|
[Sponsors] |
September 28, 2008, 03:52 |
AMG in hypre
|
#1 |
Guest
Posts: n/a
|
Dose anybody use AMG in hypre to solve fluid flow? I would appreciate your comment.
|
|
September 29, 2008, 06:18 |
Re: AMG in hypre
|
#2 |
Guest
Posts: n/a
|
Yes, I use it as the velocity preconditioner in a coupled solver for incompressible flow of power-law fluids. I use it through the PETSc interface and have found ML to frequently be more performant and use less memory, although the robustness of BoomerAMG seems to be better. What do you want to do.
|
|
September 29, 2008, 09:00 |
Re: AMG in hypre
|
#3 |
Guest
Posts: n/a
|
Thanks! I want to use it solve incompressible flow in a general unstructured fluid solver. I would appreciate your suggestions.
|
|
September 29, 2008, 09:24 |
Re: AMG in hypre
|
#4 |
Guest
Posts: n/a
|
I would recommend interfacing to PETSc instead of directly to BoomerAMG since it will be much easier to try alternatives like ML (everything becomes a command-line argument). Do you want a coupled solver or do you use projection methods? If you are solving the coupled equations, you may be interested in some Schur-complement preconditioning which is available in petsc-dev (http://www-unix.mcs.anl.gov/petsc/pe...ers/index.html) but not yet in the released version.
I am not aware of any publicly available parallel algebraic multigrid package for indefinite problems. There has been some success with Vanka smoothers, but the standard method is to use Schur complement preconditioning where multigrid is used as a preconditioner for the velocity block. Good references below. @article{benzi2005nss, title={{Numerical solution of saddle point problems}}, author={Benzi, M. and Golub, G.H. and Liesen, J.}, journal={Acta Numerica}, volume={14}, pages={1--137}, year={2005}, publisher={Cambridge Univ Press} } @article{deniet2007tps, title={{Two preconditioners for saddle point problems in fluid flows}}, author={de Niet, AC and Wubs, FW}, journal={Int. J. Numer. Meth. Fluids}, volume={54}, pages={355--377}, year={2007} } @article{kay2002pss, title={{A preconditioner for the steady-state Navier-Stokes equations}}, author={Kay, D. and Loghin, D. and Wathen, A.}, journal={SIAM Journal on Scientific Computing}, volume={24}, number={1}, pages={237--256}, year={2002}, publisher={SIAM} } |
|
June 5, 2009, 19:35 |
Hypre
|
#5 |
New Member
Join Date: Jun 2009
Posts: 1
Rep Power: 0 |
Hi every one !
I have a human head model. Simulation of this model generates large sparse symmetric positive definite matrices, in the form of Ax = b. Now I have matrices A as large as 1 gb, and same is the vectors b. I would like to use Hypre for solving linear equation, any one who is familiar with hypre and its use. I have these questions. 1 - What can be the best way to solve my matrix A and vector b which has values from my simulations, using PCG of Hypre library. 2 - I want to solve this equation on many processors say more than 100, which can be the optimal approach. I'll be really grateful for your replays. BRs. Ahnav |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
v4 & AMG warning msg | more work | Siemens | 9 | September 7, 2022 10:10 |
Quarter Burner mesh with periosic condition | SamCanuck | FLUENT | 2 | August 31, 2011 12:34 |
AMG versus ICCG | msrinath80 | OpenFOAM Running, Solving & CFD | 2 | November 7, 2006 16:15 |
Problem with jet flow | braennstroem | OpenFOAM Running, Solving & CFD | 16 | May 15, 2006 03:23 |
AMG vs ICCG | eugene | OpenFOAM Running, Solving & CFD | 2 | June 7, 2005 15:07 |