Searcharxiv⌕ Search

arXiv subjects

János Flesch

Publications and source records attributed to János Flesch.

At least 19 recordsLinked to original sources
arXiv2026

Optimal strategies in Markov decision processes with finitely additive evaluations

We study infinite-horizon Markov decision processes (MDPs) where the decision maker evaluates each of her strategies by aggregating the infinite stream of expected stage-rewards. The crucial feature of our approach is that the aggregation is performed by means of a given diffuse charge (a diffuse finitely additive probability measure) on the set of stages. The results of Neyman [2023] imply that in this setting, in every MDP with finite state and action spaces, the decision maker has a pure optimal strategy as long as the diffuse charge satisfies the time value of money principle. His result raises the question of existence of an optimal strategy without additional assumptions on the aggregation charge. We answer this question in the negative with a counterexample. With a delicately constructed aggregation charge, the MDP has no optimal strategy at all, neither pure nor randomized.

math.OC↗
arXiv2025

Games with infinite past

We study multi-player games with perfect information and general payoff function, where the set of stages is the set of non-positive integers $\{\ldots,-2,-1,0\}$. We define two related equilibrium concepts: one considering only deviations at finitely many stages and another considering all deviations. We show that (i) The sets of equilibrium plays coincide for the two equilibrium concepts, provided that at least two players are active along each infinite play. (ii) In win-lose games, the game has an equilibrium if the winning sets have Borel-rank at most 2, and we provide a counter-example showing that this is no longer true for Borel-rank 3. (iii) In general non-zero-sum games, the game has an equilibrium if the payoff functions are continuous, for example, with reversed-time discounted payoffs. The challenge for all these results is that not all strategy profiles admit a consistent infinite play, hampering the use of backward induction arguments.

math.OC↗
arXiv2025

A general definition of perfect equilibrium

We propose a general definition of perfect equilibrium which is applicable to a wide class of games. A key feature is the concept of completely mixed nets of strategies, based on a more detailed notion of carrier of a strategy. Under standard topological conditions, this definition yields a nonempty and compact set of perfect equilibria. For finite action sets, our notion of perfect equilibrium coincides with Selten's (1975) original notion. In the compact-continuous case, perfect equilibria are weak perfect equilibria in the sense of Simon and Stinchcombe (1995). In the finitely additive case, perfect equilibria in the sense of Marinacci (1997) are perfect. Under mild conditions, perfect equilibrium meets game-theoretic desiderata such as limit undominatedness and invariance. We provide a variety of examples to motivate and illustrate our definition. Notably, examples include applications to games with discontinuous payoffs and games played with finitely additive strategies.

econ.TH↗
arXiv2024

Zero-one Laws for a Control Problem with Random Action Sets

In many control problems there is only limited information about the actions that will be available at future stages. We introduce a framework where the Controller chooses actions $a_{0}, a_{1}, \ldots$, one at a time. Her goal is to maximize the probability that the infinite sequence $(a_{0}, a_{1}, \ldots)$ is an element of a given subset $G$ of $\mathbb{N}^{\mathbb{N}}$. The set $G$, called the goal, is assumed to be a Borel tail set. The Controller's choices are restricted: having taken a sequence $h_{t} = (a_{0}, \ldots, a_{t-1})$ of actions prior to stage $t \in \mathbb{N}$, she must choose an action $a_{t}$ at stage $t$ from a non-empty, finite subset $A(h_{t})$ of $\mathbb{N}$. The set $A(h_{t})$ is chosen from a distribution $p_{t}$, independently over all $t \in \mathbb{N}$ and all $h_{t} \in \mathbb{N}^{t}$. We consider several information structures defined by how far ahead into the future the Controller knows what actions will be available. In the special case where all the action sets are singletons (and thus the Controller is a dummy), Kolmogorov's 0-1 law says that the probability for the goal to be reached is 0 or 1. We construct a number of counterexamples to show that in general the value of the control problem can be strictly between 0 and 1, and derive several sufficient conditions for the 0-1 ``law" to hold.

math.OC↗
arXiv2024

Approximations of expectations under infinite product measures

We are given a bounded Borel-measurable real-valued function on a product of countably many Polish spaces, and a product probability measure. We are interested in points in the product space that can be used to approximate the expected value of this function. We define two notions. A point is called a weak $ε$-approximation, where $ε\geq 0$, if the Dirac measure on this point, except in finitely many coordinates where another measure can be taken, gives an expected value that is $ε$-close to the original expected value. A point is called a strong $ε$-approximation if the same holds under the restriction that in those finitely many coordinates the measure is equal to the original one. We prove that both the set of weak 0-approximation points and the set of strong $ε$-approximation points, for any $ε>0$, have measure 1 under the original measure. Finally, we provide two applications: (i) in Game Theory on the minmax guarantee levels of the players in games with infinitely many players, and (ii) in Decision Theory on the set of feasible expected payoffs in infinite duration problems.

