Matrix games with missing,interval, and ambiguous lottery payoffs of pure strategy profiles and compound strategy profiles

In a matrix game, the interactions among players are based on the assumption that each player has accurate information about the payoffs of their interactions and the other players are rationally self‐interested. As a result, the players should definitely take Nash equilibrium strategies. However, in real‐life, when choosing their optimal strategies, sometimes the players have to face missing, imprecise (i.e., interval), ambiguous lottery payoffs of pure strategy profiles and even compound strategy profile, which means that it is hard to determine a Nash equilibrium. To address this issue, in this paper we introduce a new solution concept, called ambiguous Nash equilibrium, which extends the concept of Nash equilibrium to the one that can handle these types of ambiguous payoff. Moreover, we will reveal some properties of matrix games of this kind. In particular, we show that a Nash equilibrium is a special case of ambiguous Nash equilibrium if the players have accurate information of each player's payoff sets. Finally, we give an example to illustrate how our approach deals with real‐life game theory problems.     

Equilibria problems on games: Complexity versus succinctness

We study the computational complexity of problems involving equilibria in strategic games and in perfect information extensive games when the number of players is large. We consider, among others, the problems of deciding the existence of a pure Nash equilibrium in strategic games or deciding the existence of a pure Nash or a subgame perfect Nash equilibrium with a given payoff in finite perfect information extensive games. We address the fundamental question of how can we represent a game with a large number of players? We propose three ways of representing a game with different degrees of succinctness for the components of the game. For perfect information extensive games we show that when the number of moves of each player is large and the input game is represented succinctly these problems are PSPACE-complete. In contraposition, when the game is described explicitly by means of its associated tree all these problems are decidable in polynomial time. For strategic games we show that the complexity of deciding the existence of a pure Nash equilibrium depends on the succinctness of the game representation and then on the size of the action sets. In particular we show that it is NP-complete, when the number of players is large and the number of actions for each player is constant, and that the problem is -complete when the number of players is a constant and the size of the action sets is exponential in the size of the game representation. Again when the game is described explicitly the problem is decidable in polynomial time.     

Reaching pareto-optimality in prisoner’s dilemma using conditional joint action learning

We consider the learning problem faced by two self-interested agents repeatedly playing a general-sum stage game. We assume that the players can observe each other’s actions but not the payoffs received by the other player. The concept of Nash Equilibrium in repeated games provides an individually rational solution for playing such games and can be achieved by playing the Nash Equilibrium strategy for the single-shot game in every iteration. Such a strategy, however can sometimes lead to a Pareto-Dominated outcome for games like Prisoner’s Dilemma. So we prefer learning strategies that converge to a Pareto-Optimal outcome that also produces a Nash Equilibrium payoff for repeated two-player, n-action general-sum games. The Folk Theorem enable us to identify such outcomes. In this paper, we introduce the Conditional Joint Action Learner (CJAL) which learns the conditional probability of an action taken by the opponent given its own actions and uses it to decide its next course of action. We empirically show that under self-play and if the payoff structure of the Prisoner’s Dilemma game satisfies certain conditions, a CJAL learner, using a random exploration strategy followed by a completely greedy exploitation technique, will learn to converge to a Pareto-Optimal solution. We also show that such learning will generate Pareto-Optimal payoffs in a large majority of other two-player general sum games. We compare the performance of CJAL with that of existing algorithms such as WOLF-PHC and JAL on all structurally distinct two-player conflict games with ordinal payoffs.     

When Can Limited Randomness Be Used in Repeated Games?

The central result of classical game theory states that every finite normal form game has a Nash equilibrium, provided that players are allowed to use randomized (mixed) strategies. However, in practice, humans are known to be bad at generating random-like sequences, and true random bits may be unavailable. Even if the players have access to enough random bits for a single instance of the game their randomness might be insufficient if the game is played many times. In this work, we ask whether randomness is necessary for equilibria to exist in finitely repeated games. We show that for a large class of games containing arbitrary two-player zero-sum games, approximate Nash equilibria of the n-stage repeated version of the game exist if and only if both players have Ω(n) random bits. In contrast, we show that there exists a class of games for which no equilibrium exists in pure strategies, yet the n-stage repeated version of the game has an exact Nash equilibrium in which each player uses only a constant number of random bits. When the players are assumed to be computationally bounded, if cryptographic pseudorandom generators (or, equivalently, one-way functions) exist, then the players can base their strategies on “random-like” sequences derived from only a small number of truly random bits. We show that, in contrast, in repeated two-player zero-sum games, if pseudorandom generators do not exist, then Ω(n) random bits remain necessary for equilibria to exist.     

