Showing 1–9 of 9 results for author: De Maio, R

A Mean Field Games model for finite mixtures of Bernoulli and Categorical distributions

Authors: Laura Aquilanti, Simone Cacace, Fabio Camilli, Raul De Maio

Abstract: Finite mixture models are an important tool in the statistical analysis of data, for example in data clustering. The optimal parameters of a mixture model are usually computed by maximizing the log-likelihood functional via the Expectation-Maximization algorithm. We propose an alternative approach based on the theory of Mean Field Games, a class of differential games with an infinite number of age… ▽ More Finite mixture models are an important tool in the statistical analysis of data, for example in data clustering. The optimal parameters of a mixture model are usually computed by maximizing the log-likelihood functional via the Expectation-Maximization algorithm. We propose an alternative approach based on the theory of Mean Field Games, a class of differential games with an infinite number of agents. We show that the solution of a finite state space multi-population Mean Field Games system characterizes the critical points of the log-likelihood functional for a Bernoulli mixture. The approach is then generalized to mixture models of categorical distributions. Hence, the Mean Field Games approach provides a method to compute the parameters of the mixture model, and we show its application to some standard examples in cluster analysis. △ Less

Submitted 27 November, 2020; v1 submitted 17 April, 2020; originally announced April 2020.

A Mean Field Games approach to Cluster Analysis

Authors: Laura Aquilanti, Simone Cacace, Fabio Camilli, Raul De Maio

Abstract: In this paper, we develop a Mean Field Games approach to Cluster Analysis. We consider a finite mixture model, given by a convex combination of probability density functions, to describe the given data set. We interpret a data point as an agent of one of the populations represented by the components of the mixture model, and we introduce a corresponding optimal control problem. In this way, we obt… ▽ More In this paper, we develop a Mean Field Games approach to Cluster Analysis. We consider a finite mixture model, given by a convex combination of probability density functions, to describe the given data set. We interpret a data point as an agent of one of the populations represented by the components of the mixture model, and we introduce a corresponding optimal control problem. In this way, we obtain a multi-population Mean Field Games system which characterizes the parameters of the finite mixture model. Our method can be interpreted as a continuous version of the classical Expectation-Maximization algorithm. △ Less

Submitted 23 December, 2019; v1 submitted 4 July, 2019; originally announced July 2019.

Superposition principle and schemes for Measure Differential Equations

Authors: Fabio Camilli, Giulia Cavagnari, Raul De Maio, Benedetto Piccoli

Abstract: Measure Differential Equations (MDE) describe the evolution of probability measures driven by probability velocity fields, i.e. probability measures on the tangent bundle. They are, on one side, a measure-theoretic generalization of ordinary differential equations; on the other side, they allow to describe concentration and diffusion phenomena typical of kinetic equations. In this paper, we analyz… ▽ More Measure Differential Equations (MDE) describe the evolution of probability measures driven by probability velocity fields, i.e. probability measures on the tangent bundle. They are, on one side, a measure-theoretic generalization of ordinary differential equations; on the other side, they allow to describe concentration and diffusion phenomena typical of kinetic equations. In this paper, we analyze some properties of this class of differential equations, especially highlighting their link with nonlocal continuity equations. We prove a representation result in the spirit of the Superposition Principle by Ambrosio-Gigli-Savaré, and we provide alternative schemes converging to a solution of the MDE, with a particular view to uniqueness/non-uniqueness phenomena. △ Less

Submitted 17 December, 2020; v1 submitted 14 February, 2019; originally announced February 2019.

Comments: Accepted for publication in Kinetic and Related Models, DOI: 10.3934/krm.2020050. Published version available at http://www.aimsciences.org/article/doi/10.3934/krm.2020050

Memory effects in measure transport equations

Authors: Fabio Camilli, Raul De Maio

Abstract: Transport equations with a nonlocal velocity field have been introduced as a continuum model for interacting particle systems arising in physics, chemistry and biology. Fractional time derivatives, given by convolution integrals of the time-derivative with power-law kernels, are typical for memory effects in complex systems. In this paper we consider a nonlinear transport equation with a fractiona… ▽ More Transport equations with a nonlocal velocity field have been introduced as a continuum model for interacting particle systems arising in physics, chemistry and biology. Fractional time derivatives, given by convolution integrals of the time-derivative with power-law kernels, are typical for memory effects in complex systems. In this paper we consider a nonlinear transport equation with a fractional time-derivative. We provide a well-posedness theory for weak measure solutions of the problem and an integral formula which generalizes the classical push-forward representation formula to this setting. △ Less

Submitted 15 April, 2019; v1 submitted 27 June, 2018; originally announced June 2018.

