What question did this study set out to answer?

To enhance the performance of algebraic multigrid methods for scalar and point-block systems through algorithmic modifications.

On improving the robustness and scalability of shared-memory AMG solvers for point-block problems

View Full Paper

Ask AI

Bookmark

Authors

Algorithms enhance performance in multilevel structures for point-block systems, indicating better scalability and accuracy.

To enhance the performance of algebraic multigrid methods for scalar and point-block systems through algorithmic modifications.
Explored basic operations like transposing and multiplying sparse matrices.
Considered techniques for selecting maximum independent subsets in graph connections.
Evaluated approaches for constructing a prolongation operator and aggressive coarsening.
An extended prolongation operator significantly improves accuracy but increases operational complexity.
Filtering small weights from the prolongation operator helps mitigate complexity issues.
Numerical experiments on various matrices confirmed the effectiveness of the proposed methods.

Konshin et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark