problem in tutorial 1 (norm of solution)

Numerical methods and mathematical models of Elmer
Post Reply
wing
Posts: 10
Joined: 14 Oct 2009, 11:49

problem in tutorial 1 (norm of solution)

Post by wing » 14 Nov 2009, 11:37

Hello everybody,
I tried Tutorial 1 (Heat Equation - Temperature field of a solid object) which use pump_carter_sub.stp according to ElmerTutorials.pdf (September 22, 2009). I found that the norm of my solution is about 432.4, but it says it should be around 389.853 in the ElmerTutorials.pdf (page 9).
I think it's unusual and I'd like to figure it out what causes the difference. The solver? or any update of the material library?

P.S.
1) I use the "default netgen mesh" for calculation, 16159 nodes and 65689 tetrahedral elements.
2) Elmer Solver 5.5.0 (Rev: 4266)
3) Linux x86_64

attached is the sif and Solver log for reference.
Attachments
Solver_log.txt
(2.37 KiB) Downloaded 162 times
case.sif
(1.95 KiB) Downloaded 108 times

raback
Site Admin
Posts: 3498
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: problem in tutorial 1 (norm of solution)

Post by raback » 14 Nov 2009, 14:42

The difference is due to the fact the original tutorial was done with a mesh provided by tetgen. After some imrpovement of the netgen interface it became the default mesher. The new norm just wasn't updated to the documentation. Thanks for pointing this out.

The reported norm is a L2 norm of the solution vector, |x|=\sqrt(\sum_i^n x_i*x_i /n), and therefore different meshes may have large variations in the norm even though the distributions would be rather similar.

-Peter

Post Reply