-
Towards Practical Second-Order Optimizers in Deep Learning: Insights from Fisher Information Analysis
Authors:
Damien Martins Gomes
Abstract:
First-order optimization methods remain the standard for training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by preconditioning the stochastic gradient with a diagonal matrix. Despite the widespread adoption of first-order methods, second-order optimization algorithms often exhibit superior convergence compared to methods like Adam and SGD. However,…
▽ More
First-order optimization methods remain the standard for training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by preconditioning the stochastic gradient with a diagonal matrix. Despite the widespread adoption of first-order methods, second-order optimization algorithms often exhibit superior convergence compared to methods like Adam and SGD. However, their practicality in training DNNs is still limited by a significantly higher per-iteration computational cost compared to first-order methods. In this thesis, we present AdaFisher, a novel adaptive second-order optimizer that leverages a diagonal block-Kronecker approximation of the Fisher information matrix to adaptively precondition gradients. AdaFisher aims to bridge the gap between the improved convergence and generalization of second-order methods and the computational efficiency needed for training DNNs. Despite the traditionally slower speed of second-order optimizers, AdaFisher is effective for tasks such as image classification and language modeling, exhibiting remarkable stability and robustness during hyperparameter tuning. We demonstrate that AdaFisher outperforms state-of-the-art optimizers in both accuracy and convergence speed. The code is available from https://github.com/AtlasAnalyticsLab/AdaFisher.
△ Less
Submitted 26 April, 2025;
originally announced April 2025.
-
Hessian Riemannian Flow For Multi-Population Wardrop Equilibrium
Authors:
Tigran Bakaryan,
Christoph Aoun,
Ricardo de Lima Ribeiro,
Naira Hovakimyan,
Diogo Gomes
Abstract:
In this paper, we address the problem of optimizing flows on generalized graphs that feature multiple entry points and multiple populations, each with varying cost structures. We tackle this problem by considering the multi-population Wardrop equilibrium, defined through variational inequalities. We rigorously analyze the existence and uniqueness of the Wardrop equilibrium. Furthermore, we introdu…
▽ More
In this paper, we address the problem of optimizing flows on generalized graphs that feature multiple entry points and multiple populations, each with varying cost structures. We tackle this problem by considering the multi-population Wardrop equilibrium, defined through variational inequalities. We rigorously analyze the existence and uniqueness of the Wardrop equilibrium. Furthermore, we introduce an efficient numerical method to find the solution. In particular, we reformulate the equilibrium problem as a distributed optimization problem over subgraphs and introduce a novel Hessian Riemannian flow method, a Riemannian-manifold-projected Hessian flow, to efficiently compute a solution. Finally, we demonstrate the effectiveness of our approach through examples in urban traffic management, including routing for diverse vehicle types and strategies for minimizing emissions in congested environments.
△ Less
Submitted 22 April, 2025;
originally announced April 2025.
-
The Riemannian geometry of the probability space of the unit circle
Authors:
André Magalhães de Sá Gomes,
Christian S. Rodrigues,
Luiz A. B. San Martin
Abstract:
This paper explores the Riemannian geometry of the Wasserstein space of the circle, namely $P(S^{1})$, the set of probability measures on the unit circle endowed with the 2-Wasserstein metric. Building on the foundational work of Otto, Lott, and Villani, the authors developed in another work an intrinsic framework for studying the differential geometry of Wasserstein spaces of compact Lie groups,…
▽ More
This paper explores the Riemannian geometry of the Wasserstein space of the circle, namely $P(S^{1})$, the set of probability measures on the unit circle endowed with the 2-Wasserstein metric. Building on the foundational work of Otto, Lott, and Villani, the authors developed in another work an intrinsic framework for studying the differential geometry of Wasserstein spaces of compact Lie groups, making use of the Peter-Weyl Theorem. This formalism allowed them to explicit an example in this paper. Key contributions include explicit computations of the Riemannian metric matrix coefficients, Lie brackets, and the Levi-Civita connection, along with its associated Christoffel symbols. The geodesic equations and curves with constant velocity fields are analysed, expliciting their PDEs. Notably, the paper demonstrates that $P(S^{1})$ is flat, with vanishing curvature. These results provide a comprehensive geometric understanding of $P(S^{1})$, connecting optimal transport theory and differential geometry, with potential applications in dynamical systems.
△ Less
Submitted 15 April, 2025;
originally announced April 2025.
-
An introduction to monotonicity methods in mean-field games
Authors:
Rita Ferreira,
Diogo Gomes,
Teruo Tada
Abstract:
This chapter examines monotonicity techniques in the theory of mean-field games(MFGs). Originally, monotonicity ideas were used to establish the uniqueness of solutions for MFGs. Later, monotonicity methods and monotone operators were further exploited to build numerical methods and to construct weak solutions under mild assumptions. Here, after a brief discussion on the mean-field game formulatio…
▽ More
This chapter examines monotonicity techniques in the theory of mean-field games(MFGs). Originally, monotonicity ideas were used to establish the uniqueness of solutions for MFGs. Later, monotonicity methods and monotone operators were further exploited to build numerical methods and to construct weak solutions under mild assumptions. Here, after a brief discussion on the mean-field game formulation, we introduce the Minty method and regularization strategies for PDEs. These are then used to address typical stationary and time-dependent monotone MFGs and to establish the existence of weak solutions for such MFGs.
△ Less
Submitted 27 February, 2025;
originally announced February 2025.
-
Weak-strong uniqueness for solutions to mean-field games
Authors:
Rita Ferreira,
Diogo Gomes,
Vardan Voskanyan
Abstract:
This paper addresses the crucial question of solution uniqueness in stationary first-order Mean-Field Games (MFGs). Despite well-established existence results, establishing uniqueness, particularly for weaker solutions in the sense of monotone operators, remains an open challenge. Building upon the framework of monotonicity methods, we introduce a linearization method that enables us to prove a we…
▽ More
This paper addresses the crucial question of solution uniqueness in stationary first-order Mean-Field Games (MFGs). Despite well-established existence results, establishing uniqueness, particularly for weaker solutions in the sense of monotone operators, remains an open challenge. Building upon the framework of monotonicity methods, we introduce a linearization method that enables us to prove a weak-strong uniqueness result for stationary MFG systems on the d-dimensional torus. In particular, we give explicit conditions under which this uniqueness holds.
△ Less
Submitted 27 February, 2025;
originally announced February 2025.
-
Regularity for Weak Solutions to First-Order Local Mean Field Games
Authors:
Abdulrahman Alharbi,
Diogo Gomes,
Giuseppe Di Fazio,
Melih Ucer
Abstract:
We establish interior regularity results for first-order, stationary, local mean-field game (MFG) systems. Specifically, we study solutions of the coupled system consisting of a Hamilton-Jacobi-Bellman equation $H(x, Du, m) = 0$ and a transport equation $-\operatorname{div}(m D_pH(x, Du, m)) = 0$ in a domain $Ω\subset \mathbb{R}^d$. Under suitable structural assumptions on the Hamiltonian $H$, wit…
▽ More
We establish interior regularity results for first-order, stationary, local mean-field game (MFG) systems. Specifically, we study solutions of the coupled system consisting of a Hamilton-Jacobi-Bellman equation $H(x, Du, m) = 0$ and a transport equation $-\operatorname{div}(m D_pH(x, Du, m)) = 0$ in a domain $Ω\subset \mathbb{R}^d$. Under suitable structural assumptions on the Hamiltonian $H$, without requiring monotonicity of the system, convexity of the Hamiltonian, separability in variables, or smoothness beyond basic continuity in $(p,m)$, we introduce a notion of weak solutions that allows the application of techniques from elliptic regularity theory. Our main contribution is to prove that the value function $u$ is locally Hölder continuous in $Ω$. The proof leverages the connection between first-order MFG systems and quasilinear equations in divergence form, adapting classical techniques to handle the specific structure of MFG systems.
