving on Discrete Environments Jorge L. Piovesan, Chaouki T. Abdallah, and Herbert G. Tanner 1 Abstract We consider the problem of controlling

More information

Vickrey Auction VCG Characterization. Mechanism Design. Algorithmic Game Theory. Alexander Skopalik Algorithmic Game Theory 2013 Mechanism Design

Vickrey Auction VCG Characterization. Mechanism Design. Algorithmic Game Theory. Alexander Skopalik Algorithmic Game Theory 2013 Mechanism Design Algorithmic Game Theory Vickrey Auction Vickrey-Clarke-Groves Mechanisms Characterization of IC Mechanisms Mechanisms with Money Player preferences are quantifiable. Common currency enables utility transfer

More information

Players as Serial or Parallel Random Access Machines. Timothy Van Zandt. INSEAD (France)

Players as Serial or Parallel Random Access Machines. Timothy Van Zandt. INSEAD (France) Timothy Van Zandt Players as Serial or Parallel Random Access Machines DIMACS 31 January 2005 1 Players as Serial or Parallel Random Access Machines (EXPLORATORY REMARKS) Timothy Van Zandt tvz@insead.edu

More information

Mathematical Economics - PhD in Economics

Mathematical Economics - PhD in Economics - PhD in Part 1: Supermodularity and complementarity in the one-dimensional and Paulo Brito ISEG - Technical University of Lisbon November 24, 2010 1 2 - Supermodular optimization 3 one-dimensional 4 Supermodular

More information

Persuading Skeptics and Reaffirming Believers

Persuading Skeptics and Reaffirming Believers Persuading Skeptics and Reaffirming Believers May, 31 st, 2014 Becker-Friedman Institute Ricardo Alonso and Odilon Camara Marshall School of Business - USC Introduction Sender wants to influence decisions

More information

A Solution to the Problem of Externalities When Agents Are Well-Informed

A Solution to the Problem of Externalities When Agents Are Well-Informed A Solution to the Problem of Externalities When Agents Are Well-Informed Hal R. Varian. The American Economic Review, Vol. 84, No. 5 (Dec., 1994), pp. 1278-1293 Introduction There is a unilateral externality

More information

Axiomatic bargaining. theory

Axiomatic bargaining. theory Axiomatic bargaining theory Objective: To formulate and analyse reasonable criteria for dividing the gains or losses from a cooperative endeavour among several agents. We begin with a non-empty set of

More information

A Price-Based Approach for Controlling Networked Distributed Energy Resources

A Price-Based Approach for Controlling Networked Distributed Energy Resources A Price-Based Approach for Controlling Networked Distributed Energy Resources Alejandro D. Domínguez-García (joint work with Bahman Gharesifard and Tamer Başar) Coordinated Science Laboratory Department

More information

Distributed Optimization over Networks Gossip-Based Algorithms

Distributed Optimization over Networks Gossip-Based Algorithms Distributed Optimization over Networks Gossip-Based Algorithms Angelia Nedić angelia@illinois.edu ISE Department and Coordinated Science Laboratory University of Illinois at Urbana-Champaign Outline Random

More information

SEQUENTIAL ESTIMATION OF DYNAMIC DISCRETE GAMES. Victor Aguirregabiria (Boston University) and. Pedro Mira (CEMFI) Applied Micro Workshop at Minnesota

SEQUENTIAL ESTIMATION OF DYNAMIC DISCRETE GAMES. Victor Aguirregabiria (Boston University) and. Pedro Mira (CEMFI) Applied Micro Workshop at Minnesota SEQUENTIAL ESTIMATION OF DYNAMIC DISCRETE GAMES Victor Aguirregabiria (Boston University) and Pedro Mira (CEMFI) Applied Micro Workshop at Minnesota February 16, 2006 CONTEXT AND MOTIVATION Many interesting

More information

On the Existence of Strong Nash Equilibria

On the Existence of Strong Nash Equilibria On the Existence of Strong Nash Equilibria Rabia Nessah CNRS-LEM (UMR 8179) IESEG School of Management 3 rue de la Digue F-59000 Lille France Guoqiang Tian Department of Economics Texas A&M University

More information

Decentralized Convergence to Nash Equilibria in Constrained Deterministic Mean Field Control

Decentralized Convergence to Nash Equilibria in Constrained Deterministic Mean Field Control Decentralized Convergence to Nash Equilibria in Constrained Deterministic Mean Field Control 1 arxiv:1410.4421v2 [cs.sy] 17 May 2015 Sergio Grammatico, Francesca Parise, Marcello Colombino, and John Lygeros

More information

EVOLUTIONARY GAMES WITH GROUP SELECTION

EVOLUTIONARY GAMES WITH GROUP SELECTION EVOLUTIONARY GAMES WITH GROUP SELECTION Martin Kaae Jensen Alexandros Rigos Department of Economics University of Leicester Controversies in Game Theory: Homo Oeconomicus vs. Homo Socialis ETH Zurich 12/09/2014

More information

Asymmetric Information Security Games 1/43

Asymmetric Information Security Games 1/43 Asymmetric Information Security Games Jeff S. Shamma with Lichun Li & Malachi Jones & IPAM Graduate Summer School Games and Contracts for Cyber-Physical Security 7 23 July 2015 Jeff S. Shamma Asymmetric

More information

Exponential Moving Average Based Multiagent Reinforcement Learning Algorithms

Exponential Moving Average Based Multiagent Reinforcement Learning Algorithms Exponential Moving Average Based Multiagent Reinforcement Learning Algorithms Mostafa D. Awheda Department of Systems and Computer Engineering Carleton University Ottawa, Canada KS 5B6 Email: mawheda@sce.carleton.ca

More information

GAMES with information and payoff externalities are

GAMES with information and payoff externalities are 2250 IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL 62, NO 9, MAY 1, 2014 Bayesian Quadratic Network Game Filters Ceyhun Eksin, Student Member, IEEE, Pooya Molavi, Alejandro Ribeiro, Member, IEEE, and Ali

More information

Existence of Nash Networks in One-Way Flow Models

Existence of Nash Networks in One-Way Flow Models Existence of Nash Networks in One-Way Flow Models pascal billand a, christophe bravard a, sudipta sarangi b a CREUSET, Jean Monnet University, Saint-Etienne, France. email: pascal.billand@univ-st-etienne.fr

More information

Opinion Dynamics on Triad Scale Free Network

Opinion Dynamics on Triad Scale Free Network Opinion Dynamics on Triad Scale Free Network Li Qianqian 1 Liu Yijun 1,* Tian Ruya 1,2 Ma Ning 1,2 1 Institute of Policy and Management, Chinese Academy of Sciences, Beijing 100190, China lqqcindy@gmail.com,

More information

Cyber Security Games with Asymmetric Information

Cyber Security Games with Asymmetric Information Cyber Security Games with Asymmetric Information Jeff S. Shamma Georgia Institute of Technology Joint work with Georgios Kotsalis & Malachi Jones ARO MURI Annual Review November 15, 2012 Research Thrust:

More information

Introduction Benchmark model Belief-based model Empirical analysis Summary. Riot Networks. Lachlan Deer Michael D. König Fernando Vega-Redondo

Introduction Benchmark model Belief-based model Empirical analysis Summary. Riot Networks. Lachlan Deer Michael D. König Fernando Vega-Redondo Riot Networks Lachlan Deer Michael D. König Fernando Vega-Redondo University of Zurich University of Zurich Bocconi University June 7, 2018 Deer & König &Vega-Redondo Riot Networks June 7, 2018 1 / 23

More information