arXiv subjects

Diogo A. Gomes

Publications and source records attributed to Diogo A. Gomes.

31 records · Page 2Linked to original sources

arXiv2014

Time dependent mean-field games in the superquadratic case

We investigate time-dependent mean-field games with superquadratic Hamiltonians and a power dependence on the measure. Such problems pose substantial mathematical challenges as the key techniques used in the subquadratic case do not extend to the superquadratic setting. Because of the superquadratic structure of the Hamiltonian, Lipschitz estimates for the solutions of the Hamilton-Jacobi equation are obtained through a novel set of techniques. These explore the parabolic nature of the problem through the non-linear adjoint method. Well-posedness is proved by combining Lipschitz regularity for the Hamilton-Jacobi equation with polynomial estimates for solutions of the Fokker-Planck equation. Existence of classical solutions can then be proved under conditions depending only on the growth of the Hamiltonian and the dimension. Our results also add to the current understanding of superquadratic Hamilton-Jacobi equations.

math.AP↗

arXiv2014

Existence for stationary mean field games with quadratic Hamiltonians with congestion

In this paper, we investigate the existence and uniqueness of solutions to a stationary mean field game model introduced by J.-M. Lasry and P.-L. Lions. This model features a quadratic Hamiltonian with possibly singular congestion effects. Thanks to a new class of a-priori bounds, combined with the continuation method, we prove the existence of smooth solutions in arbitrary dimensions.

math.AP↗

arXiv2014

Socio-economic applications of finite state mean field games

In this paper we present different applications of finite state mean field games to socio-economic sciences. Examples include paradigm shifts in the scientific community or the consumer choice behaviour in the free market. The corresponding finite state mean field game models are hyperbolic systems of partial differential equations, for which we present and validate different numerical methods. We illustrate the behaviour of solutions with various numerical experiments, which show interesting phenomena like shock formation. Hence we conclude with an investigation of the shock structure in the case of two-state problems.

math.AP↗

arXiv2013

Time dependent mean-field games in the subquadratic case

In this paper we consider time-dependent mean-field games with subquadratic Hamiltonians and power-like local dependence on the measure. We establish existence of classical solutions under a certain set of conditions depending on both the growth of the Hamiltonian and the dimension. This is done by combining regularity estimates for the Hamilton-Jacobi equation based on the Gagliardo-Nirenberg interpolation inequality with polynomial estimates for the Fokker-Planck equation. This technique improves substantially the previous results on the regularity of time-dependent mean-field games.

math.AP↗

arXiv2013

On the existence of classical solutions for stationary extended mean field games

In this paper we consider extended stationary mean field games, that is mean-field games which depend on the velocity field of the players. We prove various a-priori estimates which generalize the results for quasi-variational mean field games in [GPSM12]. In addition we use adjoint method techniques to obtain higher regularity bounds. Then we establish existence of smooth solutions under fairly general conditions by applying the continuity method. When applied to standard stationary mean-field games as in [LL06a], [GSM11] or [GPSM12] this paper yields various new estimates and regularity properties not available previously. We discuss additionally several examples where existence of classical solutions can be proved.

math.AP↗

arXiv2012

Continuous time finite state mean field games

In this paper we consider symmetric games where a large number of players can be in any one of d states. We derive a limiting mean field model and characterize its main properties. This mean field limit is a system of coupled ordinary differential equations with initial-terminal data. For this mean field problem we prove a trend to equilibrium theorem, that is convergence, in an appropriate limit, to stationary solutions. Then we study the $N+1$-player problem, which the mean field model attempts to approximate. Our main result is the convergence as $N\to \infty$ of the mean field model and an estimate of the rate of convergence. We end the paper with some further examples for potential mean field games.

