Toggle menu
71
399
28
3.6K
Medusa: Coordinate Free Meshless Method implementation
Toggle preferences menu
Toggle personal menu
Not logged in
Your IP address will be publicly visible if you make any edits.

Solving sparse systems: Difference between revisions

From Medusa: Coordinate Free Meshless Method implementation
No edit summary
No edit summary
Line 13: Line 13:
* iterative: bicgstab, cg
* iterative: bicgstab, cg


Solving a simple sparse system $A x = b$ with $A = \begin{bmatrix}  -2 & 1 & \\ 1 & \ddots & \ddots \\ & \ddots & \end{bmatrix}$ and $b = \begin{bmatrix} 1 \\ \vdots \\ 1 \end{bmatrix}$ with dimension $n$
Solving a simple sparse system $A x = b$ for steady space of heat equation in 1d with $n$ nodes.
has the following timings in seconds:
has the following timings in seconds:


Line 26: Line 26:
| 0.16
| 0.16
| 0.28
| 0.28
| /
| 0.04
|-
|-
! SparseLU
! SparseLU
Line 33: Line 33:
| 0.69
| 0.69
|-
|-
! Bicgstab / Krylov
! BICGStab / Krylov
| ??
| ??
| 0.39
| 0.39
| ??
| 0.53
|}
|}


Preconditioners matter a lot.
Incomplete LU preconditioner was used for Eigen BICGStab.

Revision as of 11:25, 16 March 2017

There are many methods available for solving sparse systems. We compare some of them here.

Mathematica has the following methods available (https://reference.wolfram.com/language/ref/LinearSolve.html#DetailsAndOptions)

  • direct: banded, cholesky, multifrontal (direct sparse LU)
  • iterative: Krylov

Matlab has the following methods:

Eigen has the following methods: (https://eigen.tuxfamily.org/dox-devel/group__TopicSparseSystems.html)

  • direct: sparse LU
  • iterative: bicgstab, cg

Solving a simple sparse system $A x = b$ for steady space of heat equation in 1d with $n$ nodes. has the following timings in seconds:

$n = 10^6$ Matlab Mathematica Eigen
Banded 0.16 0.28 0.04
SparseLU / 1.73 0.69
BICGStab / Krylov ?? 0.39 0.53

Incomplete LU preconditioner was used for Eigen BICGStab.