Abstract:Parabolic partial differential equations (PDEs) and backward stochastic differential equations (BSDEs) are key ingredients a number of models physics and financial engineering. In particular, parabolic PDEs and BSDEs are fundamental tools the state-of-the-art pricing and hedging of financial derivatives. The PDEs and BSDEs appearing such applications are often high-dimensional and nonlinear. Since explicit solutions of such PDEs and BSDEs are typically not available, it is a very active topic of research to solve such PDEs and BSDEs approximately. In this paper we introduce a family of new approximation methods for high-dimensional PDEs and BSDEs. A key idea of our methods is to combine multilevel approximations with Picard fixed-point approximations. Thereby we obtain a class of full history recursive multilevel Picard approximations. Our error analysis proves for one of the proposed approximation methods that if $epsilonin(0,infty)$ is the prescribed approximation accuracy and if $dinmathbb{N}$ is the dimension of the considered PDE, then for every $delta in (0,infty)$ it holds that the computational effort of the approximation method (number of function evaluations of the coefficient functions of the considered PDE and number of used independent scalar standard normal random variables) is at most $O(d,epsilon^{-(4+delta)})$, that is, the computational effort grows only linearly the dimension and up to an arbitrarily small order quartically the reciprocal of the prescribed approximation accuracy. We illustrate the efficiency of one of the proposed approximation methods by means of numerical simulations presenting approximation accuracy against runtime for several nonlinear PDEs from physics (such as the Allen-Cahn equation) and financial engineering (such as derivative pricing incorporating default risks) the case of $d=100$ space dimensions.

The Deep Ritz Method for Parametric $p$-Dirichlet Problems

Error analysis for a Crouzeix-Raviart approximation of the $p$-Dirichlet problem

Deep Ritz Method with Adaptive Quadrature for Linear Elasticity

Deep Ritz Methods for Laplace Equations with Dirichlet Boundary Condition

Deep Ritz -- Finite Element methods: Neural Network Methods trained with Finite Elements

Adaptive importance sampling for Deep Ritz

On full history recursive multilevel Picard approximations and numerical approximations for high-dimensional nonlinear parabolic partial differential equations and high-dimensional nonlinear backward stochastic differential equations

The Deep Parametric PDE Method: Application to Option Pricing

A Deep Double Ritz Method (D$^2$RM) for solving Partial Differential Equations using Neural Networks

A Natural Deep Ritz Method for Essential Boundary Value Problems

Uniform convergence guarantees for the deep Ritz method for nonlinear problems

Error Analysis of Deep Ritz Methods for Elliptic Equations.

Deep Ritz method for the spectral fractional Laplacian equation using the Caffarelli-Silvestre extension

Kernel Methods in the Deep Ritz framework: Theory and practice

The Deep Ritz Method: A Deep Learning-Based Numerical Algorithm for Solving Variational Problems

Error analysis for a Crouzeix-Raviart approximation of the variable exponent Dirichlet problem

Parametric PDEs: Sparse or Low-Rank Approximations?

Solving parametric partial differential equations with deep rectified quadratic unit neural networks

Parametric Integration by Magic Point Empirical Interpolation

A comparison study of deep Galerkin method and deep Ritz method for elliptic problems with different boundary conditions

Statistical Numerical PDE : Fast Rate, Neural Scaling Law and When it’s Optimal