math.OC↗

arXiv2011

Wigner measures and the semi-classical limit to the Aubry-Mather measure

In this paper we investigate the asymptotic behavior of the semi-classical limit of Wigner measures defined on the tangent bundle of the one-dimensional torus. In particular we show the convergence of Wigner measures to the Mather measure on the tangent bundle, for energy levels above the minimum of the effective Hamiltonian. The Wigner measures $μ_h$ we consider are associated to $ψ_h,$ a distinguished critical solution of the Evans' quantum action given by $ψ_h=a_h\,e^{i\frac{u_h}h}$, with $a_h(x)=e^{\frac{v^*_h(x)-v_h(x)}{2h}}$, $u_h(x)=P\cdot x+\frac{v^*_h(x)+v_h(x)}{2},$ and $v_h,v^*_h$ satisfying the equations -\frac{h\, Δv_h}{2}+ 1/2 \, | P + D v_h \,|^2 + V &= \bar{H}_h(P), \frac{h\, Δv_h^*}{2}+ 1/2 \, | P + D v_h^* \,|^2 + V &= \bar{H}_h(P), where the constant $\bar{H}_h(P)$ is the $h$ effective potential and $x$ is on the torus. L.\ C.\ Evans considered limit measures $|ψ_h|^2$ in $\mathbb{T}^n$, when $h\to 0$, for any $n\geq 1$. We consider the limit measures on the phase space $\mathbb{T}^n\times\mathbb{R}^n$, for $n=1$, and, in addition, we obtain rigorous asymptotic expansions for the functions $v_h$, and $v^*_h$, when $h\to 0$.

math.DS↗

arXiv2011

Mean field limit of a continuous time finite state game

Mean field games is a recent area of study introduced by Lions and Lasry in a series of seminal papers in 2006. Mean field games model situations of competition between large number of rational agents that play non-cooperative dynamic games under certain symmetry assumptions. They key step is to develop a mean field model, in a similar way that what is done in statistical physics in order to construct a mathematically tractable model. A main question that arises in the study of such mean field problems is the rigorous justification of the mean field models by a limiting procedure. In this paper we consider the mean field limit of two-state Markov decision problem as the number of players $N\to \infty$. First we establish the existence and uniqueness of a symmetric partial information Markov perfect equilibrium. Then we derive a mean field model and characterize its main properties. This mean field limit is a system of coupled ordinary differential equations with initial-terminal data. Our main result is the convergence as $N\to \infty$ of the $N$ player game to the mean field model and an estimate of the rate of convergence.

math.OC↗

arXiv2009

Minimax probabilities for Aubry-Mather Problems

In this paper we study minimax Aubry-Mather measures and its main properties. We consider first the discrete time problem and then the continuous time case. In the discrete time problem we establish existence, study some of the main properties using duality theory and present some examples. In the continuous time case, we establish both existence and non-existence results. First we give some examples that show that in continuous time stationary minimax Mather measures are either trivial or fail to exist. A more natural definition in continuous time are $T$-periodic minimax Mather measures. We give a complete characterization of these measures and discuss several examples.

math.DS↗

arXiv2009

Mather problem and viscosity solutions in the stationary setting

In this paper we discuss the Mather problem for stationary Lagrangians, that is Lagrangians $L:\Rr^n\times \Rr^n\times Ω\to \Rr$, where $Ω$ is a compact metric space on which $\Rr^n$ acts through an action which leaves $L$ invariant. This setting allow us to generalize the standard Mather problem for quasi-periodic and almost-periodic Lagrangians. Our main result is the existence of stationary Mather measures invariant under the Euler-Lagrange flow which are supported in a graph. We also obtain several estimates for viscosity solutions of Hamilton-Jacobi equations for the discounted cost infinite horizon problem.

