Elliptic Solvers for Adaptive Mesh Refinement Grids

Technical Report · Thu Jun 03 04:00:00 EDT 1999

DOI:https://doi.org/10.2172/763231· OSTI ID:763231

Quinlan, D J ^[1]; Dendy, Jr, J E; Shapira, Y

---

1. Los Alamos National Laboratory

We are developing multigrid methods that will efficiently solve elliptic problems with anisotropic and discontinuous coefficients on adaptive grids. The final product will be a library that provides for the simplified solution of such problems. This library will directly benefit the efforts of other Laboratory groups. The focus of this work is research on serial and parallel elliptic algorithms and the inclusion of our black-box multigrid techniques into this new setting. The approach applies the Los Alamos object-oriented class libraries that greatly simplify the development of serial and parallel adaptive mesh refinement applications. In the final year of this LDRD, we focused on putting the software together; in particular we completed the final AMR++ library, we wrote tutorials and manuals, and we built example applications. We implemented the Fast Adaptive Composite Grid method as the principal elliptic solver. We presented results at the Overset Grid Conference and other more AMR specific conferences. We worked on optimization of serial and parallel performance and published several papers on the details of this work. Performance remains an important issue and is the subject of continuing research work.

• Research Organization: Los Alamos National Lab., NM (US)
• Sponsoring Organization: M and A (US)
• DOE Contract Number: W-7405-ENG-36
• OSTI ID: 763231
• Report Number(s): LA-UR-99-2673
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
ALGORITHMS
LOS ALAMOS
OPTIMIZATION
PERFORMANCE