△ Less
Submitted 17 November, 2024;
originally announced November 2024.
-
Convergence Analysis of a Schrodinger Problem with Moving Boundary
Authors:
Daniel G. Alfaro Vigo,
Daniele C. R. Gomes,
Bruno A. do Carmo,
Mauro A. Rincon
Abstract:
In this article, we present the mathematical analysis of the convergence of the linearized Crank-Nicolson Galerkin method for a nonlinear Schrodinger problem related to a domain with a moving boundary. The convergence analysis of the numerical method is carried out for both semi-discrete and fully discrete problems. An optimal error estimate in the $L^2$-norm with order…
▽ More
In this article, we present the mathematical analysis of the convergence of the linearized Crank-Nicolson Galerkin method for a nonlinear Schrodinger problem related to a domain with a moving boundary. The convergence analysis of the numerical method is carried out for both semi-discrete and fully discrete problems. An optimal error estimate in the $L^2$-norm with order ${O}(τ^2+ h^s),~ 2\leq s\leq r$, where $h$ is the finite element mesh size parameter, $τ$ is the time step, and $r-1$ represents the degree of the finite element polynomial basis. Numerical simulations are provided to confirm the consistency between theoretical and numerical results, validating the method and the order of convergence for different degrees $p\geq 1$ of the Lagrange polynomials and also for Hermite polynomials (degree $p=3$), which form the basis of the approximate solution.
△ Less
Submitted 30 April, 2025; v1 submitted 11 October, 2024;
originally announced October 2024.
-
Kitai's Criterion for Composition Operators
Authors:
Daniel Gomes,
Karl-G. Grosse-Erdmann
Abstract:
We present a general and natural framework to study the dynamics of composition operators on spaces of measurable functions, in which we then reconsider the characterizations for hypercyclic and mixing composition operators obtained by Bayart, Darji and Pires. We show that the notions of hypercyclicity and weak mixing coincide in this context and, if the system is dissipative, the recurrent compos…
▽ More
We present a general and natural framework to study the dynamics of composition operators on spaces of measurable functions, in which we then reconsider the characterizations for hypercyclic and mixing composition operators obtained by Bayart, Darji and Pires. We show that the notions of hypercyclicity and weak mixing coincide in this context and, if the system is dissipative, the recurrent composition operators agree with the hypercyclic ones. We also give a characterization for invertible composition operators satisfying Kitai's Criterion, and we construct an example of a mixing composition operator not satisfying Kitai's Criterion. For invertible dissipative systems with bounded distortion we show that composition operators satisfying Kitai's Criterion coincide with the mixing operators.
△ Less
Submitted 14 September, 2024;
originally announced September 2024.
-
On Differential and Riemannian Calculus on Wasserstein Spaces
Authors:
André Magalhães de Sá Gomes,
Christian S. Rodrigues,
Luiz A. B. San Martin
Abstract:
In this paper we develop an intrinsic formalism to study the topology, smooth structure, and Riemannian geometry of the Wasserstein space of a closed Riemannian manifold. Our formalism allows for a new characterisation of the Weak topology via convergent sequences of the subjacent space. Applying it we also provide a new proof that Wasserstein spaces of closed manifolds are geodesically convex. Ou…
▽ More
In this paper we develop an intrinsic formalism to study the topology, smooth structure, and Riemannian geometry of the Wasserstein space of a closed Riemannian manifold. Our formalism allows for a new characterisation of the Weak topology via convergent sequences of the subjacent space. Applying it we also provide a new proof that Wasserstein spaces of closed manifolds are geodesically convex. Our framework is particularly handy to address the Wasserstein spaces of compact Lie groups, where we refine our formalism and present an explicit example.
△ Less
Submitted 15 April, 2025; v1 submitted 7 June, 2024;
originally announced June 2024.
-
AdaFisher: Adaptive Second Order Optimization via Fisher Information
Authors:
Damien Martins Gomes,
Yanlei Zhang,
Eugene Belilovsky,
Guy Wolf,
Mahdi S. Hosseini
Abstract:
First-order optimization methods are currently the mainstream in training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by employing the diagonal matrix preconditioning of the stochastic gradient during the training. Despite their widespread, second-order optimization algorithms exhibit superior convergence properties compared to their first-order coun…
▽ More
First-order optimization methods are currently the mainstream in training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by employing the diagonal matrix preconditioning of the stochastic gradient during the training. Despite their widespread, second-order optimization algorithms exhibit superior convergence properties compared to their first-order counterparts e.g. Adam and SGD. However, their practicality in training DNNs is still limited due to increased per-iteration computations compared to the first-order methods. We present \emph{AdaFisher}--an adaptive second-order optimizer that leverages a \emph{diagonal block-Kronecker} approximation of the Fisher information matrix for adaptive gradient preconditioning. AdaFisher aims to bridge the gap between enhanced \emph{convergence/generalization} capabilities and computational efficiency in second-order optimization framework for training DNNs. Despite the slow pace of second-order optimizers, we showcase that AdaFisher can be reliably adopted for image classification, language modeling and stands out for its stability and robustness in hyper-parameter tuning. We demonstrate that AdaFisher \textbf{outperforms the SOTA optimizers} in terms of both accuracy and convergence speed. Code is available from https://github.com/AtlasAnalyticsLab/AdaFisher.
△ Less
Submitted 10 March, 2025; v1 submitted 25 May, 2024;
originally announced May 2024.
-
A Fully-discrete Semi-Lagrangian scheme for a price formation MFG model
Authors:
Yuri Ashrafyan,
Diogo Gomes
Abstract:
Here, we examine a fully-discrete Semi-Lagrangian scheme for a mean-field game price formation model. We show the existence of the solution of the discretized problem and that it is monotone as a multivalued operator. Moreover, we show that the limit of the discretization converges to the weak solution of the continuous price formation mean-field game using monotonicity methods. Numerical simulati…
▽ More
Here, we examine a fully-discrete Semi-Lagrangian scheme for a mean-field game price formation model. We show the existence of the solution of the discretized problem and that it is monotone as a multivalued operator. Moreover, we show that the limit of the discretization converges to the weak solution of the continuous price formation mean-field game using monotonicity methods. Numerical simulations demonstrate that this scheme can provide results efficiently, comparing favorably with other methods in the examples we tested.
△ Less
Submitted 30 January, 2025; v1 submitted 5 March, 2024;
originally announced March 2024.
-
$C^{1,α}$ Regularity For Stationary Mean-Field Games With Logarithmic Coupling
Authors:
Tigran Bakaryan,
Giuseppe Di Fazio,
Diogo A. Gomes
Abstract:
This paper investigates stationary mean-field games (MFGs) on the torus with Lipschitz non-homogeneous diffusion and logarithmic-like couplings. The primary objective is to understand the existence of $C^{1,α}$ solutions to address the research gap between low-regularity results for bounded and measurable diffusions and the smooth results modeled by the Laplacian.