math.AP↗

arXiv2009

Discrete mean field games

In this paper we study a mean field model for discrete time, finite number of states, dynamic games. These models arise in situations that involve a very large number of agents moving from state to state according to certain optimality criteria. The mean field approach for optimal control and differential games (continuous state and time) was introduced by Lasry and Lions. Here we consider a discrete version of the problem. Our setting is the following: we assume that there is a very large number of identical agents which can be in a finite number of states. Because the number of agents is very large, we assume the mean field hypothesis, that is, that the only relevant information for the global evolution is the fraction $π^n_i$ of players in each state $i$ at time $n$. The agents look for minimizing a running cost, which depends on $π$, plus a terminal cost $V^N$. In contrast with optimal control, where usually only the terminal cost $V^N$ is necessary to solve the problem, in mean-field games both the initial distribution of agents $π^0$ and the terminal cost $V^N$ are necessary to determine the solutions, that is, the distribution of players $π^n$ and value function $V^n$, for $0\leq n\leq N$. Because both initial and terminal data needs to be specified, we call this problem the initial-terminal value problem. Existence of solutions is non-trivial. Nevertheless, following the ideas of Lasry and Lions, we were able to establish existence and uniqueness, both for the stationary and for the initial-terminal value problems. In the last part of the paper we prove the main result of the paper, namely the exponential convergence to a stationary solution of $(π^0, V^0)$, as $N\to \infty$, for the initial-terminal value problem with (fixed) data $π^{-N}$ and $V^N$.

math.OC↗

arXiv2008

The Mather measure and a Large Deviation Principle for the Entropy Penalized Method

We present a large deviation principle for the entropy penalized Mather problem when the Lagrangian L is generic (in this case the Mather measure $μ$ is unique and the support of $μ$ is the Aubry set). Consider, for each value of $ε$ and h, the entropy penalized Mather problem $\min \{\int_{\tn\times\rn} L(x,v)dμ(x,v)+εS[μ]\},$ where the entropy S is given by $S[μ]=\int_{\tn\times\rn}μ(x,v)\ln\frac{μ(x,v)}{\int_{\rn}μ(x,w)dw}dxdv,$ and the minimization is performed over the space of probability densities $μ(x,v)$ that satisfy the holonomy constraint It follows from D. Gomes and E. Valdinoci that there exists a minimizing measure $μ_{ε, h}$ which converges to the Mather measure $μ$. We show a LDP $\lim_{ε,h\to0} ε\ln μ_{ε,h}(A),$ where $A\subset \mathbb{T}^N\times\mathbb{R}^N$. The deviation function I is given by $I(x,v)= L(x,v)+\nablaϕ_0(x)(v)-\bar{H}_{0},$ where $ϕ_0$ is the unique viscosity solution for L.

math.DS↗

arXiv2007

Exponential decay of correlation for the Stochastic Process associated to the Entropy Penalized Method

In this paper we present an upper bound for the decay of correlation for the stationary stochastic process associated with the Entropy Penalized Method. Let $L(x, v):\Tt^n\times\Rr^n\to \Rr$ be a Lagrangian of the form L(x,v) = {1/2}|v|^2 - U(x) + < P, v>. For each value of $ε$ and $h$, consider the operator \Gg[ϕ](x):= -εh {ln}[\int_{\re^N} e ^{-\frac{hL(x,v)+ϕ(x+hv)}{εh}}dv], as well as the reversed operator \bar \Gg[ϕ](x):= -εh {ln}[\int_{\re^N} e^{-\frac{hL(x+hv,-v)+ϕ(x+hv)}{εh}}dv], both acting on continuous functions $ϕ:\Tt^n\to \Rr$. Denote by $ϕ_{ε,h} $ the solution of $\Gg[ϕ_{ε,h}]=ϕ_{ε,h}+λ_{ε,h}$, and by $\bar ϕ_{ε,h} $ the solution of $\bar \Gg[ϕ_{ε,h}]=\bar ϕ_{ε,h}+λ_{ε,h}$. In order to analyze the decay of correlation for this process we show that the operator $ {\cal L} (ϕ) (x) = \int e^{- \frac{h L (x,v)}ε} ϕ(x+h v) d v,$ has a maximal eigenvalue isolated from the rest of the spectrum.

math.DS↗