Sampling from the Gibbs Distribution in Congestion Games

One of the most widely used solution concepts for strategic games is the concept of Nash equilibrium. A Nash equilibrium is a state in which no player can improve its objective by unilaterally changing its strategy. A Nash equilibrium is called pure if all players choose a pure strategy, and mixed if players choose probability distributions over strategies. Of special interest are fully mixed Nash equilibria where each player chooses each strategy with non-zero probability. Rosenthal [12] introduced a special class of strategic games, now widely known as congestion games. Here, the strategy set of each player is a subset of the power set of given resources. The players share a private objective function, defined as the sum (over their chosen resources) of functions in the number of players sharing this resource. In his seminal work, Rosenthal showed with help of a potential function that congestion games (in sharp contrast to general strategic games) always admit at least one pure Nash equilibrium. An extension to congestion games are weighted congestion games, in which the players have weights and thus different influence on the congestion of the resources. In (weighted) network congestion games the strategy sets of the players correspond to paths in a network. In order to measure the degradation of social welfare due to the selfish behavior of the players, Koutsoupias and Papadimitriou [8] introduced a global objective function, usually coined as social cost. They defined the price of anarchy, also called coordination ratio, as the worst-case ratio between the value of social cost in a Nash equilibrium and that of some social optimum. Thus the coordination ratio measures the extent to which non-cooperation approximates cooperation. As a starting point for studying the coordination ratio, Koutsoupias and Papadimitriou considered a very simple weighted network congestion game, now known as KP-model. Here, the network consists of a single source and a single destination which are connected by parallel links. The load on a link is the total weight of players assigned to this link. Associated with each link is a capacity representing the rate at which the link processes load. Each of the players selfishly routes from the source to the destination by choosing a probability distribution over the links. The private objective function of a player is defined as its expected latency. In the KP-model the social cost is defined as the expected maximum latency on a link, where the expectation is taken over all random choices of the players. In this paper, we give a thorough survey on the most exciting results on finite (weighted) congestion games and on special classes which are related to the KP-model. In particular, we review the findings on the existence and computational complexity of pure Nash equilibria. Furthermore, we discuss results on the price of anarchy. Last but not least, we survey known facts on fully mixed Nash equilibria.

We study the inefficiency of equilibria for congestion games when players are (partially) altruistic. We model altruistic behavior by assuming that player i 's perceived cost is a convex combination of α i times his direct cost and α i times the social cost. Tuning the parameters α i allows smooth interpolation between purely selfish and purely altruistic behavior. Within this framework, we study primarily altruistic extensions of (atomic and nonatomic) congestion games, but also obtain some results on fair cost-sharing games and valid utility games. We derive (tight) bounds on the price of anarchy of these games for several solution concepts. Thereto, we suitably adapt the smoothness notion introduced by Roughgarden and show that it captures the essential properties to determine the robust price of anarchy of these games. Our bounds show that for atomic congestion games and cost-sharing games, the robust price of anarchy gets worse with increasing altruism, while for valid utility games, it remains constant and is not affected by altruism. However, the increase in the price of anarchy is not a universal phenomenon: For general nonatomic congestion games with uniform altruism, the price of anarchy improves with increasing altruism. For atomic and nonatomic symmetric singleton congestion games, we derive bounds on the pure price of anarchy that improve as the average level of altruism increases. (For atomic games, we only derive such bounds when cost functions are linear.) Since the bounds are also strictly lower than the robust price of anarchy, these games exhibit natural examples in which pure Nash equilibria are more efficient than more permissive notions of equilibrium.

We study convergence of ?-Nash dynamics in congestion games when delay functions of all resources are similar. Delay functions are said to be similar if their values are within a polynomial factor at every congestion level. We show that for any ?> 0, an ?-Nash dynamics in symmetric congestion games with similar resources converge in steps polynomial in 1/? and the number of players are resources, yielding an FPTAS. Our result can be contrasted with that of Chien and Sinclair [3], which showed polynomial convergence result for symmetric congestion games where the delay functions have polynomially bounded jumps. Our assumption of similar delay functions is orthogonal to that of bounded jumps in that neither assumption implies the other. Our convergence result also hold for several natural variants of ?-Nash dynamics, including the most general polynomial liveness dynamics, where each player is given a chance to move frequently enough. We also extend our positive results to give an FPTAS for computing equilibrium in asymmetric games with similar resources, in which players share k distinct strategy spaces for any constant k. We complement our positive results by showing that computing an exact pure Nash equilibrium in symmetric congestion game with similar resources is PLS-complete. Furthermore, we show that for any ?> 0, all sequences of ?-Nash dynamics takes exponential steps to reach an approximate equilibrium in general congestion games with similar resources, as well as in symmetric congestion games with two groups of similar resources.

