arXiv subjects

John Fearnley

Publications and source records attributed to John Fearnley.

48 records · Page 3Linked to original sources

arXiv2014

The Complexity of the Simplex Method

The simplex method is a well-studied and widely-used pivoting method for solving linear programs. When Dantzig originally formulated the simplex method, he gave a natural pivot rule that pivots into the basis a variable with the most violated reduced cost. In their seminal work, Klee and Minty showed that this pivot rule takes exponential time in the worst case. We prove two main results on the simplex method. Firstly, we show that it is PSPACE-complete to find the solution that is computed by the simplex method using Dantzig's pivot rule. Secondly, we prove that deciding whether Dantzig's rule ever chooses a specific variable to enter the basis is PSPACE-complete. We use the known connection between Markov decision processes (MDPs) and linear programming, and an equivalence between Dantzig's pivot rule and a natural variant of policy iteration for average-reward MDPs. We construct MDPs and show PSPACE-completeness results for single-switch policy iteration, which in turn imply our main results for the simplex method.

cs.DS↗

arXiv2014

Learning Equilibria of Games via Payoff Queries

A recent body of experimental literature has studied empirical game-theoretical analysis, in which we have partial knowledge of a game, consisting of observations of a subset of the pure-strategy profiles and their associated payoffs to players. The aim is to find an exact or approximate Nash equilibrium of the game, based on these observations. It is usually assumed that the strategy profiles may be chosen in an on-line manner by the algorithm. We study a corresponding computational learning model, and the query complexity of learning equilibria for various classes of games. We give basic results for bimatrix and graphical games. Our focus is on symmetric network congestion games. For directed acyclic networks, we can learn the cost functions (and hence compute an equilibrium) while querying just a small fraction of pure-strategy profiles. For the special case of parallel links, we have the stronger result that an equilibrium can be identified while only learning a small fraction of the cost values.

cs.GT↗

arXiv2014

Finding Approximate Nash Equilibria of Bimatrix Games via Payoff Queries

We study the deterministic and randomized query complexity of finding approximate equilibria in bimatrix games. We show that the deterministic query complexity of finding an $ε$-Nash equilibrium when $ε< \frac{1}{2}$ is $Ω(k^2)$, even in zero-one constant-sum games. In combination with previous results \cite{FGGS13}, this provides a complete characterization of the deterministic query complexity of approximate Nash equilibria. We also study randomized querying algorithms. We give a randomized algorithm for finding a $(\frac{3 - \sqrt{5}}{2} + ε)$-Nash equilibrium using $O(\frac{k \cdot \log k}{ε^2})$ payoff queries, which shows that the $\frac{1}{2}$ barrier for deterministic algorithms can be broken by randomization. For well-supported Nash equilibria (WSNE), we first give a randomized algorithm for finding an $ε$-WSNE of a zero-sum bimatrix game using $O(\frac{k \cdot \log k}{ε^4})$ payoff queries, and we then use this to obtain a randomized algorithm for finding a $(\frac{2}{3} + ε)$-WSNE in a general bimatrix game using $O(\frac{k \cdot \log k}{ε^4})$ payoff queries. Finally, we initiate the study of lower bounds against randomized algorithms in the context of bimatrix games, by showing that randomized algorithms require $Ω(k^2)$ payoff queries in order to find a $\frac{1}{6k}$-Nash equilibrium, even in zero-one constant-sum games. In particular, this rules out query-efficient randomized algorithms for finding exact Nash equilibria.

cs.GT↗

arXiv2013

Time and Parallelizability Results for Parity Games with Bounded Tree and DAG Width

Parity games are a much researched class of games in NP intersect CoNP that are not known to be in P. Consequently, researchers have considered specialised algorithms for the case where certain graph parameters are small. In this paper, we study parity games on graphs with bounded treewidth, and graphs with bounded DAG width. We show that parity games with bounded DAG width can be solved in O(n^(k+3) k^(k + 2) (d + 1)^(3k + 2)) time, where n, k, and d are the size, treewidth, and number of priorities in the parity game. This is an improvement over the previous best algorithm, given by Berwanger et al., which runs in n^O(k^2) time. We also show that, if a tree decomposition is provided, then parity games with bounded treewidth can be solved in O(n k^(k + 5) (d + 1)^(3k + 5)) time. This improves over previous best algorithm, given by Obdrzalek, which runs in O(n d^(2(k+1)^2)) time. Our techniques can also be adapted to show that the problem of solving parity games with bounded treewidth lies in the complexity class NC^2, which is the class of problems that can be efficiently parallelized. This is in stark contrast to the general parity game problem, which is known to be P-hard, and thus unlikely to be contained in NC.

cs.GT↗

arXiv2013

Reachability in Two-Clock Timed Automata is PSPACE-complete

A recent result of Haase et al. has shown that reachability in two-clock timed automata is log-space equivalent to reachability in bounded one-counter automata. We show that reachability in bounded one-counter automata is PSPACE-complete.