Discovering theorems in game theory: Two-person games with unique pure Nash equilibrium payoffs

In this paper we provide a logical framework for two-person finite games in strategic form, and use it to design a computer program for discovering some classes of games that have unique pure Nash equilibrium payoffs. The classes of games that we consider are those that can be expressed by a conjunction of two binary clauses, and our program re-discovered Kats and Thisse?s class of weakly unilaterally competitive two-person games, and came up with several other classes of games that have unique pure Nash equilibrium payoffs. It also came up with new classes of strict games that have unique pure Nash equilibria, where a game is strict if for both player different profiles have different payoffs.     

On Dynamics and Nash Equilibriums of Networked Games

Networked noncooperative games are investigated, where each player (or agent) plays with all other players in its neighborhood. Assume the evolution is based on the fact that each player uses its neighbors' current information to decide its next strategy. By using sub-neighborhood, the dynamics of the evolution is obtained. Then a method for calculating Nash equilibriums from mixed strategies of multi-players is proposed. The relationship between local Nash equilibriums based on individual neighborhoods and global Nash equilibriums of overall network is revealed. Then a technique is proposed to construct Nash equilibriums of an evolutionary game from its one step static Nash equilibriums. The basic tool of this approach is the semi-tensor product of matrices, which converts strategies into logical matrices and payoffs into pseudo-Boolean functions, then networked evolutionary games become discrete time dynamic systems.      

Nash strategies for M-person differential games with mixed information structures

This paper is concerned with a class of M-person linear-quadratic nonzero-sum differential games in which a subset of the players have access to closed-loop (CL) information and the rest to open-loop (OL) information. The state equation contains an additive random perturbation term, inclusion of which has been shown to be necessary in order to obtain, a unique globally optimal Nash equilibrium solution. For each player with CL information, the optimal strategy is a linear function of the current and the initial states, and for each player with OL information, the optimal strategy is a linear function of the initial state.     

Incompetence and impact of training in bimatrix games

The concept of games with incompetence has been introduced to better represent games where players may not be capable of executing strategies that they select. In particular this paper introduces incompetence into bimatrix games and investigates the properties of such games. The results obtained describe both the general dependence of “extreme Nash equilibrium payoffs” on incompetence and special behaviour arising in particular cases. The dependence of the payoffs can be complex and include non-linearities and transition points. Transition points occur when kernels change and may result in the number of “extreme Nash equilibria” changing. Understanding these changes allows the determination of the benefits of regimes that seek to decrease a player’s incompetence. While the games we consider are normally static, in our context there is a hidden dynamics resulting from the fact that players will strive to improve their equilibrium payoffs by changing their incompetence levels. This might require training, in the case of games like tennis, or it might require the purchase of new equipment costing billions of dollars, in the case of military applications.     

Convergence analysis for pure stationary strategies in repeated potential games: Nash,Lyapunov and correlated equilibria

