Stéphane Le Roux

Infinite sequential games with real-valued payoffs

We investigate the existence of certain types of equilibria (Nash, ε-Nash, subgame perfect, ε-subgame perfect) in infinite sequential games with real-valued payoff functions depending on the class of payoff functions (continuous, upper semi-continuous, Borel) and whether the game is zero-sum. Our results hold for games with two or up to countably many players. Several of these results are corollaries of stronger results that we establish about equilibria in infinite sequential games with some weak conditions on the occurring preference relations. We also formulate an abstract equilibrium transfer result about games with compact strategy spaces and open preferences. Finally, we consider a dynamical improvement rule for infinite sequential games with continuous payoff functions.

Weihrauch Degrees of Finding Equilibria in Sequential Games

We consider the degrees of non-computability (Weihrauch degrees) of finding winning strategies (or more generally, Nash equilibria) in infinite sequential games with certain winning sets (or more generally, outcome sets). In particular, we show that as the complexity of the winning sets increases in the difference hierarchy, the complexity of constructing winning strategies increases in the effective Borel hierarchy.

On the computational content of the brouwer fixed point theorem

We study the computational content of the Brouwer Fixed Point Theorem in the Weihrauch lattice. One of our main results is that for any fixed dimension the Brouwer Fixed Point Theorem of that dimension is computably equivalent to connected choice of the Euclidean unit cube of the same dimension. Connected choice is the operation that finds a point in a non-empty connected closed set given by negative information. Another main result is that connected choice is complete for dimension greater or equal to three in the sense that it is computably equivalent to Weak Kőnigs Lemma. In contrast to this, the connected choice operations in dimensions zero, one and two form a strictly increasing sequence of Weihrauch degrees, where connected choice of dimension one is known to be equivalent to the Intermediate Value Theorem. Whether connected choice of dimension two is strictly below connected choice of dimension three or equivalent to it is unknown, but we conjecture that the reduction is strict. As a side result we also prove that finding a connectedness component in a closed subset of the Euclidean unit cube of any dimension greater than or equal to one is equivalent to Weak Kőnigs Lemma.

Closed Choice for Finite and for Convex Sets

We investigate choice principles in the Weihrauch lattice for finite sets on the one hand, and convex sets on the other hand. Increasing cardinality and increasing dimension both correspond to increasing Weihrauch degrees. Moreover, we demonstrate that the dimension of convex sets can be characterized by the cardinality of finite sets encodable into them. Precisely, choice from an n + 1 point set is reducible to choice from a convex set of dimension n, but not reducible to choice from a convex set of dimension n − 1.

FINITE CHOICE, CONVEX CHOICE AND FINDING ROOTS

We investigate choice principles in the Weihrauch lattice for finite sets on the one hand, and convex sets on the other hand. Increasing cardinality and increasing dimension both correspond to increasing Weihrauch degrees. Moreover, we demonstrate that the dimension of convex sets can be characterized by the cardinality of finite sets encodable into them. Precisely, choice from an n + 1 point set is reducible to choice from a convex set of dimension n, but not reducible to choice from a convex set of dimension n–1. Furthermore we consider searching for zeros of continuous functions. We provide an algorithm producing 3n real numbers containing all zeros of a continuous function with up to n local minima. This demonstrates that having finitely many zeros is a strictly weaker condition than having finitely many local extrema. We can prove 3n to be optimal.

Infinite subgame perfect equilibrium in the hausdorff difference hierarchy

Subgame perfect equilibria are specific Nash equilibria in perfect information games in extensive form. They are important because they relate to the rationality of the players. They always exist in infinite games with continuous real-valued payoffs, but may fail to exist even in simple games with slightly discontinuous payoffs. This article considers only games whose outcome functions are measurable in the Hausdorff difference hierarchy of the open sets (i.e. \({ {\Delta }}^0_2\) when in the Baire space), and it characterizes the families of linear preferences such that every game using these preferences has a subgame perfect equilibrium: the preferences without infinite ascending chains (of course), and such that for all players a and b and outcomes x, y, z we have \(\lnot (z <_a y <_a x \,\wedge \, x <_b z <_b y)\). Moreover at each node of the game, the equilibrium constructed for the proof is Pareto-optimal among all the outcomes occurring in the subgame. Additional results for non-linear preferences are presented.

Extending finite memory determinacy to multiplayer games

We show that under some general conditions the finite memory determinacy of a class of two-player win/lose games played on finite graphs implies the existence of a Nash equilibrium built from finite memory strategies for the corresponding class of multi-player multi-outcome games. This generalizes a previous result by Brihaye, De Pril and Schewe. For most of our conditions we provide counterexamples showing that they cannot be dispensed with. Our proofs are generally constructive, that is, provide upper bounds for the memory required, as well as algorithms to compute the relevant winning strategies.

On the Existence of Weak Subgame Perfect Equilibria

We study multi-player turn-based games played on a directed graph, where the number of players and vertices can be infinite. An outcome is assigned to every play of the game. Each player has a preference relation on the set of outcomes which allows him to compare plays. We focus on the recently introduced notion of weak subgame perfect equilibrium weak SPE, a variant of the classical notion of SPE, where players who deviate can only use strategies deviating from their initial strategy in a finite number of histories. We give general conditions on the structure of the game graph and the preference relations of the players that guarantee the existence of a weak SPE, which moreover is finite-memory.

An existence theorem of nash equilibrium in Coq and Isabelle

Nash equilibrium (NE) is a central concept in game theory. Here we prove formally a published theorem on existence of an NE in two proof assistants, Coq and Isabelle: starting from a game with finitely many outcomes, one may derive a game by rewriting each of these outcomes with either of two basic outcomes, namely that Player 1 wins or that Player 2 wins. If all ways of deriving such a win/lose game lead to a game where one player has a winning strategy, the original game also has a Nash equilibrium. This article makes three other contributions: first, while the original proof invoked linear extension of strict partial orders, here we avoid it by generalizing the relevant definition. Second, we notice that the theorem also implies the existence of a secure equilibrium, a stronger version of NE that was introduced for model checking. Third, we also notice that the constructive proof of the theorem computes secure equilibria for non-zero-sum priority games (generalizing parity games) in quasi-polynomial time.

Reduction Techniques for Model Checking and Learning in MDPs

Omega-regular objectives in Markov decision processes (MDPs) reduce to reachability: find a policy which maximizes the probability of reaching a target set of states. Given an MDP, an initial distribution, and a target set of states, such a policy can be computed by most probabilistic model checking tools. If the MDP is only partially specified, i.e., some probabilities are unknown, then model-learning techniques can be used to statistically approximate the probabilities and enable the computation of the desired policy. For fully specified MDPs, reducing the size of the MDP translates into faster model checking; for partially specified MDPs, into faster learning. We provide reduction techniques that allow us to remove irrelevant transition probabilities: transition probabilities (known, or to be learned) that do not influence the maximal reachability probability. Among other applications, these reductions can be seen as a pre-processing of MDPs before model checking or as a way to reduce the number of experiments required to obtain a good approximation of an unknown MDP.