We use the Hopf--Cole transform…
▽ More
This paper investigates stationary mean-field games (MFGs) on the torus with Lipschitz non-homogeneous diffusion and logarithmic-like couplings. The primary objective is to understand the existence of $C^{1,α}$ solutions to address the research gap between low-regularity results for bounded and measurable diffusions and the smooth results modeled by the Laplacian.
We use the Hopf--Cole transformation to convert the MFG system into a scalar elliptic equation. Then, we apply Morrey space methods to establish the existence and regularity of solutions. The introduction of Morrey space methods offers a novel approach to address regularity issues in the context of MFGs.
△ Less
Submitted 25 October, 2023;
originally announced October 2023.
-
Rigidity of Curvature Bounds of Quotient Spaces Of Isometric Actions
Authors:
André Magalhães de Sá Gomes,
Christian S. Rodrigues
Abstract:
Let $G\curvearrowright M$ be an isometric action of a Lie Group on a complete orientable Riemannian manifold. We disintegrate absolutely continuous measures with respect to the volume measure of $M$ along the principal orbits of $G\curvearrowright M$ and define a functional on the probability measures with support on the principal orbits of the action to further prove that the convexity properties…
▽ More
Let $G\curvearrowright M$ be an isometric action of a Lie Group on a complete orientable Riemannian manifold. We disintegrate absolutely continuous measures with respect to the volume measure of $M$ along the principal orbits of $G\curvearrowright M$ and define a functional on the probability measures with support on the principal orbits of the action to further prove that the convexity properties of this functional guarantees necessary and sufficient conditions to the Ricci curvature of $M$ to be bound below by a given real number $K$.
△ Less
Submitted 23 October, 2023;
originally announced October 2023.
-
Time Dependent First-Order Mean Field Games with Neumann Boundary Conditions
Authors:
Diogo A. Gomes,
Michele Ricciardi
Abstract:
The primary objective of this paper is to understand first-order, time-dependent mean-field games with Neumann boundary conditions, a question that remains under-explored in the literature. This matter is particularly relevant given the importance of boundary conditions in crowd models. In our model, the Neumann conditions result from players entering the domain according to a prescribed current,…
▽ More
The primary objective of this paper is to understand first-order, time-dependent mean-field games with Neumann boundary conditions, a question that remains under-explored in the literature. This matter is particularly relevant given the importance of boundary conditions in crowd models. In our model, the Neumann conditions result from players entering the domain according to a prescribed current, for instance, in a crowd entry scenario into an open-air concert or stadium. We formulate the model as a standard mean-field game coupling a Hamilton-Jacobi equation with a Fokker-Planck equation. Then, we introduce a relaxed variational problem and use Fenchel-Rockafellar duality to study the relation between these problems. Finally, we prove the existence and uniqueness of solutions for the system using variational methods.
△ Less
Submitted 23 October, 2024; v1 submitted 17 October, 2023;
originally announced October 2023.
-
Derivative-Free Global Minimization in One Dimension: Relaxation, Monte Carlo, and Sampling
Authors:
Alexandra A. Gomes,
Diogo A. Gomes
Abstract:
We introduce a derivative-free global optimization algorithm that efficiently computes minima for various classes of one-dimensional functions, including non-convex, and non-smooth functions.This algorithm numerically approximates the gradient flow of a relaxed functional, integrating strategies such as Monte Carlos methods, rejection sampling, and adaptive techniques. These strategies enhance per…
▽ More
We introduce a derivative-free global optimization algorithm that efficiently computes minima for various classes of one-dimensional functions, including non-convex, and non-smooth functions.This algorithm numerically approximates the gradient flow of a relaxed functional, integrating strategies such as Monte Carlos methods, rejection sampling, and adaptive techniques. These strategies enhance performance in solving a diverse range of optimization problems while significantly reducing the number of required function evaluations compared to established methods. We present a proof of the convergence of the algorithm and illustrate its performance by comprehensive benchmarking. The proposed algorithm offers a substantial potential for real-world models. It is particularly advantageous in situations requiring computationally intensive objective function evaluations.
△ Less
Submitted 17 August, 2023;
originally announced August 2023.
-
Machine Learning architectures for price formation models with common noise
Authors:
Diogo Gomes,
Julian Gutierrez,
Mathieu Laurière
Abstract:
We propose a machine learning method to solve a mean-field game price formation model with common noise. This involves determining the price of a commodity traded among rational agents subject to a market clearing condition imposed by random supply, which presents additional challenges compared to the deterministic counterpart. Our approach uses a dual recurrent neural network architecture encodin…
▽ More
We propose a machine learning method to solve a mean-field game price formation model with common noise. This involves determining the price of a commodity traded among rational agents subject to a market clearing condition imposed by random supply, which presents additional challenges compared to the deterministic counterpart. Our approach uses a dual recurrent neural network architecture encoding noise dependence and a particle approximation of the mean-field model with a single loss function optimized by adversarial training. We provide a posteriori estimates for convergence and illustrate our method through numerical experiments.
△ Less
Submitted 27 May, 2023;
originally announced May 2023.
-
A First-Order Mean-Field Game on a Bounded Domain with Mixed Boundary Conditions
Authors:
AbdulRahman M. Alharbi,
Yuri Ashrafyan,
Diogo Gomes
Abstract:
Entry-exit dynamics are crucial in modeling crowd movement. Here, we present a novel first-order, stationary mean-field game model on a bounded domain that accurately captures these dynamics. The interior dynamics of the system are governed by a standard first-order stationary MFG system consisting of a Hamilton-Jacobi equation coupled with a transport equation. The model incorporates nonstandard…
▽ More
Entry-exit dynamics are crucial in modeling crowd movement. Here, we present a novel first-order, stationary mean-field game model on a bounded domain that accurately captures these dynamics. The interior dynamics of the system are governed by a standard first-order stationary MFG system consisting of a Hamilton-Jacobi equation coupled with a transport equation. The model incorporates nonstandard mixed boundary conditions corresponding to an entry region $Γ_N$, where a Neumann condition prescribes agent inflow, and an exit region $Γ_D$, where a no-entry condition prevents inward flow. Additionally, we impose an upper bound on the exit cost through $Γ_D$, combined with a complementary contact-set condition.
The contact-set condition distinguishes boundary contact points, where the exit cost is attained and exit is permitted, from non-contact points, where a strict no-penetration condition is enforced. This mixed approach overcomes the limitations of classical Dirichlet conditions, which can artificially force boundary points to serve as both entry and exit locations. We analyze the system through a variational formulation, applying the direct method of the calculus of variations to establish the existence of solutions under minimal regularity assumptions. Furthermore, we prove a partial uniqueness result for the gradient of the value function (particularly in regions with positive agent density) and establish the uniqueness of the density function.
Several examples, including one- and two-dimensional cases, illustrate the proper assignment of entry and exit roles and demonstrate that contact does not necessarily enforce exit. Additionally, they showcase first-order MFG phenomena, such as the formation of empty regions, where agent density vanishes. These results provide a rigorous mathematical foundation for modeling realistic entry-exit scenarios.
△ Less
Submitted 17 March, 2025; v1 submitted 25 May, 2023;
originally announced May 2023.
