[Gmsh] Random behaviour in Gmsh

[Serban.Georgescu at uk.fujitsu.com]

Hi Geordie and thanks for your quick reply.

> Are you optimizing the mesh? This isn't done by default but can be
> quite important for subsequent finite element analysis.

For this particular test case I was not optimizing the mesh. I am aware of the fact that optimizing the mesh will in general lead to better results, however the fact that I get a different result still does not change and that is the real problem. 

> Dipl.-Ing. A. P. reported a few weeks ago being able to fix the
> randomness by meshing from the command line.  I haven't tried it, but
> it seems like an easy fix.

I have read the post, but in my case this behavior appears from the command line as well. Going inside the source code I saw plenty of calls to rand() ... that's why I was thinking about fixing the seed.

Cheers,
Serban


-----Original Message-----
From: gdmcbain at gmail.com [mailto:gdmcbain at gmail.com] On Behalf Of Geordie McBain
Sent: 26 May 2011 08:49
To: Serban Georgescu
Cc: gmsh at geuz.org
Subject: Re: [Gmsh] Random behaviour in Gmsh

> Basically, every time I run Gmsh on the same file I get a different mesh. Sometimes the differernce is very small, but there are times when it quite large. For example, for the same CAD file and meshing parameters, half of the time I am getting a good mesh while half of the time I get one that crashes the solver (because of a few "flat" elements).

Are you optimizing the mesh? This isn't done by default but can be
quite important for subsequent finite element analysis.

 http://geuz.org/gmsh/doc/texinfo/gmsh.html#index-g_t_002doptimize_005b_005fnetgen_005d-34

> Doing some googleing on the subject I found out that this is the intended behaviour of Gmsh and that this randomness cannot be stopped.  However, it is very inconvenient for us, one reason being that we cannot support other users since we cannot reproduce their exact problems.

Dipl.-Ing. A. P. reported a few weeks ago being able to fix the
randomness by meshing from the command line.  I haven't tried it, but
it seems like an easy fix.

 http://www.geuz.org/pipermail/gmsh/2011/006347.html
______________________________________________________________________
                                        
 Fujitsu Laboratories of Europe Limited
 Hayes Park Central, Hayes End Road, Hayes, Middlesex, UB4 8FE
 Registered No. 4153469
 
 This e-mail and any attachments are for the sole use of addressee(s) and
 may contain information which is privileged and confidential. Unauthorised
 use or copying for disclosure is strictly prohibited. The fact that this
 e-mail has been scanned by Trendmicro Interscan does not guarantee that 
 it has not been intercepted or amended nor that it is virus-free.