We characterize the price of anarchy in weighted congestion games, as a function of the allowable resource cost functions. Our results provide as thorough an understanding of this quantity as is already known for nonatomic and unweighted congestion games, and take the form of universal (cost function-independent) worst-case examples. One noteworthy byproduct of our proofs is the fact that weighted congestion games are “tight”, which implies that the worst-case price of anarchy with respect to pure Nash, mixed Nash, correlated, and coarse correlated equilibria are always equal (under mild conditions on the allowable cost functions). Another is the fact that, like nonatomic but unlike atomic (unweighted) congestion games, weighted congestion games with trivial structure already realize the worst-case POA, at least for polynomial cost functions.We also prove a new result about unweighted congestion games: the worst-case price of anarchy in symmetric games is, as the number of players goes to infinity, as large as in their more general asymmetric counterparts.KeywordsCost FunctionNash EquilibriumCongestion GamePure Nash EquilibriumSymmetric GameThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

We investigate the speed of convergence of best response dynamics to approximately optimal solutions in weighted congestion games with polynomial delay functions. Awerbuch et al. (Fast convergence to nearly optimal solutions in potential games. ACM Conference on Electronic Commerce, 2008) showed that the convergence time of such dynamics to Nash equilibrium may be exponential in the number of players n even for unweighted players and linear delay functions. Nevertheless, we show that Θ(n log log W) (where W is the sum of all the players’ weights) best responses are necessary and sufficient to achieve states that approximate the optimal solution by a constant factor, under the assumption that every O(n) steps each player performs a constant (and non-null) number of best responses. For congestion games in which computing a best response is a polynomial time solvable problem, such a dynamics naturally implies a polynomial time distributed algorithm for the problem of computing the social optimum in congestion games, approximating the optimal solution by a constant factor.

Rosenthal's congestion games constitute one of the few known classes of noncooperative games possessing pure-strategy Nash equilibria. In the network version, each player wants to route one unit of flow on a single path from her origin to her destination at minimum cost, and the cost of using an arc depends only on the total number of players using that arc. A natural extension is to allow for players controlling different amounts of flow, which results in so-called weighted congestion games. While examples have been exhibited showing that pure-strategy Nash equilibria need not exist anymore, we prove that it is actually strongly NP-hard to determine whether a given weighted network congestion game has a pure-strategy Nash equilibrium. This is true regardless of whether flow is unsplittable or not. In the unsplittable case, the problem remains strongly NP-hard for a fixed number of players. In addition to congestion games, we provide complexity results on the existence and computability of pure-strategy Nash equilibria for the closely related family of bidirectional local-effect games. Therein, the cost of a player taking a particular action depends not only on the number of players choosing the same action, but also on the number of players settling for (locally) related actions.

What is the price of anarchy when unsplittable demands are routed selfishly in general networks with load-dependent edge delays? Motivated by this question we generalize the model of [14] to the case of weighted congestion games. We show that varying demands of users crucially affect the nature of these games, which are no longer isomorphic to exact potential games, even for very simple instances. Indeed we construct examples where even a single-commodity (weighted) network congestion game may have no pure Nash equilibrium.On the other hand, we study a special family of networks (which we call the l-layered networks) and we prove that any weighted congestion game on such a network with resource delays equal to the congestions, possesses a pure Nash Equilibrium. We also show how to construct one in pseudo-polynomial time. Finally, we give a surprising answer to the question above for such games: The price of anarchy of any weighted ℓ-layered network congestion game with m edges and edge delays equal to the loads, is \(\Theta\left(\frac{\log m}{\log\log m}\right)\).KeywordsNash EquilibriumMixed StrategyLayered NetworkMaximum LatencyDelay FunctionThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Selfish unsplittable flows