-
Large Deviations for Lévy Diffusions in small regime
Authors:
Pedro Catuogno,
André de Oliveira Gomes
Abstract:
This article concerns the large deviations regime and the consequent solution of the Kramers problem for a two-time scale stochastic system driven by a common jump noise signal perturbed in small intensity $\varepsilon>0$ and with accelerated jumps by intensity $\frac{1}{\varepsilon}$. We establish Freidlin-Wentzell estimates for the slow process of the multiscale system in the small noise limit…
▽ More
This article concerns the large deviations regime and the consequent solution of the Kramers problem for a two-time scale stochastic system driven by a common jump noise signal perturbed in small intensity $\varepsilon>0$ and with accelerated jumps by intensity $\frac{1}{\varepsilon}$. We establish Freidlin-Wentzell estimates for the slow process of the multiscale system in the small noise limit $\varepsilon \rightarrow 0$ using the weak convergence approach to large deviations theory. The core of our proof is the reduction of the large deviations principle to the establishment of a stochastic averaging principle for auxiliary controlled processes. As consequence we solve the first exit time/ exit locus problem from a bounded domain containing the stable state of the averaged dynamics for the family of the slow processes in the small noise limit.
△ Less
Submitted 14 July, 2022;
originally announced July 2022.
-
First-order mean-field games on networks and Wardrop equilibrium
Authors:
Fatimah Al Saleh,
Tigran Bakaryan,
Diogo A. Gomes,
Ricardo Ribeiro
Abstract:
Here, we examine the Wardrop equilibrium model on networks with flow-dependent costs and its connection with stationary mean-field games (MFG). In the first part of this paper, we present the Wardrop and the first-order MFG models on networks. Then, we show how to reformulate the MFG problem into a Wardrop problem and prove that the MFG solution is the Wardrop equilibrium for the corresponding War…
▽ More
Here, we examine the Wardrop equilibrium model on networks with flow-dependent costs and its connection with stationary mean-field games (MFG). In the first part of this paper, we present the Wardrop and the first-order MFG models on networks. Then, we show how to reformulate the MFG problem into a Wardrop problem and prove that the MFG solution is the Wardrop equilibrium for the corresponding Wardrop problem. Moreover, we prove that the solution of the MFG problem can be recovered using the solution to the associated Wardrop problem. Finally, we study the cost properties and the calibration of MFG with Wardrop travel cost problems. We describe a novel approach to the calibration of MFGs. Further, we show that even simple travel costs can give rise to non-monotone MFGs.
△ Less
Submitted 4 July, 2022;
originally announced July 2022.
-
Machine Learning architectures for price formation models
Authors:
Diogo Gomes,
Julián Gutiérrez,
Mathieu Laurière
Abstract:
Here, we study machine learning (ML) architectures to solve a mean-field games (MFGs) system arising in price formation models. We formulate a training process that relies on a min-max characterization of the optimal control and price variables. Our main theoretical contribution is the development of a posteriori estimates as a tool to evaluate the convergence of the training process. We illustrat…
▽ More
Here, we study machine learning (ML) architectures to solve a mean-field games (MFGs) system arising in price formation models. We formulate a training process that relies on a min-max characterization of the optimal control and price variables. Our main theoretical contribution is the development of a posteriori estimates as a tool to evaluate the convergence of the training process. We illustrate our results with numerical experiments for linear dynamics and both quadratic and non-quadratic models.
△ Less
Submitted 25 January, 2023; v1 submitted 8 April, 2022;
originally announced April 2022.
-
The Potential Method For Price-Formation Models
Authors:
Yuri Ashrafyan,
Tigran Bakaryan,
Diogo Gomes,
Julian Gutierrez
Abstract:
We consider the mean-field game price formation model introduced by Gomes and Saúde.
In this MFG model, agents trade a commodity whose supply can be deterministic or stochastic. Agents maximize profit, taking into account current and future prices. The balance between supply and demand determines the price. We introduce a potential function that converts the MFG into a convex variational problem.…
▽ More
We consider the mean-field game price formation model introduced by Gomes and Saúde.
In this MFG model, agents trade a commodity whose supply can be deterministic or stochastic. Agents maximize profit, taking into account current and future prices. The balance between supply and demand determines the price. We introduce a potential function that converts the MFG into a convex variational problem. This variational formulation is particularly suitable for machine learning approaches. Here, we use a recurrent neural network to solve this problem. In the last section of the paper, we compare our results with known analytical solutions.
△ Less
Submitted 4 April, 2022;
originally announced April 2022.
-
A Variational Approach For Price Formation Models In One Dimension
Authors:
Yuri Ashrafyan,
Tigran Bakaryan,
Diogo Gomes,
Julian Gutierrez
Abstract:
In this paper, we study a class of first-order mean-field games (MFGs) that model price formation. Using Poincar{é} Lemma, we eliminate one of the equations and obtain a variational problem for a single function. This variational problem offers an alternative approach for the numerical solution of the original MFGs system. We show a correspondence between solutions of the MFGs system and the varia…
▽ More
In this paper, we study a class of first-order mean-field games (MFGs) that model price formation. Using Poincar{é} Lemma, we eliminate one of the equations and obtain a variational problem for a single function. This variational problem offers an alternative approach for the numerical solution of the original MFGs system. We show a correspondence between solutions of the MFGs system and the variational problem. Moreover, we address the existence of solutions for the variational problem using the direct method in the calculus of variations. We end the paper with numerical results for a linear-quadratic model.
△ Less
Submitted 4 April, 2022;
originally announced April 2022.
-
Discrete Approximation Of Stationary Mean Field Games
Authors:
Tigran Bakaryan,
Diogo Gomes,
Héctor Sánchez Morgado
Abstract:
In this paper, we focus on stationary (ergodic) mean-field games (MFGs). These games arise in the study of the long-time behavior of finite-horizon MFGs. Motivated by a prior scheme for Hamilton-Jacobi equations introduced in Aubry-Mather's theory, we introduce a discrete approximation to stationary MFGs. Relying on Kakutani's fixed-point theorem, we prove the existence and uniqueness (up to addit…
▽ More
In this paper, we focus on stationary (ergodic) mean-field games (MFGs). These games arise in the study of the long-time behavior of finite-horizon MFGs. Motivated by a prior scheme for Hamilton-Jacobi equations introduced in Aubry-Mather's theory, we introduce a discrete approximation to stationary MFGs. Relying on Kakutani's fixed-point theorem, we prove the existence and uniqueness (up to additive constant) of solutions to the discrete problem. Moreover, we show that the solutions to the discrete problem converge, uniformly in the nonlocal case and weakly in the local case, to the classical solutions of the stationary problem.
△ Less
Submitted 26 September, 2021;
originally announced September 2021.
-
Particle approximation of one-dimensional Mean-Field-Games with local interactions
Authors:
Marco Di Francesco,
Serikbolsyn Duisembay,
Diogo Aguiar Gomes,
Ricardo Ribeiro
Abstract:
We study a particle approximation for one-dimensional first-order Mean-Field-Games (MFGs) with local interactions with planning conditions. Our problem comprises a system of a Hamilton-Jacobi equation coupled with a transport equation. As we deal with the planning problem, we prescribe initial and terminal distributions for the transport equation. The particle approximation builds on a semi-discre…
▽ More
We study a particle approximation for one-dimensional first-order Mean-Field-Games (MFGs) with local interactions with planning conditions. Our problem comprises a system of a Hamilton-Jacobi equation coupled with a transport equation. As we deal with the planning problem, we prescribe initial and terminal distributions for the transport equation. The particle approximation builds on a semi-discrete variational problem. First, we address the existence and uniqueness of a solution to the semi-discrete variational problem. Next, we show that our discretization preserves some previously identified conserved quantities. Finally, we prove that the approximation by particle systems preserves displacement convexity. We use this last property to establish uniform estimates for the discrete problem. We illustrate our results for the discrete problem with numerical examples.