MSC Class: 35R11; 35Q35; 26A33

A Hopf-Lax formula for Hamilton-Jacobi equations with Caputo time derivative

Authors: Fabio Camilli, Raul De Maio, Elisa Iacomini

Abstract: We prove a representation formula of Hopf-Lax type for the solution of a Hamilton-Jacobi equation involving Caputo time-fractional derivative. Equations of these type are associated with optimal control problems where the controlled dynamics is replaced by a time-changed stochastic process describing the trajectory of a particle subject to random trapping effects. We prove a representation formula of Hopf-Lax type for the solution of a Hamilton-Jacobi equation involving Caputo time-fractional derivative. Equations of these type are associated with optimal control problems where the controlled dynamics is replaced by a time-changed stochastic process describing the trajectory of a particle subject to random trapping effects. △ Less

Submitted 25 March, 2018; originally announced March 2018.

MSC Class: 35R11; 26A33; 49L20

A measure theoretic approach to traffic flow optimization on networks

Authors: Simone Cacace, Fabio Camilli, Raul De Maio, Andrea Tosin

Abstract: We consider a class of optimal control problems for measure-valued nonlinear transport equations describing traffic flow problems on networks. The objective isto minimise/maximise macroscopic quantities, such as traffic volume or average speed,controlling few agents, for example smart traffic lights and automated cars. The measuretheoretic approach allows to study in a same setting local and nonlo… ▽ More We consider a class of optimal control problems for measure-valued nonlinear transport equations describing traffic flow problems on networks. The objective isto minimise/maximise macroscopic quantities, such as traffic volume or average speed,controlling few agents, for example smart traffic lights and automated cars. The measuretheoretic approach allows to study in a same setting local and nonlocal drivers interactionsand to consider the control variables as additional measures interacting with the driversdistribution. We also propose a gradient descent adjoint-based optimization method, ob-tained by deriving first-order optimality conditions for the control problem, and we providesome numerical experiments in the case of smart traffic lights for a 2-1 junction. △ Less

Submitted 2 March, 2018; originally announced March 2018.

Comments: 20 pages, 6 figures

MSC Class: 49J20; 35B37; 35R02; 49M07; 65M99

Journal ref: European J. Appl. Math., 30(6):1187-1209, 2019

Measure-valued solutions to nonlocal transport equations on networks

Authors: Fabio Camilli, Raul De Maio, Andrea Tosin

Abstract: We study a nonlinear transport equation defined on an oriented network where the velocity field depends not only on the state variable, but also on the solution itself. We prove existence, uniqueness and continuous dependence results for the solution of the problem intended in a suitable measure-theoretic sense. We also provide a representation formula in terms of the push-forward of the initial a… ▽ More We study a nonlinear transport equation defined on an oriented network where the velocity field depends not only on the state variable, but also on the solution itself. We prove existence, uniqueness and continuous dependence results for the solution of the problem intended in a suitable measure-theoretic sense. We also provide a representation formula in terms of the push-forward of the initial and boundary data along the network and discuss an example of nonlocal velocity field fitting our framework. △ Less

Submitted 22 January, 2018; v1 submitted 12 September, 2017; originally announced September 2017.

MSC Class: 35R02; 35Q35; 28A50

Journal ref: J. Differential Equations, 264(12):7213-7241, 2018

A time-fractional mean field game

Authors: Fabio Camilli, Raul De Maio

Abstract: We consider a Mean Field Games model where the dynamics of the agents is subdiffusive. According to the optimal control interpretation of the problem, we get a system involving fractional time-derivatives for the Hamilton-Jacobi-Bellman and the Fokker-Planck equations. We discuss separately the well-posedness for each of the two equations and then we prove existence and uniqueness of the solution… ▽ More We consider a Mean Field Games model where the dynamics of the agents is subdiffusive. According to the optimal control interpretation of the problem, we get a system involving fractional time-derivatives for the Hamilton-Jacobi-Bellman and the Fokker-Planck equations. We discuss separately the well-posedness for each of the two equations and then we prove existence and uniqueness of the solution to the Mean Field Games system △ Less

Submitted 22 January, 2018; v1 submitted 1 September, 2017; originally announced September 2017.

MSC Class: 35R11; 60H05; 26A33; 40L20; 49N70

Transport of measures on networks

Authors: Fabio Camilli, Raul De Maio, Andrea Tosin

Abstract: In this paper we formulate a theory of measure-valued linear transport equations on networks. The building block of our approach is the initial/boundary-value problem for the measure-valued linear transport equation on a bounded interval, which is the prototype of an arc of the network. For this problem we give an explicit representation formula of the solution, which also considers the total mass… ▽ More In this paper we formulate a theory of measure-valued linear transport equations on networks. The building block of our approach is the initial/boundary-value problem for the measure-valued linear transport equation on a bounded interval, which is the prototype of an arc of the network. For this problem we give an explicit representation formula of the solution, which also considers the total mass flowing out of the interval. Then we construct the global solution on the network by gluing all the measure-valued solutions on the arcs by means of appropriate distribution rules at the vertexes. The measure-valued approach makes our framework suitable to deal with multiscale flows on networks, with the microscopic and macroscopic phases represented by Lebesgue-singular and Lebesgue-absolutely continuous measures, respectively, in time and space. △ Less

Submitted 13 February, 2017; v1 submitted 3 October, 2016; originally announced October 2016.

MSC Class: 35R02; 35Q35; 28A50

Journal ref: Netw. Heretog. Media, 12(2):191-215, 2017