math.PR↗
arXiv2022

Stochastic Games with General Payoff Functions

We consider multiplayer stochastic games in which the payoff of each player is a bounded and Borel-measurable function of the infinite play. By using a generalization of the technique of Martin (1998) and Maitra and Sudderth (1998), we show four different existence results. In each stochastic game, it holds for every $ε>0$ that (i) each player has a strategy that guarantees in each subgame that this player's payoff is at least her maxmin value up to $ε$, (ii) there exists a strategy profile under which in each subgame each player's payoff is at least her minmax value up to $ε$, (iii) the game admits an extensive-form correlated $ε$-equilibrium, and (iv) there exists a subgame that admits an $ε$-equilibrium.

math.OC↗
arXiv2022

Absorbing Blackwell Games

It was shown in Flesch and Solan (2022) with a rather involved proof that all two-player stochastic games with finite state and action spaces and shift-invariant payoffs admit an $ε$-equilibrium, for every $ε>0$. Their proof also holds for two-player absorbing games with tail-measurable payoffs. In this paper we provide a simpler proof for the existence of $ε$-equilibrium in two-player absorbing games with tail-measurable payoffs, by combining recent mathematical tools for such payoff functions with classical tools for absorbing games.

math.OC↗
arXiv2022

Repeated Games with Tail-Measurable Payoffs

We study multiplayer Blackwell games, which are repeated games where the payoff of each player is a bounded and Borel-measurable function of the infinite stream of actions played by the players during the game. These games are an extension of the two-player perfect-information games studied by David Gale and Frank Stewart (1953). Recently, various new ideas have been discovered to study Blackwell games. In this paper, we give an overview of these ideas by proving, in four different ways, that Blackwell games with a finite number of players, finite action sets, and tail-measurable payoffs admit an $\varepsilon$-equilibrium, for all $\varepsilon>0$.

math.OC↗
arXiv2022

Regularity of the minmax value and equilibria in multiplayer Blackwell games

A real-valued function $φ$ that is defined over all Borel sets of a topological space is \emph{regular} if for every Borel set $W$, $φ(W)$ is the supremum of $φ(C)$, over all closed sets $C$ that are contained in $W$, and the infimum of $φ(O)$, over all open sets $O$ that contain $W$. We study Blackwell games with finitely many players. We show that when each player has a countable set of actions and the objective of a certain player is represented by a Borel winning set, that player's minmax value is regular. We then use the regularity of the minmax value to establish the existence of $\varepsilon$-equilibria in two distinct classes of Blackwell games. One is the class of $n$-player Blackwell games where each player has a finite action space and an analytic winning set, and the sum of the minmax values over the players exceeds $n-1$. The other class is that of Blackwell games with bounded upper semi-analytic payoff functions, history-independent finite action spaces, and history-independent minmax values. For the latter class, we obtain a characterization of the set of equilibrium payoffs.

math.OC↗
arXiv2021

Random perfect information games

The paper proposes a natural measure space of zero-sum perfect information games with upper semicontinuous payoffs. Each game is specified by the game tree, and by the assignment of the active player and of the capacity to each node of the tree. The payoff in a game is defined as the infimum of the capacity over the nodes that have been visited during the play. The active player, the number of children, and the capacity are drawn from a given joint distribution independently across the nodes. We characterize the cumulative distribution function of the value $v$ using the fixed points of the so-called value generating function. The characterization leads to a necessary and sufficient condition for the event $v \geq k$ to occur with positive probability. We also study probabilistic properties of the set of Player I's $k$-optimal strategies and the corresponding plays.

cs.GT↗
arXiv2020

Games characterizing limsup functions and Baire class 1 functions

We consider a real-valued function $f$ defined on the set of infinite branches $X$ of a countably branching pruned tree $T$. The function $f$ is said to be a \textit{limsup function} if there is a function $u \colon T \to \mathbb{R}$ such that $f(x) = \limsup_{t \to \infty} u(x_{0},\dots,x_{t})$ for each $x \in X$. We study a game characterization of limsup functions, as well as a novel game characterization of functions of Baire class 1.

math.GN↗
arXiv2020

A competitive search game with a moving target

We introduce a discrete-time search game, in which two players compete to find an object first. The object moves according to a time-varying Markov chain on finitely many states. The players know the Markov chain and the initial probability distribution of the object, but do not observe the current state of the object. The players are active in turns. The active player chooses a state, and this choice is observed by the other player. If the object is in the chosen state, this player wins and the game ends. Otherwise, the object moves according to the Markov chain and the game continues at the next period. We show that this game admits a value, and for any error-term $\veps>0$, each player has a pure (subgame-perfect) $\veps$-optimal strategy. Interestingly, a 0-optimal strategy does not always exist. The $\veps$-optimal strategies are robust in the sense that they are $2\veps$-optimal on all finite but sufficiently long horizons, and also $2\veps$-optimal in the discounted version of the game provided that the discount factor is close to 1. We derive results on the analytic and structural properties of the value and the $\veps$-optimal strategies. Moreover, we examine the performance of the finite truncation strategies, which are easy to calculate and to implement. We devote special attention to the important time-homogeneous case, where additional results hold.