△ Less
Submitted 6 September, 2021;
originally announced September 2021.
-
A duality approach to a price formation MFG model
Authors:
Yuri Ashrafyan,
Tigran Bakaryan,
Diogo Gomes,
Julian Gutierrez
Abstract:
We study the connection between the Aubry-Mather theory and a mean-field game (MFG) price-formation model. We introduce a framework for Mather measures that is suited for constrained time-dependent problems in R. Then, we propose a variational problem on a space of measures, from which we obtain a duality relation involving the MFG problem examined in [36].
We study the connection between the Aubry-Mather theory and a mean-field game (MFG) price-formation model. We introduce a framework for Mather measures that is suited for constrained time-dependent problems in R. Then, we propose a variational problem on a space of measures, from which we obtain a duality relation involving the MFG problem examined in [36].
△ Less
Submitted 11 September, 2021; v1 submitted 4 September, 2021;
originally announced September 2021.
-
A random-supply Mean Field Game price model
Authors:
Diogo Gomes,
Julian Gutierrez,
Ricardo Ribeiro
Abstract:
We consider a market where a finite number of players trade an asset whose supply is a stochastic process. The price formation problem consists of finding a price process that ensures that when agents act optimally to minimize their trading costs, the market clears, and supply meets demand. This problem arises in market economies, including electricity generation from renewable sources in smart gr…
▽ More
We consider a market where a finite number of players trade an asset whose supply is a stochastic process. The price formation problem consists of finding a price process that ensures that when agents act optimally to minimize their trading costs, the market clears, and supply meets demand. This problem arises in market economies, including electricity generation from renewable sources in smart grids. Our model includes noise on the supply side, which is counterbalanced on the consumption side by storing energy or reducing the demand according to a dynamic price process. By solving a constrained minimization problem, we prove that the Lagrange multiplier corresponding to the market-clearing condition defines the solution of the price formation problem. For the linear-quadratic structure, we characterize the price process of a continuum population using optimal control techniques. We include numerical schemes for the price computation in the finite and infinite games, and we illustrate the model using real data.
△ Less
Submitted 12 August, 2022; v1 submitted 3 September, 2021;
originally announced September 2021.
-
Representations of low copolarity
Authors:
André Magalhães de Sá Gomes,
Claudio Gorodski
Abstract:
We classify irreducible representations of compact connected Lie groups whose orbit space is isometric to the orbit space of a representation of a compact Lie group of dimension~$7$, $8$ or $9$. They turn out to be closely related to symmetric spaces, with one exception only.
We classify irreducible representations of compact connected Lie groups whose orbit space is isometric to the orbit space of a representation of a compact Lie group of dimension~$7$, $8$ or $9$. They turn out to be closely related to symmetric spaces, with one exception only.
△ Less
Submitted 1 June, 2021;
originally announced June 2021.
-
A Potential Approach for Planning Mean-Field Games in One Dimension
Authors:
Tigran Bakaryan,
Rita Ferreira,
Diogo Gomes
Abstract:
This manuscript discusses planning problems for first- and second-order one-dimensional mean-field games (MFGs). These games are comprised of a Hamilton-Jacobi equation coupled with a Fokker-Planck equation. Applying Poincaré's Lemma to the Fokker-Planck equation, we deduce the existence of a potential. Rewriting the Hamilton-Jacobi equation in terms of the potential, we obtain a system of Euler-L…
▽ More
This manuscript discusses planning problems for first- and second-order one-dimensional mean-field games (MFGs). These games are comprised of a Hamilton-Jacobi equation coupled with a Fokker-Planck equation. Applying Poincaré's Lemma to the Fokker-Planck equation, we deduce the existence of a potential. Rewriting the Hamilton-Jacobi equation in terms of the potential, we obtain a system of Euler-Lagrange equations for certain variational problems. Instead of the mean-field planning problem (MFP), we study this variational problem. By the direct method in the calculus of variations, we prove the existence and uniqueness of solutions to the variational problem. The variational approach has the advantage of eliminating the continuity equation.
We also consider a first-order MFP with congestion. We prove that the congestion problem has a weak solution by introducing a potential and relying on the theory of variational inequalities. We end the paper by presenting an application to the one-dimensional Hughes' model.
△ Less
Submitted 25 April, 2021;
originally announced April 2021.
-
The large time profile for Hamilton--Jacobi--Bellman equations
Authors:
Diogo A. Gomes,
Hiroyoshi Mitake,
Hung V. Tran
Abstract:
Here, we study the large-time limit of viscosity solutions of the Cauchy problem for second-order Hamilton--Jacobi--Bellman equations with convex Hamiltonians in the torus. This large-time limit solves the corresponding stationary problem, sometimes called the ergodic problem. This problem, however, has multiple viscosity solutions and, thus, a key question is which of these solutions is selected…
▽ More
Here, we study the large-time limit of viscosity solutions of the Cauchy problem for second-order Hamilton--Jacobi--Bellman equations with convex Hamiltonians in the torus. This large-time limit solves the corresponding stationary problem, sometimes called the ergodic problem. This problem, however, has multiple viscosity solutions and, thus, a key question is which of these solutions is selected by the limit. Here, we provide a representation for the viscosity solution to the Cauchy problem in terms of generalized holonomic measures. Then, we use this representation to characterize the large-time limit in terms of the initial data and generalized Mather measures. In addition, we establish various results on generalized Mather measures and duality theorems that are of independent interest.
△ Less
Submitted 8 June, 2020;
originally announced June 2020.
-
Uniform estimates for the planning problem with potential
Authors:
Tigran Bakaryan,
Rita Ferreira,
Diogo Gomes
Abstract:
In this paper, we study a priori estimates for a first-order mean-field planning problem with a potential. In the theory of mean-field games (MFGs), a priori estimates play a crucial role to prove the existence of classical solutions. In particular, uniform bounds for the density of players' distribution and its inverse are of utmost importance. Here, we investigate a priori bounds for those quant…
▽ More
In this paper, we study a priori estimates for a first-order mean-field planning problem with a potential. In the theory of mean-field games (MFGs), a priori estimates play a crucial role to prove the existence of classical solutions. In particular, uniform bounds for the density of players' distribution and its inverse are of utmost importance. Here, we investigate a priori bounds for those quantities for a planning problem with a non-vanishing potential. The presence of a potential raises non-trivial difficulties, which we overcome by exploring a displacement-convexity property for the mean-field planning problem with a potential together with Moser's iteration method. We show that if the potential satisfies a certain smallness condition, then a displacement-convexity property holds. This property enables $L^q$ bounds for the density. In the one-dimensional case, the displacement-convexity property also gives $L^q$ bounds for the inverse of the density. Finally, using these $L^q$ estimates and Moser's iteration method, we obtain $L^\infty$ estimates for the density of the distribution of the players and its inverse.
△ Less
Submitted 5 March, 2020;
originally announced March 2020.
