H-Matrix and Block Error Tolerances
arXiv:1110.28071.213 citationsh-index: 20
Originality Synthesis-oriented
AI Analysis
Provides a practical improvement for H-matrix-based solvers in computational science, though incremental in nature.
The paper presents a method for mapping error tolerances in H-matrix approximations to block error tolerances, achieving 1.5–5x efficiency gains for singular kernels with order >1 at lower computational cost.
We describe a new method to map the requested error tolerance on an H-matrix approximation to the block error tolerances. Numerical experiments show that the method produces more efficient approximations than the standard method for kernels having singularity order greater than one, often by factors of 1.5 to 5 and at a lower computational cost.