Singular perturbations in stochastic optimal control with unbounded data

DOI10.1051/COCV/2023020zbMATH Open1526.49014arXiv2208.00655OpenAlexW4361008785MaRDI QIDQ6138481FDOQ6138481

Authors: Martino Bardi, Hicham Kouhkouh 

Publication date: 5 September 2023

Published in: ESAIM: Control, Optimisation and Calculus of Variations (Search for Journal in Brave)

Abstract: We study singular perturbations of a class of two-scale stochastic control systems with unbounded data. The assumptions are designed to cover some relaxation problems for deep neural networks. We construct effective Hamiltonian and initial data and prove the convergence of the value function to the solution of a limit (effective) Cauchy problem for a parabolic equation of HJB type. We use methods of probability, viscosity solutions and homogenization.

Full work available at URL: https://arxiv.org/abs/2208.00655

Recommendations

• Deep relaxation of controlled stochastic gradient descent via singular perturbations
• Singular perturbations and optimal control of stochastic systems in infinite dimension: HJB equations and viscosity solutions
• Singular Perturbations in Stochastic Ergodic Control Problems
• scientific article; zbMATH DE number 4132079
• Ergodicity, stabilization, and singular perturbations for Bellman-Isaacs equations

zbMATH Keywords

Hamilton-Jacobi-Bellman equationshomogenizationinvariant measuressingular perturbationsstochastic optimal controlviscosity solutionstwo-scale systems

Mathematics Subject Classification ID

Neural networks for/in biological studies, artificial life and related topics (92B20) Viscosity solutions to PDEs (35D40) Hamilton-Jacobi equations (35F21) Methods involving semicontinuity and convergence; relaxation (49J45) Existence of optimal solutions to problems involving randomness (49J55) Viscosity solutions to Hamilton-Jacobi equations in optimal control and differential games (49L25) Time-scale analysis and singular perturbations in control/observation systems (93C70) Optimal stochastic control (93E20)

Cites Work

• Weak convergence methods and singularly perturbed stochastic control and filtering problems
• Singular perturbations of nonlinear degenerate parabolic pDEs: A general convergence result
• Large deviations and importance sampling for systems of slow-fast motion
• Averaging of singularly perturbed controlled stochastic differential equations
• Multiscale stochastic volatility for equity, interest rate, and credit derivatives.
• Uniqueness Results for Second-Order Bellman--Isaacs Equations under Quadratic Growth Assumptions and Applications
• Convergence by viscosity methods in multiscale financial models with stochastic volatility
• Title not available (Why is that?)
• Optimal control with random parameters: a multiscale approach
• Small-time asymptotics for fast mean-reverting stochastic volatility models
• Viscosity Solutions Methods for Singular Perturbations in Deterministic and Stochastic Control
• The perturbed test function method for viscosity solutions of nonlinear PDE
• Stochastic differential equations and applications.
• On polynomial mixing bounds for stochastic differential equations
• On Poisson equation and diffusion approximation. II.
• Measurable relations
• Title not available (Why is that?)
• On the Poisson equation and diffusion approximation. I
• Liouville properties and critical value of fully nonlinear elliptic operators
• The ergodic problem for some subelliptic operators with unbounded coefficients
• Optimal control and viscosity solutions of Hamilton-Jacobi-Bellman equations
• Title not available (Why is that?)
• Transition times and stochastic resonance for multidimensional diffusions with time periodic drift: a large deviations approach
• On the Poisson equation and diffusion approximation. III
• Singular Perturbations in Ergodic Control of Diffusions
• Singular perturbations and optimal control of stochastic systems in infinite dimension: HJB equations and viscosity solutions
• Averaging principle for slow-fast stochastic differential equations with time dependent locally Lipschitz coefficients
• Deep relaxation: partial differential equations for optimizing deep neural networks
• Entropy-SGD: biasing gradient descent into wide valleys
• The Kantorovich and variation distances between invariant measures of diffusions and nonlinear stationary Fokker-Planck-Kolmogorov equations
• Averaging principle and normal deviations for multiscale stochastic systems
• Singular limit of BSDEs and optimal control of two scale stochastic systems in infinite dimensional spaces
• Viscosity methods for large deviations estimates of multiscale stochastic processes
• Simultaneous small noise limit for singularly perturbed slow-fast coupled diffusions
• The averaging principle for non-autonomous slow-fast stochastic differential equations and an application to a local stochastic volatility model
• Rate of convergence for singular perturbations of Hamilton-Jacobi equations in unbounded spaces

Cited In (4)

• A Viscous Ergodic Problem with Unbounded and Measurable Ingredients, Part 1: HJB Equation
• Deep relaxation of controlled stochastic gradient descent via singular perturbations
• A singular stochastic control problem in an unbounded domain
• Stationary equilibria and their stability in a Kuramoto MFG with strong interaction

This page was built for publication: Singular perturbations in stochastic optimal control with unbounded data

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6138481)