Яндекс.Метрика

T. Khachkova, V. Lisitsa

Том: 21st International Conference on Computational Science and Its Applications, ICCSA 2021. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Cagliari, Italy, September 13-16, 2021)
Том: 12949 LNCS , Год издания: 2021
Многотомное издание: 21st International Conference on Computational Science and Its Applications, ICCSA 2021. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Cagliari, Italy, September 13-16, 2021)
Издатель: Springer International Publishing , Место издания: Berlin
Страницы: 532-545

Аннотация

In this paper, we present an iterative solver for Poisson equation. The approach is essentially oriented on upscaling the physical parameters of porous materials, such as electrical resistivity, thermal conductivity, pressure field computation. The algorithm allows solving Poisson equation for strongly heterogeneous media with small-scale high-contrast heterogeneities. The solver is based on the Krylov-type iterative method with a pseudo-spectral preconditionner, thus the convergence rate is independent on the sample size, but sensitive to the contrast of physical properties in the model. GPU-based implementation of the algorithm allows performing simulations for the samples of the size 400 3 using a single GPU.
индекс в базе ИАЦ: 035223