In game theory the interaction among players obligates each player to develop a belief about the possible strategies of the other players, to choose a best-reply given those beliefs, and to look for an adjustment of the best-reply and the beliefs using a learning mechanism until they reach an equilibrium point. Usually, the behavior of an individual cost-function, when such best-reply strategies are applied, turns out to be non-monotonic and concluding that such strategies lead to some equilibrium point is a non-trivial task. Even in repeated games the convergence to a stationary equilibrium is not always guaranteed. The best-reply strategies analyzed in this paper represent the most frequent type of behavior applied in practice in problems of bounded rationality of agents considered within the Artificial Intelligence research area. They are naturally related with the, so-called, fixed-local-optimal actions or, in other words, with one step-ahead optimization algorithms widely used in the modern Intelligent Systems theory.This paper shows that for an ergodic class of finite controllable Markov games the best-reply strategies lead necessarily to a Lyapunov/Nash equilibrium point. One of the most interesting properties of this approach is that an expedient (or absolutely expedient) behavior of an ergodic system (repeated game) can be represented by a Lyapunov-like function non-decreasing in time. We present a method for constructing a Lyapunov-like function: the Lyapunov-like function replaces the recursive mechanism with the elements of the ergodic system that model how players are likely to behave in one-shot games. To show our statement, we first propose a non-converging state-value function that fluctuates (increases and decreases) between states of the Markov game. Then, we prove that it is possible to represent that function in a recursive format using a one-step-ahead fixed-local-optimal strategy. As a result, we prove that a Lyapunov-like function can be built using the previous recursive expression for the Markov game, i.e., the resulting Lyapunov-like function is a monotonic function which can only decrease (or remain the same) over time, whatever the initial distribution of probabilities. As a result, a new concept called Lyapunov games is suggested for a class of repeated games. Lyapunov games allow to conclude during the game whether the applied strategy provides the convergence to an equilibrium point (or not). The time for constructing a Potential (Lyapunov-like) function is exponential. Our algorithm tractably computes the Nash, Lyapunov and the correlated equilibria: a Lyapunov equilibrium is a Nash equilibrium, as well it is also a correlated equilibrium. Validity of the proposed method is successfully demonstrated both theoretically and practically by a simulated experiment related to the Duel game.     

The Quantum Ultimatum Game

A quantum version of the ultimatum game is studied. Both a restricted version with classical moves and the unitary version are considered. With entangled initial states, Nash equilibria in quantum games are in general different from those of classical games. Quantum versions might therefore be useful as a framework for modeling deviations from classical Nash equilibrium in experimental games.PACS:02.50.Le; 03.67.-a     

Altruistic behavior in a nonantagonistic positional differential game

We consider a two-person nonantagonistic positional differential game (NPDG) whose dynamics is described by an ordinary nonlinear vector differential equation. Constraints on values of players’ controls are geometric. Final time of the game is fixed. Payoff functionals of both players are terminal. The formalization of positional strategies in an NPDG is based on the formalization and results of the general theory of antagonistic positional differential games (APDGs) (see monographs by N.N. Krasovskii and A.I. Subbotin [3, 4]). Additionally, in the present paper we assume that each player, together with the usual, normal (nor), type of behavior aimed at maximizing his own functional, can use other behavior types introduced in [2, 5]. In particular, these may be altruistic (alt), aggressive (agg), and paradoxical (par) types. It is assumed that in the course of the game players can switch their behavior from one type to another. Using the possibility of such switches in a repeated bimatrix 2 × 2 game in [5, 6] allowed to obtain new solutions of this game. In the present paper, extension of this approach to NPDGs leads to a new formulation of the problem. In particular, of interest is the question of how players’ outcomes at Nash solutions are transformed. An urgent problem is minimizing the time of “abnormal” behavior while achieving a good result. The paper proposes a formalization of an NPDG with behavior types (NPDGwBT). It is assumed that in an NPDGwBT each player, simultaneously with choosing a positional strategy, chooses also his own indicator function defined on the whole game horizon and taking values in the set {normal, altruistic, aggressive, paradoxical}. The indicator function of a player shows the dynamics of changes in the behavior type demonstrated by the player. Thus, in this NPDGwBT each player controls the choice of a pair {positional strategy, indicator function}. We define the notion of a BT-solution of such a game. It is expected that using behavior types in the NPDGwBT which differ from the normal one (so-called abnormal types) in some cases may lead to more favorable outcomes for the players than in the NPDG. We consider two examples of an NPDGwBT with simple dynamics in the plane in each of which one player keeps to altruistic behavior type over some time period. It is shown that in the first example payoffs of both players increase on a BT-solution as compared to the game with the normal behavior type, and in the second example, the sum of players’ payoffs is increased.     

Random Bimatrix Games Are Asymptotically Easy to Solve (A Simple Proof)

We focus on the problem of computing approximate Nash equilibria and well-supported approximate Nash equilibria in random bimatrix games, where each player’s payoffs are bounded and independent random variables, not necessarily identically distributed, but with almost common expectations. We show that the completely mixed uniform strategy profile, i.e., the combination of mixed strategies (one per player) where each player plays with equal probability each one of her available pure strategies, is with high probability a $\sqrt{\frac{\ln n}{n}}$ -Nash equilibrium and a $\sqrt{\frac{3\ln n}{n}}$ -well supported Nash equilibrium, where n is the number of pure strategies available to each player. This asserts that the completely mixed, uniform strategy profile is an almost Nash equilibrium for random bimatrix games, since it is, with high probability, an ?-well-supported Nash equilibrium where ? tends to zero as n tends to infinity.     

