Post-graduate theses
Current Record: 120 of 125
|
Identifier |
000402866 |
Title |
Performance comparison of preconditioners & solvers for big, sparse, complex symmetric linear systems |
Author
|
Αυγερινός, Σταύρος
|
Thesis advisor
|
Πλεξουσάκης, Μιχαήλ
|
Abstract |
The aim of this thesis is to investigate and compare two solvers for big, sparse, complex,
symmetric, linear system of equations. Large sparse linear systems of equations appear
in most applications of scientific computing. In particular, discretization of PDEs with
the finite element method (FEM) or with the finite difference method (FDM) leads
to such problems. First we investigate the PARDISO package, a high-performance,
memory efficient direct solver. Then we analyze the QMRPACK, an iterative solver
implementing several of the QMR algorithms. More specifically, we use the QMR algorithm
based on the look-ahead coupled two-term recurrence Lanczos process and the
simplified no-look-ahead version of the same algorithm, both equipped with ILUT and
SSOR preconditioners. In conclusion, we compare the runtime between the two packages
with test matrices formed by numerical methods.
|
Language |
English |
Issue date |
2015-11-20 |
Collection
|
School/Department--School of Sciences and Engineering--Department of Mathematics and Applied Mathematics--Post-graduate theses
|
|
Type of Work--Post-graduate theses
|
Permanent Link |
https://elocus.lib.uoc.gr//dlib/8/c/e/metadata-dlib-1474282229-642715-1633.tkl
|
Views |
328 |