-
A mean-field game price model with noise
Authors:
Diogo Gomes,
Julian Gutierrez,
Ricardo Ribeiro
Abstract:
In this paper, we propose a mean-field game model for the price formation of a commodity whose production is subjected to random fluctuations. The model generalizes existing deterministic price formation models. Agents seek to minimize their average cost by choosing their trading rates with a price that is characterized by a balance between supply and demand. The supply and the price processes are…
▽ More
In this paper, we propose a mean-field game model for the price formation of a commodity whose production is subjected to random fluctuations. The model generalizes existing deterministic price formation models. Agents seek to minimize their average cost by choosing their trading rates with a price that is characterized by a balance between supply and demand. The supply and the price processes are assumed to follow stochastic differential equations. Here, we show that, for linear dynamics and quadratic costs, the optimal trading rates are determined in feedback form. Hence, the price arises as the solution to a stochastic differential equation, whose coefficients depend on the solution of a system of ordinary differential equations.
△ Less
Submitted 4 March, 2020;
originally announced March 2020.
-
Existence of weak solutions to time-dependent mean-field games
Authors:
Rita Ferreira,
Diogo Gomes,
Teruo Tada
Abstract:
Here, we establish the existence of weak solutions to a wide class of time-dependent monotone mean-field games (MFGs). These MFGs are given as a system of degenerate parabolic equations with initial and terminal conditions. To construct these solutions, we consider a high-order elliptic regularization in space-time. Then, using Schaefer's fixed-point theorem, we obtain the existence and uniqueness…
▽ More
Here, we establish the existence of weak solutions to a wide class of time-dependent monotone mean-field games (MFGs). These MFGs are given as a system of degenerate parabolic equations with initial and terminal conditions. To construct these solutions, we consider a high-order elliptic regularization in space-time. Then, using Schaefer's fixed-point theorem, we obtain the existence and uniqueness for this regularized problem. Using Minty's method, we prove the existence of a weak solution to the original MFG. Finally, the paper ends with a discussion on congestion problems and density constrained MFGs.
△ Less
Submitted 12 January, 2020;
originally announced January 2020.
-
Moderate averaged deviations for a multi-scale system with jumps and memory
Authors:
André de Oliveira Gomes,
Pedro Catuogno
Abstract:
This work studies a two-time-scale functional system given by two jump-diffusions under the scale separation by a small parameter $\varepsilon \rightarrow 0$. The coefficients of the equations that govern the dynamics of the system depend on the segment process of the slow variable (responsible for capturing delay effects on the slow component) and on the state of the fast variable. We derive a mo…
▽ More
This work studies a two-time-scale functional system given by two jump-diffusions under the scale separation by a small parameter $\varepsilon \rightarrow 0$. The coefficients of the equations that govern the dynamics of the system depend on the segment process of the slow variable (responsible for capturing delay effects on the slow component) and on the state of the fast variable. We derive a moderate deviations principle for the slow component of the system in the small noise limit using the weak convergence approach. The rate function is written in terms of the averaged dynamics associated to the multi-scale system. The core of the proof of the moderate deviations principle is the establishment of an averaging principle for the controlled processes associated to the slow variable in the framework of the weak convergence approach. The controlled version of the averaging principle for jump multi-scale diffusions relies on the classical Khasminkii's technique.
△ Less
Submitted 14 July, 2022; v1 submitted 24 September, 2019;
originally announced September 2019.
-
The selection problem for some first-order stationary mean-field games
Authors:
Diogo A. Gomes,
Hiroyoshi Mitake,
Kengo Terai
Abstract:
Here, we study the existence and the convergence of solutions for the vanishing discount MFG problem with a quadratic Hamiltonian. We give conditions under which the discounted problem has a unique classical solution and prove convergence of the vanishing-discount limit to a unique solution up to constants. Then, we establish refined asymptotics for the limit. When those conditions do not hold, th…
▽ More
Here, we study the existence and the convergence of solutions for the vanishing discount MFG problem with a quadratic Hamiltonian. We give conditions under which the discounted problem has a unique classical solution and prove convergence of the vanishing-discount limit to a unique solution up to constants. Then, we establish refined asymptotics for the limit. When those conditions do not hold, the limit problem may not have a unique solution and its solutions may not be smooth, as we illustrate in an elementary example. Finally, we investigate the stability of regular weak solutions and address the selection problem. Using ideas from Aubry-Mather theory, we establish a selection criterion for the limit.
△ Less
Submitted 18 August, 2019;
originally announced August 2019.
-
Well-posedness and controllability of Kawahara equation in weighted Sobolev spaces
Authors:
Roberto de A. Capistrano Filho,
Milena de S. Gomes
Abstract:
We consider the Kawahara equation, a fifth order Korteweg-de Vries type equation, posed on a bounded interval. The first result of the article is related to the well-posedness in weighted Sobolev spaces, which one was shown using a general version of the Lax--Milgram Theorem. With respect to the control problems, we will prove two results. First, if the control region is a neighborhood of the righ…
▽ More
We consider the Kawahara equation, a fifth order Korteweg-de Vries type equation, posed on a bounded interval. The first result of the article is related to the well-posedness in weighted Sobolev spaces, which one was shown using a general version of the Lax--Milgram Theorem. With respect to the control problems, we will prove two results. First, if the control region is a neighborhood of the right endpoint, an exact controllability result in weighted Sobolev spaces is established. Lastly, we show that the Kawahara equation is controllable by regions on $L^2$ Sobolev space, the so-called regional controllability, that is, the state function is exact controlled on the left part of the complement of the control region and null controlled on the right part of the complement of the control region.
△ Less
Submitted 18 January, 2021; v1 submitted 17 May, 2019;
originally announced May 2019.
-
Two-scale homogenization of a stationary mean-field game
Authors:
Rita Ferreira,
Diogo Gomes,
Xianjin Yang
Abstract:
In this paper, we characterize the asymptotic behavior of a first-order stationary mean-field game (MFG) with a logarithm coupling, a quadratic Hamiltonian, and a periodically oscillating potential. This study falls into the realm of the homogenization theory, and our main tool is the two-scale convergence. Using this convergence, we rigorously derive the two-scale homogenized and the homogenized…
▽ More
In this paper, we characterize the asymptotic behavior of a first-order stationary mean-field game (MFG) with a logarithm coupling, a quadratic Hamiltonian, and a periodically oscillating potential. This study falls into the realm of the homogenization theory, and our main tool is the two-scale convergence. Using this convergence, we rigorously derive the two-scale homogenized and the homogenized MFG problems, which encode the so-called macroscopic or effective behavior of the original oscillating MFG. Moreover, we prove existence and uniqueness of the solution to these limit problems.
△ Less
Submitted 6 May, 2019;
originally announced May 2019.
-
The Kramers problem for SDEs driven by small, accelerated Lévy noise with exponentially light jumps
Authors:
André de Oliveira Gomes,
Michael A. Högele
Abstract:
We establish Freidlin-Wentzell results for a nonlinear ordinary differential equation starting close to the stable state $0$, say, subject to a perturbation by a stochastic integral which is driven by an $\varepsilon$-small and $(1/\varepsilon)$-accelerated Lévy process with exponentially light jumps. For this purpose we derive a large deviations principle for the stochastically perturbed system u…
▽ More
We establish Freidlin-Wentzell results for a nonlinear ordinary differential equation starting close to the stable state $0$, say, subject to a perturbation by a stochastic integral which is driven by an $\varepsilon$-small and $(1/\varepsilon)$-accelerated Lévy process with exponentially light jumps. For this purpose we derive a large deviations principle for the stochastically perturbed system using the weak convergence approach developed by Budhiraja, Dupuis, Maroulas and collaborators in recent years. In the sequel we solve the associated asymptotic first escape problem from the bounded neighborhood of $0$ in the limit as $\varepsilon \rightarrow 0$ which is also known as the Kramers problem in the literature.