cs.FL↗

arXiv2012

Synthesis of Succinct Systems

Synthesis of correct by design systems from specification has recently attracted much attention. The theoretical results imply that this problem is highly intractable, e.g., synthesizing a system is 2EXPTIME-complete for an LTL specification, and EXPTIME-complete for a CTL specification. However, an argument against it is that the temporal specification is highly compact, and the complexity reflects the large size of the system constructed. In that respect, the complexity should, perhaps, be specified relative to the size of the minimal satisfying system. A careful observation reveals that the size of the system is presented in such arguments as the size of its state space. This view is a bit nonstandard, in the sense that the state space can be exponentially larger than the size of a reasonable implementation such as a circuit or a program. Although this alternative measure of the size of the synthesized system is more intuitive (e.g., this is the standard way model checking problems are measured), research on synthesis has so far stayed with measuring the system in terms of the explicit state space. This raises the question of whether or not there always exists a small system. In this paper, we show that this is the case if, and only if, PSPACE = EXPTIME.

cs.FL↗

arXiv2012

Bounded Satisfiability for PCTL

While model checking PCTL for Markov chains is decidable in polynomial-time, the decidability of PCTL satisfiability, as well as its finite model property, are long standing open problems. While general satisfiability is an intriguing challenge from a purely theoretical point of view, we argue that general solutions would not be of interest to practitioners: such solutions could be too big to be implementable or even infinite. Inspired by bounded synthesis techniques, we turn to the more applied problem of seeking models of a bounded size: we restrict our search to implementable -- and therefore reasonably simple -- models. We propose a procedure to decide whether or not a given PCTL formula has an implementable model by reducing it to an SMT problem. We have implemented our techniques and found that they can be applied to the practical problem of sanity checking -- a procedure that allows a system designer to check whether their formula has an unexpectedly small model.

cs.LO↗

arXiv2011

Efficient Approximation of Optimal Control for Markov Games

We study the time-bounded reachability problem for continuous-time Markov decision processes (CTMDPs) and games (CTMGs). Existing techniques for this problem use discretisation techniques to break time into discrete intervals, and optimal control is approximated for each interval separately. Current techniques provide an accuracy of O(ε^2) on each interval, which leads to an infeasibly large number of intervals. We propose a sequence of approximations that achieve accuracies of O(ε^3), O(ε^4), and O(ε^5), that allow us to drastically reduce the number of intervals that are considered. For CTMDPs, the performance of the resulting algorithms is comparable to the heuristic approach given by Buckholz and Schulz, while also being theoretically justified. All of our results generalise to CTMGs, where our results yield the first practically implementable algorithms for this problem. We also provide positional strategies for both players that achieve similar error bounds.

cs.GT↗

arXiv2010

Playing Muller Games in a Hurry

This work studies the following question: can plays in a Muller game be stopped after a finite number of moves and a winner be declared. A criterion to do this is sound if Player 0 wins an infinite-duration Muller game if and only if she wins the finite-duration version. A sound criterion is presented that stops a play after at most 3^n moves, where n is the size of the arena. This improves the bound (n!+1)^n obtained by McNaughton and the bound n!+1 derived from a reduction to parity games.

cs.GT↗

arXiv2010

Exponential Lower Bounds For Policy Iteration

We study policy iteration for infinite-horizon Markov decision processes. It has recently been shown policy iteration style algorithms have exponential lower bounds in a two player game setting. We extend these lower bounds to Markov decision processes with the total reward and average-reward optimality criteria.

cs.DS↗

arXiv2010

Non-oblivious Strategy Improvement

We study strategy improvement algorithms for mean-payoff and parity games. We describe a structural property of these games, and we show that these structures can affect the behaviour of strategy improvement. We show how awareness of these structures can be used to accelerate strategy improvement algorithms. We call our algorithms non-oblivious because they remember properties of the game that they have discovered in previous iterations. We show that non-oblivious strategy improvement algorithms perform well on examples that are known to be hard for oblivious strategy improvement. Hence, we argue that previous strategy improvement algorithms fail because they ignore the structural properties of the game that they are solving.

cs.GT↗

arXiv2009

Linear Complementarity Algorithms for Infinite Games

The performance of two pivoting algorithms, due to Lemke and Cottle and Dantzig, is studied on linear complementarity problems (LCPs) that arise from infinite games, such as parity, average-reward, and discounted games. The algorithms have not been previously studied in the context of infinite games, and they offer alternatives to the classical strategy-improvement algorithms. The two algorithms are described purely in terms of discounted games, thus bypassing the reduction from the games to LCPs, and hence facilitating a better understanding of the algorithms when applied to games. A family of parity games is given, on which both algorithms run in exponential time, indicating that in the worst case they perform no better for parity, average-reward, or discounted games than they do for general P-matrix LCPs.

cs.GT↗