Title | ||
---|---|---|
FLLOP: A Massively Parallel Solver Combining FETI Domain Decomposition Method and Quadratic Programming |
Abstract | ||
---|---|---|
FLLOP (FETI Light Layer On top of PETSc) is a novel solver for linear systems and more general quadratic programming problems. It connects quadratic programming algorithms with FETI domain decomposition methods that bring high parallel scalability and at the same time high accuracy and robustness needed in engineering applications. According to its name, FLLOP forms an extension of the renowned PETSc software framework for parallel numerical solution of partial differential equations. It can be thus adopted by PETSc users with relative ease. The above-mentioned facts makes FLLOP a unique project. It now forms a solver part of our emerging PERMON toolbox for Parallel, Efficient, Robust, Modular, Object-oriented, Numerical simulations. We introduce one more PERMON tool, Permon Cube, which serves as a benchmark generation tool for FLLOP in massively parallel environment. |
Year | DOI | Venue |
---|---|---|
2014 | 10.1109/HPCC.2014.56 | HPCC/CSS/ICESS |
Keywords | Field | DocType |
linear systems,matrix decomposition,quadratic programming,finite element analysis,scalability,vectors | FETI,Massively parallel,Computer science,Parallel computing,Numerical partial differential equations,Robustness (computer science),Theoretical computer science,Quadratic programming,Solver,Software framework,Domain decomposition methods,Distributed computing | Conference |
Citations | PageRank | References |
0 | 0.34 | 5 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Václav Hapla | 1 | 22 | 5.30 |
martin cermak | 2 | 11 | 4.44 |
Alexandros Markopoulos | 3 | 31 | 7.16 |
David Horák | 4 | 8 | 2.24 |