Core Concepts
Developing implicit rank-adaptive schemes for matrix differential equations.
Abstract
The study introduces robust implicit adaptive low-rank time-stepping methods for matrix differential equations. Inspired by the dynamic low-rank approximation (DLRA) technique, the schemes aim to address convergence issues in equations with cross terms. By merging row and column spaces, stability is proven, and local truncation errors are estimated. The proposed methods are benchmarked in various tests to demonstrate robust convergence properties.
Stats
arXiv:2402.05347v3 [math.NA] 17 Mar 2024
Research supported by DOE Office of Advanced Scientific Computing Research under the Advanced Research in Quantum Computing program, subcontracted from award 2019-LLNL-SCW-1683, NSF DMS-2208164, and Virginia Tech.