We extend the study of the complexity of computing an ε-approximate Nash equilibrium in symmetric congestion games from the case of positive delay functions to delays of arbitrary sign. Our results show that with this extension the complexity has a richer structure, and it depends on the exact nature of the signs allowed. We first prove that in symmetric games with increasing delay functions and with α-bounded jump the ε-Nash dynamic converges in polynomial time when all delays are negative, similarly to the case of positive delays. We are able to extend this result to monotone delay functions. We then establish a hardness result for symmetric games with increasing delay functions and with α-bounded jump when the delays can be both positive and negative: in that case computing an ε-approximate Nash equilibrium becomes PLS-complete, even if each delay function is of constant sign or of constant absolute value.KeywordsNash EquilibriumPolynomial TimePure StrategyHardness ResultDelay FunctionThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

We give exponential lower bounds on the Price of Stability (PoS) of weighted congestion games with polynomial cost functions. In particular, for any positive integer $d$ we construct rather simple games with cost functions of degree at most $d$ which have a PoS of at least $\varOmega(Φ_d)^{d+1}$, where $Φ_d\sim d/\ln d$ is the unique positive root of equation $x^{d+1}=(x+1)^d$. This almost closes the huge gap between $\varTheta(d)$ and $Φ_d^{d+1}$. Our bound extends also to network congestion games. We further show that the PoS remains exponential even for singleton games. More generally, we provide a lower bound of $\varOmega((1+1/α)^d/d)$ on the PoS of $α$-approximate Nash equilibria for singleton games. All our lower bounds hold for mixed and correlated equilibria as well. On the positive side, we give a general upper bound on the PoS of $α$-approximate Nash equilibria, which is sensitive to the range $W$ of the player weights and the approximation parameter $α$. We do this by explicitly constructing a novel approximate potential function, based on Faulhaber's formula, that generalizes Rosenthal's potential in a continuous, analytic way. From the general theorem, we deduce two interesting corollaries. First, we derive the existence of an approximate pure Nash equilibrium with PoS at most $(d+3)/2$; the equilibrium's approximation parameter ranges from $\varTheta(1)$ to $d+1$ in a smooth way with respect to $W$. Secondly, we show that for unweighted congestion games, the PoS of $α$-approximate Nash equilibria is at most $(d+1)/α$.

In this paper we initiate the study of how collusion alters the quality of solutions obtained in competitive games. The price of anarchy aims to measure the cost of the lack of coordination by comparing the quality of a Nash equilibrium to that of a centrally designed optimal solution. This notion assumes that players act not only selfishly, but also independently. We propose a framework for modeling groups of colluding players, in which members of a coalition cooperate so as to selfishly maximize their collective welfare. Clearly, such coalitions can improve the social welfare of the participants, but they can also harm the welfare of those outside the coalition. One might hope that the improvement for the coalition participants outweighs the negative effects on the others. This would imply that increased cooperation can only improved the overall solution quality of stable outcomes. However, increases in coordination can actually lead to significant decreases in total social welfare. In light of this, we propose the price of collusion as a measure of the possible negative effect of collusion, specifying the factor by which solution quality can deteriorate in the presence of coalitions. We give examples to show that the price of collusion can be arbitrarily high even in convex games. Our main results show that in the context of load-balancing games, the price of collusion depends upon the disparity in market power among the game participants. We show that in some symmetric nonatomic games (where all users have access to the same set of strategies) increased cooperation always improves the solution quality, and in the discrete analogs of such games, the price of collusion is bounded by two.

This chapter deals with multi-agent systems whose objective is modeled by means of potential games and which are endowed with oracle-based information. As it was discussed in the previous chapter,

In this thesis we explore extremal graph theory, focusing on new methods which apply to different notions of regular graph. The first notion is dregularity, which means each vertex of a graph is contained in exactly d edges, and the second notion is Szemeredi regularity, which is a strong, approximate version of this property that relates to pseudorandomness. We begin with a novel method for optimising observables of Gibbs distributions in sparse graphs. The simplest application of the method is to the hard-core model, concerning independent sets in d-regular graphs, where we prove a tight upper bound on an observable known as the occupancy fraction. We also cover applications to matchings and colourings, in each case proving a tight bound on an observable of a Gibbs distribution and deriving an extremal result on the number of a relevant combinatorial structure in regular graphs. The results relate to a wide range of topics including statistical physics and Ramsey theory. We then turn to a form of Szemeredi regularity in sparse hypergraphs, and develop a method for embedding complexes that generalises a widely-applied method for counting in pseudorandom graphs. We prove an inheritance lemma which shows that the neighbourhood of a sparse, regular subgraph of a highly pseudorandom hypergraph typically inherits regularity in a natural way. This shows that we may embed complexes into suitable regular hypergraphs vertex-by-vertex, in much the same way as one can prove a counting lemma for regular graphs. Finally, we consider the multicolour Ramsey number of paths and even cycles. A well-known density argument shows that when the edges of a complete graph on kn vertices are coloured with k colours, one can find a monochromatic path on n vertices. We give an improvement to this bound by exploiting the structure of the densest colour, and use the regularity method to extend the result to even cycles.

