Monte Carlo difference schemes for the wave equation
Consider iteration procedures of the form \[ x^n = A_0 x^n + A_1 x^{n-1} A_2 x^{n-2} + f, n \geq 2 \] where \( x^0, x^1 \in\mathbb{R}^m\) are given initial values, \(A_i \in \mathbb{R}^{m \times m}\) matrices, as commonly met in equidistant difference schemes for partial differential equations or in fixed point iterations for linear systems \(x = A x + f \) with \(A = A_0 + A_1 + A_2\). The authors study the qualitative behavior of random estimators \(\eta^n\) for the calculation of \(x^n\). There estimators using appropriate Monte Carlo algorithms (direct, recurrent) are constructed, based on the well-known von Neumann-Ulam scheme and aiming at the control on stochastic errors for large \(n\). For this purpose, they treat the problem of stability (boundedness) of corresponding Monte Carlo algorithms. According to the paper, a Monte Carlo algorithm (MC algorithm) is called stable if \[ \sup_{n \in \mathbb N} \|\text{Cov}(\eta^n) \|< + \infty \] for related covariance matrices \(\text{Cov} (\eta^n)\). An explicit representation theorem for these covariance matrices related to the MC algorithms is proven. The main result provides a necessary and sufficient condition for the stability of a recurrent Monte Carlo algorithm. Eventually, the analysis is illustrated by an application to the numerical approximations of the (deterministic) one-dimensional wave and heat flow equations. Some test results are given to discuss numerical aspects of the presented theoretical results. The problem of parallelization of MC algorithms is tangentially touched as well. Further work is necessary to find precise estimates of the necessary number of averaging steps which is sufficient for the efficent work of stable MC algorithms in multidimensional situation.
- Stochastic stability and parallelism in the Monte Carlo method.
- Stochastic and quasistochastic computations
- Monte Carlo method for solving ODE systems
- Comparing M/G/1 queue estimators in Monte Carlo simulation through the tested generator ``getRDS and the proposed ``getLHS using variance reduction
- Analysis of the Monte-Carlo error in a hybrid semi-Lagrangian scheme
- Sparsified Randomization Algorithms for large systems of linear equations and a new version of the Random Walk on Boundary method
- scientific article; zbMATH DE number 4094741 (Why is no real title available?)
- scientific article; zbMATH DE number 2206850 (Why is no real title available?)
- An efficient Monte Carlo scheme for Zakai equations
- Stochastic computational methods and experiment design
- On stochastic and quasistochastic methods for solving equations
- On construction of parallel algorithms in problems of computational mathematics
- Stochastic stability of the Monte Carlo method (the case of operators)
- Optimization of the quasi-Monte Carlo algorithm for solving systems of linear algebraic equations
This page was built for publication: Monte Carlo difference schemes for the wave equation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4331089)