[CIG-MC] cookbook4 and v.3.2.0

[Shijie Zhong]

Eric,

Typically, div/v and dp/p are coupled, i.e., if one is convergent, the other must be as well, because they are both associated with how well the continuity equation is solved. Therefore, there is no need to check for both div/v and dp/p, in most cases. 

Cheers,

Shijie


Shijie Zhong, Professor
Department of Physics
University of Colorado at Boulder
Boulder, CO 80309
Tel: 303-735-5095; Fax: 303-492-7935
Web: http://anquetil.colorado.edu/szhong
________________________________________
From: cig-mc-bounces at geodynamics.org [cig-mc-bounces at geodynamics.org] On Behalf Of Eric Heien [emheien at ucdavis.edu]
Sent: Wednesday, May 16, 2012 3:00 PM
To: CIG-MC at geodynamics.org
Subject: Re: [CIG-MC] cookbook4 and v.3.2.0

Hi everyone,

After working with Juan to understand this, we believe we have discovered the cause.  In CitcomS 3.2.0, an additional convergence criterion was added for continuity (in the output this is referred to as "div/v") such that solver iterations continue until div/v is lower than a certain threshold.  In CitcomS 3.1.1 the iterations were ended based only on pressure and/or velocity convergence.

For the model in Cookbook 4 this causes the kind of isolated cell creation Juan discovered, but it is not clear whether this isolated cell is physically reasonable or caused by the changed element size implemented in the cookbook. Juan is investigating this and will report back to the mailing list.

If anyone wants the same solver behavior in 3.2.0 that was used in 3.1.1, they should set the following parameter:

[CitcomS.solver.vsolver]
check_continuity_convergence=0

We are still investigating how this change in convergence criterion will affect other models.  Let me know if you have any questions.

-Eric

On May 8, 2012, at 2:01 AM, Juan Rodríguez wrote:

> Hello,
>
> I am having some troubles with cookbook4 while using version 3.2.0 and I would really appreciate some help.
>
> I am running the cookbook using two different compiled versions of CitcomS.
>
> -       CitcomS-3.1.1 with pyre, Python 2.6.5 and OpenMPI 1.2.8-13.1
>
> -       CitcomS-3.2.0 with pyre, Python 2.6.5 and OpenMPI 1.2.8-13.1
>
> I am running everything on the same machine: 24 processors with 64 GB RAM running under linux (OpenSUSE 11.3).
>
>
> When I run the cookbook with version 3.1.1 the results seem correct and match what is shown in Fig. 6.5 of the manual.  The attached figures v.3.1.1_inital and v.3.1.1_final show the temperature and the velocity field for the initial and final time steps of the cookbook.
>
> However, when I run the cookbook using version 3.2.0 I get completely different results. First, the model does not converge after the maximum number of p-iterations (1000). Second, the final velocity field and temperature do not match the expected result (se figures v.3.2.0_inital and v.3.2.0_final
>
>
>
> I am using the same files to run the cookbook (cookbook4.cfg and coor.dat, also attached).
>
>
>
> I am not sure if there is some kind of problem with the new version or if I am missing something, but I hope someone can help.
>
> Thank you
>
>
> - Juan Rodriguez
>
> <v.3.1.1_final.pdf><v.3.1.1_initial.pdf><v.3.2_final.pdf><v.3.2_initial.pdf>_______________________________________________
> CIG-MC mailing list
> CIG-MC at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-mc

_______________________________________________
CIG-MC mailing list
CIG-MC at geodynamics.org
http://geodynamics.org/cgi-bin/mailman/listinfo/cig-mc