△ Less
Submitted 18 June, 2020; v1 submitted 3 April, 2019;
originally announced April 2019.
-
A new class of mild and strong solutions of integro-differential equation of arbitrary order in Banach space
Authors:
J. Vanterler da C. Sousa,
D. F. Gomes,
E. Capelas de Oliveira
Abstract:
The motivation that the field of differential equations provide to several researchers for the challenges that have been challenging them over the decades has contributed to the strengthening of the area within mathematics. In this sense, investigating important properties of solutions of differential equations, in particular fractional, has been object of study due to the exponential growth of th…
▽ More
The motivation that the field of differential equations provide to several researchers for the challenges that have been challenging them over the decades has contributed to the strengthening of the area within mathematics. In this sense, investigating important properties of solutions of differential equations, in particular fractional, has been object of study due to the exponential growth of the fractional calculus. In this paper, we investigate the existence and uniqueness of a new class of mild and strong solution of the fractional integro-differential equations in the Hilfer fractional derivative sense in Banach space, by means of the continuously $C_{0}$-semigroup, Banach fixed point theorem and Gronwall inequality.
△ Less
Submitted 28 December, 2018;
originally announced December 2018.
-
The Hessian Riemannian flow and Newton's method for Effective Hamiltonians and Mather measures
Authors:
Diogo A. Gomes,
Xianjin Yang
Abstract:
Effective Hamiltonians arise in several problems, including homogenization of Hamilton--Jacobi equations, nonlinear control systems, Hamiltonian dynamics, and Aubry--Mather theory. In Aubry--Mather theory, related objects, Mather measures, are also of great importance. Here, we combine ideas from mean-field games with the Hessian Riemannian flow to compute effective Hamiltonians and Mather measure…
▽ More
Effective Hamiltonians arise in several problems, including homogenization of Hamilton--Jacobi equations, nonlinear control systems, Hamiltonian dynamics, and Aubry--Mather theory. In Aubry--Mather theory, related objects, Mather measures, are also of great importance. Here, we combine ideas from mean-field games with the Hessian Riemannian flow to compute effective Hamiltonians and Mather measures simultaneously. We prove the convergence of the Hessian Riemannian flow in the continuous setting. For the discrete case, we give both the existence and the convergence of the Hessian Riemannian flow. In addition, we explore a variant of Newton's method that greatly improves the performance of the Hessian Riemannian flow. In our numerical experiments, we see that our algorithms preserve the non-negativity of Mather measures and are more stable than {related} methods in problems that are close to singular. Furthermore, our method also provides a way to approximate stationary MFGs.
△ Less
Submitted 17 April, 2020; v1 submitted 8 October, 2018;
originally announced October 2018.
-
Displacement convexity for first-order mean-field games
Authors:
Diogo Gomes,
Tommaso Seneci
Abstract:
Here, we consider the planning problem for first-order mean-field games (MFG). When there is no coupling between players, MFG degenerate into optimal transport problems. Displacement convexity is a fundamental tool in optimal transport that often reveals hidden convexity of functionals and, thus, has numerous applications in the calculus of variations. We explore the similarities between the Benam…
▽ More
Here, we consider the planning problem for first-order mean-field games (MFG). When there is no coupling between players, MFG degenerate into optimal transport problems. Displacement convexity is a fundamental tool in optimal transport that often reveals hidden convexity of functionals and, thus, has numerous applications in the calculus of variations. We explore the similarities between the Benamou-Brenier formulation of optimal transport and MFG to extend displacement convexity methods from to MFG. In particular, we identify a class of functions, that depend on solutions of MFG, that are convex in time and, thus, obtain new a priori bounds for solutions of MFG. A remarkable consequence is the log-convexity of $L^q$ norms. This convexity gives bounds for the density of solutions of the planning problem and extends displacement convexity of $L^q$ norms from optimal transport. Additionally, we prove the convexity of $L^q$ norms for MFG with congestion.
△ Less
Submitted 18 July, 2018;
originally announced July 2018.
-
A mean-field game approach to price formation in electricity markets
Authors:
Diogo Gomes,
João Saúde
Abstract:
Here, we introduce a price-formation model where a large number of small players can store and trade electricity. Our model is a constrained mean-field game (MFG) where the price is a Lagrange multiplier for the supply vs. demand balance condition. We establish the existence of a unique solution using a fixed-point argument. In particular, we show that the price is well-defined and it is a Lipschi…
▽ More
Here, we introduce a price-formation model where a large number of small players can store and trade electricity. Our model is a constrained mean-field game (MFG) where the price is a Lagrange multiplier for the supply vs. demand balance condition. We establish the existence of a unique solution using a fixed-point argument. In particular, we show that the price is well-defined and it is a Lipschitz function of time. Then, we study linear-quadratic models that can be solved explicitly and compare our model with real data.
△ Less
Submitted 18 July, 2018;
originally announced July 2018.
-
Existence of weak solutions to first-order stationary mean-field games with Dirichlet conditions
Authors:
Rita Ferreira,
Diogo Gomes,
Teruo Tada
Abstract:
In this paper, we study first-order stationary monotone mean-field games (MFGs) with Dirichlet boundary conditions. While for Hamilton--Jacobi equations Dirichlet conditions may not be satisfied, here, we establish the existence of solutions of MFGs that satisfy those conditions. To construct these solutions, we introduce a monotone regularized problem. Applying Schaefer's fixed-point theorem and…
▽ More
In this paper, we study first-order stationary monotone mean-field games (MFGs) with Dirichlet boundary conditions. While for Hamilton--Jacobi equations Dirichlet conditions may not be satisfied, here, we establish the existence of solutions of MFGs that satisfy those conditions. To construct these solutions, we introduce a monotone regularized problem. Applying Schaefer's fixed-point theorem and using the monotonicity of the MFG, we verify that there exists a unique weak solution to the regularized problem. Finally, we take the limit of the solutions of the regularized problem and using Minty's method, we show the existence of weak solutions to the original MFG.
△ Less
Submitted 19 April, 2018;
originally announced April 2018.
-
Orthogonal foliations on riemannian manifolds
Authors:
André de Oliveira Gomes,
Eurípedes Carvalho da Silva
Abstract:
In this work, we find an equation that relates the Ricci curvature of a riemannian manifold $M$ and the second fundamental forms of two orthogonal foliations of complementary dimensions, $\mathcal{F}$ and $\mathcal{F}^{\bot}$, defined on $M$. Using this equation, we show a sufficient condition for the manifold M to be locally a riemannian product of the leaves of $\mathcal{F}$ and…
▽ More
In this work, we find an equation that relates the Ricci curvature of a riemannian manifold $M$ and the second fundamental forms of two orthogonal foliations of complementary dimensions, $\mathcal{F}$ and $\mathcal{F}^{\bot}$, defined on $M$. Using this equation, we show a sufficient condition for the manifold M to be locally a riemannian product of the leaves of $\mathcal{F}$ and $\mathcal{F}^{\bot}$, if one of the foliations is totally umbilical. We also prove an integral formula for such foliations.