Noisy relativistic quantum games in noninertial frames

The influence of noise and of Unruh effect on quantum Prisoners’ dilemma is investigated both for entangled and unentangled initial states. The noise is incorporated through amplitude damping channel. For unentangled initial state, the decoherence compensates for the adverse effect of acceleration of the frame and the effect of acceleration becomes irrelevant provided the game is fully decohered. It is shown that the inertial player always out scores the noninertial player by choosing defection. For maximally entangled initially state, we show that for fully decohered case every strategy profile results in either of the two possible equilibrium outcomes. Two of the four possible strategy profiles become Pareto optimal and Nash equilibrium and no dilemma is leftover. It is shown that other equilibrium points emerge for different region of values of decoherence parameter that are either Pareto optimal or Pareto inefficient in the quantum strategic spaces. It is shown that the Eisert et al. (Phys Rev Lett 83:3077, 1999) miracle move is a special move that leads always to distinguishable results compare to other moves. We show that the dilemma like situation is resolved in favor of one player or the other.     

Equilibrium Trajectories in Dynamical Bimatrix Games with Average Integral Payoff Functionals

This paper considers models of evolutionary non-zero-sum games on the infinite time interval. Methods of differential game theory are used for the analysis of game interactions between two groups of participants. We assume that participants in these groups are controlled by signals for the behavior change. The payoffs of coalitions are defined as average integral functionals on the infinite horizon. We pose the design problem of a dynamical Nash equilibrium for the evolutionary game under consideration. The ideas and approaches of non-zero-sum differential games are employed for the determination of the Nash equilibrium solutions. The results derived in this paper involve the dynamic constructions and methods of evolutionary games. Much attention is focused on the formation of the dynamical Nash equilibrium with players strategies that maximize the corresponding payoff functions and have the guaranteed properties according to the minimax approach. An application of the minimax approach for constructing optimal control strategies generates dynamical Nash equilibrium trajectories yielding better results in comparison to static solutions and evolutionary models with the replicator dynamics. Finally, we make a comparison of the dynamical Nash equilibrium trajectories for evolutionary games with the average integral payoff functionals and the trajectories for evolutionary games with the global terminal payoff functionals on the infinite horizon.     

On the Rate of Convergence of Fictitious Play

Fictitious play is a simple learning algorithm for strategic games that proceeds in rounds. In each round, the players play a best response to a mixed strategy that is given by the empirical frequencies of actions played in previous rounds. There is a close relationship between fictitious play and the Nash equilibria of a game: if the empirical frequencies of fictitious play converge to a strategy profile, this strategy profile is a Nash equilibrium. While fictitious play does not converge in general, it is known to do so for certain restricted classes of games, such as constant-sum games, non-degenerate 2×n games, and potential games. We study the rate of convergence of fictitious play and show that, in all the classes of games mentioned above, fictitious play may require an exponential number of rounds (in the size of the representation of the game) before some equilibrium action is eventually played. In particular, we show the above statement for symmetric constant-sum win-lose-tie games.     

Quantum game simulator, using the circuit model of quantum computation

We present a general two-player quantum game simulator that can simulate any two-player quantum game described by a 2×2 payoff matrix (two strategy games).The user can determine the payoff matrices for both players, their strategies and the amount of entanglement between their initial strategies. The outputs of the simulator are the expected payoffs of each player as a function of the other player's strategy parameters and the amount of entanglement. The simulator also produces contour plots that divide the strategy spaces of the game in regions in which players can get larger payoffs if they choose to use a quantum strategy against any classical one. We also apply the simulator to two well-known quantum games, the Battle of Sexes and the Chicken game.