cs.GT↗
arXiv2020

Search for a moving target in a competitive environment

We consider a discrete-time dynamic search game in which a number of players compete to find an invisible object that is moving according to a time-varying Markov chain. We examine the subgame perfect equilibria of these games. The main result of the paper is that the set of subgame perfect equilibria is exactly the set of greedy strategy profiles, i.e. those strategy profiles in which the players always choose an action that maximizes their probability of immediately finding the object. We discuss various variations and extensions of the model.

math.OC↗
arXiv2020

Incentive compatibility in sender-receiver stopping games

We introduce a model of sender-receiver stopping games, where the state of the world follows an iid--process throughout the game. At each period, the sender observes the current state, and sends a message to the receiver, suggesting either to stop or to continue. The receiver, only seeing the message but not the state, decides either to stop the game, or to continue which takes the game to the next period. The payoff to each player is a function of the state when the receiver quits, with higher states leading to better payoffs. The horizon of the game can be finite or infinite. We prove existence and uniqueness of responsive (i.e. non-babbling) Perfect Bayesian Equilibrium (PBE) under mild conditions on the game primitives in the case where the players are sufficiently patient. The responsive PBE has a remarkably simple structure, which builds on the identification of an easy-to-implement and compute class of threshold strategies for the sender. With the help of these threshold strategies, we derive simple expressions describing this PBE. It turns out that in this PBE the receiver obediently follows the recommendations of the sender. Hence, surprisingly, the sender alone plays the decisive role, and regardless of the payoff function of the receiver the sender always obtains the best possible payoff for himself.

cs.GT↗
arXiv2020

The doubling metric and doubling measures

We introduce the so--called doubling metric on the collection of non--empty bounded open subsets of a metric space. Given a subset $U$ of a metric space $X$, the predecessor $U_{*}$ of $U$ is defined by doubling the radii of all open balls contained inside $U$, and taking their union. If $U$ is open, the predecessor of $U$ is an open set containing $U$. The directed doubling distance between $U$ and another subset $V$ is the number of times that the predecessor operation needs to be applied to $U$ to obtain a set that contains $V$. Finally, the doubling distance between $U$ and $V$ is the maximum of the directed distance between $U$ and $V$ and the directed distance between $V$ and $U$.

math.GN↗
arXiv2019

Reachability and safety objectives in Markov decision processes on long but finite horizons

We consider discrete-time Markov decision processes in which the decision maker is interested in long but finite horizons. First we consider reachability objective: the decision maker's goal is to reach a specific target state with the highest possible probability. Formally, strategy $σ$ overtakes another strategy $σ'$, if the probability of reaching the target state within horizon $t$ is larger under $σ$ than under $σ'$, for all sufficiently large $t\in\NN$. We prove that there exists a pure stationary strategy that is not overtaken by any pure strategy nor by any stationary strategy, under some condition on the transition structure and respectively under genericity. A strategy that is not overtaken by any other strategy, called an overtaking optimal strategy, does not always exist. We provide sufficient conditions for its existence. Next we consider safety objective: the decision maker's goal is to avoid a specific state with the highest possible probability. We argue that the results proven for reachability objective extend to this model. We finally discuss extensions of our results to two-player zero-sum perfect information games.

math.OC↗
arXiv2016

Approachability of convex sets in generalized quitting games

We consider Blackwell approachability, a very powerful and geometric tool in game theory, used for example to design strategies of the uninformed player in repeated games with incomplete information. We extend this theory to "generalized quitting games" , a class of repeated stochastic games in which each player may have quitting actions, such as the Big-Match. We provide three simple geometric and strongly related conditions for the weak approachability of a convex target set. The first is sufficient: it guarantees that, for any fixed horizon, a player has a strategy ensuring that the expected time-average payoff vector converges to the target set as horizon goes to infinity. The third is necessary: if it is not satisfied, the opponent can weakly exclude the target set. In the special case where only the approaching player can quit the game (Big-Match of type I), the three conditions are equivalent and coincide with Blackwell's condition. Consequently, we obtain a full characterization and prove that the game is weakly determined-every convex set is either weakly approachable or weakly excludable. In games where only the opponent can quit (Big-Match of type II), none of our conditions is both sufficient and necessary for weak approachability. We provide a continuous time sufficient condition using techniques coming from differential games, and show its usefulness in practice, in the spirit of Vieille's seminal work for weak approachability.Finally, we study uniform approachability where the strategy should not depend on the horizon and demonstrate that, in contrast with classical Blackwell approacha-bility for convex sets, weak approachability does not imply uniform approachability.

cs.GT↗