Hi all,
Im currently learning Elmer and Ive got a few questions:
-is there any website where new users could download a tutorial cases solved, it would be great to have them to compare,
-why when I do some tutorial simulation, exactly as instruction say there is a convergence problem,
-is there any references about preconditioners, i mean ILU (i found sth like this: http://ta.twi.tudelft.nl/nw/users/vuik/ ... e_2007.pdf). But Elmer have ILU from 0 to 9, what is the difference, how it works and how it depends on convergence behavior.
Anyone knows the answer?
Best regards
Althenor
Few question about convergence and preconditionals (ILU)
Re: Few question about convergence and preconditionals (ILU)
Hi Althenor,
In that page, you can download example cases, some with a .sif file
http://www.csc.fi/english/pages/elmer/examples
You even have a course in the glaciological section
Cheers
lionel
In that page, you can download example cases, some with a .sif file
http://www.csc.fi/english/pages/elmer/examples
You even have a course in the glaciological section
Cheers
lionel
-
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Few question about convergence and preconditionals (ILU)
Hi Althenor
1) I created yesterday a subforum in which people can submit their cases of moderate size. One could place to look is the "tests" directory in your Elmer installation which includes around 140 minimalistic Elmer cases.
2) If you do *exactly* as in the tutorial there should not be a convergence issue, if things haven't gotten broken. However, there are some cases that will diverge if the defaults are not changed. I think at least the bending beam and step flow with k-epsilon model belongs to this category.
3) ILUn is the incomplete LU decomposition of A with the same sparsity structure as that of A^n.
-Peter
1) I created yesterday a subforum in which people can submit their cases of moderate size. One could place to look is the "tests" directory in your Elmer installation which includes around 140 minimalistic Elmer cases.
2) If you do *exactly* as in the tutorial there should not be a convergence issue, if things haven't gotten broken. However, there are some cases that will diverge if the defaults are not changed. I think at least the bending beam and step flow with k-epsilon model belongs to this category.
3) ILUn is the incomplete LU decomposition of A with the same sparsity structure as that of A^n.
-Peter