Program summary

Program title: Quantum Game Simulator (QGS)Catalogue identifier: AEED_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AEED_v1_0.htmlProgram obtainable from: CPC Program Library, Queen's University, Belfast, N. IrelandLicensing provisions: Standard CPC licence, http://cpc.cs.qub.ac.uk/licence/licence.htmlNo. of lines in distributed program, including test data, etc.: 3416No. of bytes in distributed program, including test data, etc.: 583 553Distribution format: tar.gzProgramming language: Matlab R2008a (C)Computer: Any computer that can sufficiently run Matlab R2008aOperating system: Any system that can sufficiently run Matlab R2008aClassification: 4.15Nature of problem: Simulation of two player quantum games described by a payoff matrix.Solution method: The program calculates the matrices that comprise the Eisert setup for quantum games based on the quantum circuit model. There are 5 parameters that can be altered. We define 3 of them as constant. We play the quantum game for all possible values for the other 2 parameters and store the results in a matrix.Unusual features: The software provides an easy way of simulating any two-player quantum games.Running time: Approximately 0.4 sec (Region Feature) and 0.3 sec (Payoff Feature) on a Intel Core 2 Duo GHz with 2 GB of memory under Windows XP.     

Establishing Nash equilibria of strategic games: a multistart Fuzzy Adaptive Simulated Annealing approach

This paper's proposal is to show some significant results obtained by the application of the optimization algorithm known as Fuzzy Adaptive Simulated Annealing (Fuzzy ASA) to the task of finding all Nash equilibria of normal form games. To that end, a special version of Fuzzy ASA, that utilizes space-filling curves to find good seeds, is applied to several well-known strategic games, showing its effectiveness in obtaining all Nash equilibria in all cases. The results are compared to previous work that also used computational intelligence techniques in order to solve the same problem but could not find all equilibria in all tests. Game theory is a very important subject, modeling interactions between generic agents, and Nash equilibrium represents a powerful concept portraying situations in which joint strategies are optimal in the sense that no player can benefit from changing her/his strategy while the other players do not change their strategies as well. So, new techniques are always welcome, mainly those that can find the whole set of solutions for a given strategic game.     

Distributed Nash Equilibrium Seeking for General Networked Games With Bounded Disturbances

This paper is concerned with anti-disturbance Nash equilibrium seeking for games with partial information. First, reduced-order disturbance observer-based algorithms are proposed to achieve Nash equilibrium seeking for games with first-order and second-order players, respectively. In the developed algorithms, the observed disturbance values are included in control signals to eliminate the influence of disturbances, based on which a gradient-like optimization method is implemented for each player. Second, a signum function based distributed algorithm is proposed to attenuate disturbances for games with second-order integrator-type players. To be more specific, a signum function is involved in the proposed seeking strategy to dominate disturbances, based on which the feedback of the velocity-like states and the gradients of the functions associated with players achieves stabilization of system dynamics and optimization of players’ objective functions. Through Lyapunov stability analysis, it is proven that the players’ actions can approach a small region around the Nash equilibrium by utilizing disturbance observer-based strategies with appropriate control gains. Moreover, exponential (asymptotic) convergence can be achieved when the signum function based control strategy (with an adaptive control gain) is employed. The performance of the proposed algorithms is tested by utilizing an integrated simulation platform of virtual robot experimentation platform (V-REP) and MATLAB.      

Graphical Congestion Games

We consider congestion games with linear latency functions in which each player is aware only of a subset of all the other players. This is modeled by means of a social knowledge graph G in which nodes represent players and there is an edge from i to j if i knows j. Under the assumption that the payoff of each player is affected only by the strategies of the adjacent ones, we first give a complete characterization of the games possessing pure Nash equilibria. Namely, if the social graph G is undirected, the game is an exact potential game and thus isomorphic to a classical congestion game. As a consequence, it always converges and possesses Nash equilibria. On the other hand, if G is directed an equilibrium is not guaranteed to exist, but the game is always convergent and an equilibrium can be found in polynomial time if G is acyclic, even if finding the best equilibrium remains an intractable problem.     

Numerical Construction of Stackelberg Solutions in a Linear Positional Differential Game Based on the Method of Polyhedra

We consider the problem of constructing approximate Stackelberg solutions in a linear non-zero-sum positional differential game of two players with terminal payoffs and player controls chosen on convex polyhedra. A formalization of player strategies and motions generated by them is based on the formalization and results of the theory of zero-sum positional differential games developed by N.N. Krasovskii and his scientific school. The problem of finding a Stackelberg solution reduces to solving nonstandard optimal control problems. We propose an approach based on operations with convex polyhedra.