Following this advice, we should look over the big questions, then doodle and sketch out some approaches. If you are an expert, then this is easy to do, but most people do not want to wait to become an expert before looking at interesting problems. Graph theory, our area of expertise, has many hard-to-solve questions. Some hark back to the recreational roots of the area yet still keep their mystery. These ‘‘acorns’’ can be planted on the backs of envelopes, on a blackboard, and over a coffee. Our goal is to collect some of these conjectures—arguably some of the most intriguing—in one place. We present ten conjectures in graph theory, and you can read about each one in at most ten minutes. As we live in the era of Twitter, all the conjectures we state are 140 characters or fewer (so ‘‘minute’’ here has a double meaning). We might even call these sketchy tweets, as we present examples for each conjecture that you can doodle on as you read. Hamming also references ambiguity: good researchers can work both on proving and disproving the same statement, so we approach the conjectures with an open mind. He also mentions that a good approach is to reframe the problem, and change the point of view. One example from Vizing’s Conjecture (which is discussed as our second-to-last conjecture in the following text), is the three-page paper [2] which, with a new way of thinking, reduced most of the published work of twenty years to a corollary of its main result! Given the size of modern graph theory, with its many smaller subfields (such as structural graph theory, random graphs, topological graph theory, graph algorithms, spectral graph theory, graph minors, and graph homomorphisms, to name a few), it would be impossible to list all, or even the bulk of the conjectures in the field. We are content instead to focus on a few family jewels, which have an intrinsic beauty and have provided some challenges for graph-theorists for at least two decades. There is something for everyone here, from undergraduate students taking their first course in graph theory, to seasoned researchers in the field. For additional reading on problems and conjectures in graph theory and other fields, see the Open Problem Garden maintained by IRMACS at Simon Fraser University [24]. We consider only finite and undirected graphs, with no multiple edges or loops (unless otherwise stated). We assume the reader has some basic familiarity with graphs and their terminology, including notions such as cycles, paths, complete graphs, complete bipartite graphs, vertex degrees, and connected graphs. We use the notation Cn for the cycle with n vertices, Pn for the path with n vertices, and Kn for the complete graph with n vertices. The complete bipartite graph with m and n vertices of the respective colors is denoted by Km,n. All the background we need can be found in any text in graph theory, such as those of Diestel [9] and West [41], or online (see for example [42]).

The Maxwell–Boltzmann (MB) distribution is important because it provides the statistical foundation for connecting microscopic particle motion to macroscopic gas properties by statistically describing molecular speeds and energies, making it essential for understanding and predicting the behavior of classical ideal gases. This study advances the statistical modeling of lifetime distributions by developing a comprehensive reliability analysis of the MB distribution under an improved adaptive progressive censoring framework. The proposed scheme strategically enhances experimental flexibility by dynamically adjusting censoring protocols, thereby preserving more information from test samples compared to conventional designs. Maximum likelihood estimation, interval estimation, and Bayesian inference are rigorously derived for the MB parameters, with asymptotic properties established to ensure methodological soundness. To address computational challenges, Markov chain Monte Carlo algorithms are employed for efficient Bayesian implementation. A detailed exploration of reliability measures—including hazard rate, mean residual life, and stress–strength models—demonstrates the MB distribution’s suitability for complex reliability settings. Extensive Monte Carlo simulations validate the efficiency and precision of the proposed inferential procedures, highlighting significant gains over traditional censoring approaches. Finally, the utility of the methodology is showcased through real-world applications to physics and engineering datasets, where the MB distribution coupled with such censoring yields superior predictive performance. This genuine examination is conducted through two datasets (including the failure times of aircraft windshields, capturing degradation under extreme environmental and operational stress, and mechanical component failure times) that represent recurrent challenges in industrial systems. This work contributes a unified statistical framework that broadens the applicability of the Maxwell–Boltzmann model in reliability contexts and provides practitioners with a powerful tool for decision making under censored data environments.