△ Less
Submitted 15 November, 2017;
originally announced November 2017.
-
First-order, stationary mean-field games with congestion
Authors:
David Evangelista,
Rita Ferreira,
Diogo A. Gomes,
Levon Nurbekyan,
Vardan Voskanyan
Abstract:
Mean-field games (MFGs) are models for large populations of competing rational agents that seek to optimize a suitable functional. In the case of congestion, this functional takes into account the difficulty of moving in high-density areas. Here, we study stationary MFGs with congestion with quadratic or power-like Hamiltonians. First, using explicit examples, we illustrate two main difficulties:…
▽ More
Mean-field games (MFGs) are models for large populations of competing rational agents that seek to optimize a suitable functional. In the case of congestion, this functional takes into account the difficulty of moving in high-density areas. Here, we study stationary MFGs with congestion with quadratic or power-like Hamiltonians. First, using explicit examples, we illustrate two main difficulties: the lack of classical solutions and the existence of areas with vanishing density. Our main contribution is a new variational formulation for MFGs with congestion. This formulation was not previously known, and, thanks to it, we prove the existence and uniqueness of solutions. Finally, we consider applications to numerical methods.
△ Less
Submitted 4 October, 2017;
originally announced October 2017.
-
Monotone numerical methods for finite-state mean-field games
Authors:
Diogo Gomes,
Joao Saude
Abstract:
Here, we develop numerical methods for finite-state mean-field games (MFGs) that satisfy a monotonicity condition. MFGs are determined by a system of differential equations with initial and terminal boundary conditions. These non-standard conditions are the main difficulty in the numerical approximation of solutions. Using the monotonicity condition, we build a flow that is a contraction and whose…
▽ More
Here, we develop numerical methods for finite-state mean-field games (MFGs) that satisfy a monotonicity condition. MFGs are determined by a system of differential equations with initial and terminal boundary conditions. These non-standard conditions are the main difficulty in the numerical approximation of solutions. Using the monotonicity condition, we build a flow that is a contraction and whose fixed points solve the MFG, both for stationary and time-dependent problems. We illustrate our methods in a MFG modeling the paradigm-shift problem.
△ Less
Submitted 29 April, 2017;
originally announced May 2017.
-
Conservation laws arising in the study of forward-forward Mean-Field Games
Authors:
Diogo Gomes,
Levon Nurbekyan,
Marc Sedjro
Abstract:
We consider forward-forward Mean Field Game (MFG) models that arise in numerical approximations of stationary MFGs. First, we establish a link between these models and a class of hyperbolic conservation laws as well as certain nonlinear wave equations. Second, we investigate existence and long-time behavior of solutions for such models.
We consider forward-forward Mean Field Game (MFG) models that arise in numerical approximations of stationary MFGs. First, we establish a link between these models and a class of hyperbolic conservation laws as well as certain nonlinear wave equations. Second, we investigate existence and long-time behavior of solutions for such models.
△ Less
Submitted 24 April, 2017;
originally announced April 2017.
-
One-dimensional, forward-forward mean-field games with congestion
Authors:
Diogo Gomes,
Marc Sedjro
Abstract:
Here, we consider one-dimensional forward-forward mean-field games (MFGs) with congestion, which were introduced to approximate stationary MFGs. We use methods from the theory of conservation laws to examine the qualitative properties of these games. First, by computing Riemann invariants and corresponding invariant regions, we develop a method to prove lower bounds for the density. Next, by combi…
▽ More
Here, we consider one-dimensional forward-forward mean-field games (MFGs) with congestion, which were introduced to approximate stationary MFGs. We use methods from the theory of conservation laws to examine the qualitative properties of these games. First, by computing Riemann invariants and corresponding invariant regions, we develop a method to prove lower bounds for the density. Next, by combining the lower bound with an entropy function, we prove the existence of global solutions for parabolic forward-forward MFGs. Finally, we construct traveling-wave solutions, which settles in a negative way the convergence problem for forward-forward MFGs. A similar technique gives the existence of time-periodic solutions for non-monotonic MFGs.
△ Less
Submitted 29 March, 2017;
originally announced March 2017.
-
An Adjoint-based Numerical Method for a class of nonlinear Fokker-Planck Equations
Authors:
Adriano Festa,
Diogo A. Gomes,
Roberto M. Velho
Abstract:
Here, we introduce a numerical approach for a class of Fokker-Planck (FP) equations. These equations are the adjoint of the linearization of Hamilton-Jacobi (HJ) equations. Using this structure, we show how to transfer the properties of schemes for HJ equations to the FP equations. Hence, we get numerical schemes with desirable features such as positivity and mass-preservation. We illustrate this…
▽ More
Here, we introduce a numerical approach for a class of Fokker-Planck (FP) equations. These equations are the adjoint of the linearization of Hamilton-Jacobi (HJ) equations. Using this structure, we show how to transfer the properties of schemes for HJ equations to the FP equations. Hence, we get numerical schemes with desirable features such as positivity and mass-preservation. We illustrate this approach in examples that include mean-field games and a crowd motion model.
△ Less
Submitted 22 March, 2017;
originally announced March 2017.
-
Radially Symmetric Mean-Field Games with Congestion
Authors:
David Evangelista,
Diogo A. Gomes,
Levon Nurbekyan
Abstract:
Here, we study radial solutions for first- and second-order stationary Mean-Field Games (MFG) with congestion on $\mathbb{R}^d$. MFGs with congestion model problems where the agents' motion is hampered in high-density regions. The radial case, which is one of the simplest non one-dimensional MFG, is relatively tractable. As we observe in this paper, the Fokker-Planck equation is integrable with re…
▽ More
Here, we study radial solutions for first- and second-order stationary Mean-Field Games (MFG) with congestion on $\mathbb{R}^d$. MFGs with congestion model problems where the agents' motion is hampered in high-density regions. The radial case, which is one of the simplest non one-dimensional MFG, is relatively tractable. As we observe in this paper, the Fokker-Planck equation is integrable with respect to one of the unknowns. Consequently, we obtain a single equation substituting this solution into the Hamilton-Jacobi equation. For the first-order case, we derive explicit formulas; for the elliptic case, we study a variational formulation of the resulting equation. In both cases, we use our approach to compute numerical approximations to the solutions of the corresponding MFG systems.
△ Less
Submitted 22 March, 2017;
originally announced March 2017.
-
On the existence of solutions for stationary mean-field games with congestion
Authors:
David Evangelista,
Diogo A. Gomes
Abstract:
Mean-field games (MFGs) are models of large populations of rational agents who seek to optimize an objective function that takes into account their location and the distribution of the remaining agents. Here, we consider stationary MFGs with congestion and prove the existence of stationary solutions. Because moving in congested areas is difficult, agents prefer to move in non-congested areas. As a…
▽ More
Mean-field games (MFGs) are models of large populations of rational agents who seek to optimize an objective function that takes into account their location and the distribution of the remaining agents. Here, we consider stationary MFGs with congestion and prove the existence of stationary solutions. Because moving in congested areas is difficult, agents prefer to move in non-congested areas. As a consequence, the model becomes singular near the zero density. The existence of stationary solutions was previously obtained for MFGs with quadratic Hamiltonians thanks to a very particular identity. Here, we develop robust estimates that give the existence of a solution for general subquadratic Hamiltonians.
△ Less
Submitted 24 November, 2016;
originally announced November 2016.
