[aspect-devel] Boundary composition

[John Naliboff]

Hi Lev,

The DG option appears to be working fine on my end.  Can you send over 
the full output file so I can see at what stage the segmentation fault 
occurred?
If you like, I can also try running your parameter file to see if it 
works on my end.

Cheers,
John

*************************************************
John Naliboff
Assistant Project Scientist, CIG
Earth & Planetary Sciences Dept., UC Davis

On 09/29/2016 05:14 AM, Lev Karatun wrote:
> Hi John,
>
> thank you for the suggestion. When I tried to use the option (using 
> the latest versions of dealII and aspect) I got a segfault unfortunately:
>
> MXM: Got signal 11 (Segmentation fault)
> ==== backtrace ====
>     0  /lib64/libc.so.6(+0x326a0) [0x7fffe7b8b6a0]
>     1 
>  /home/r/russ/lkaratun/aspect/dealiidev/lib/libdeal_II.so.8.5.0-pre(_ZNK6dealii12CellAccessorILi3ELi3EE29neighbor_of_neighbor_internalEj+0x55) 
> [0x7ffff
> 6bb1715]
>     2 
>  /home/r/russ/lkaratun/aspect/dealiidev/lib/libdeal_II.so.8.5.0-pre(+0x3a736cc) 
> [0x7ffff60f76cc]
>     3 
>  ../aspect(_ZN6aspect9SimulatorILi3EE19setup_system_matrixERKSt6vectorIN6dealii8IndexSetESaIS4_EE+0x6ff) 
> [0xb059cf]
>     4  ../aspect(_ZN6aspect9SimulatorILi3EE10setup_dofsEv+0x414) 
> [0xb293e4]
>     5  ../aspect(_ZN6aspect9SimulatorILi3EE3runEv+0x1d6) [0xb29996]
>     6  ../aspect(main+0x1a9) [0x865f59]
>     7  /lib64/libc.so.6(__libc_start_main+0xfd) [0x7fffe7b77d5d]
>     8  ../aspect() [0x7ca27d]
>
>
> Best regards,
> Lev Karatun.
>
> 2016-09-28 12:58 GMT-04:00 John Naliboff <jbnaliboff at ucdavis.edu 
> <mailto:jbnaliboff at ucdavis.edu>>:
>
>     Hi Lev,
>
>     Initial composition would be the correct boundary choice for what
>     you are trying to do.
>
>     One thing that may help is to use the discontinuous composition
>     discretization option, if you have not tried it already.  This
>     helps tracks compositional field boundaries more accurately
>     through time and removes compositional field diffusion effects. 
>     Certainly helped with some of the models I am running.
>
>     Below is a section from a .prm file where I use this option.  FYI,
>     you will want to use the development version of dealii (master
>     branch from Github) if you use the discontinuous composition
>     discretization option. The model will run with dealii version
>     8.4.1, but the development version helps reduce the memory
>     requirements dramatically.
>
>     If this option helps, please let us know.
>
>     Cheers,
>     John
>
>     *************************************************
>     John Naliboff
>     Assistant Project Scientist, CIG
>     Earth & Planetary Sciences Dept., UC Davis
>     **
>
>     *# Use disccontinous composition bound preserving limiter*
>     *subsection Discretization*
>     *set Use discontinuous composition discretization = true*
>     *subsection Stabilization parameters*
>     *set Use limiter for discontinuous composition solution = true #
>     apply the limiter to the DG solutions*
>     *set Global composition maximum = 1.0, 1.0, 1.0, 1.0*
>     *set Global composition minimum = 0.0, 0.0, 0.0, 0.0*
>     *end*
>     *end*
>
>     On 09/27/2016 01:09 PM, Lev Karatun wrote:
>>     Hi,
>>     I'm doing a lithospheric scale convergence model, imposing influx
>>     (along the x axis) at the top half of the side walls (y=0, YMAX),
>>     and outflux at the bottom half. Over the course of the
>>     simulation, the lithosphere at the side of the model (the part
>>     that was injected) gets progressively thinner. The vertical (z)
>>     and lateral (y) velocities are set to zero at the boundary, but I
>>     noticed that a lateral flow is still happening very close to the
>>     side wall, and my guess is that due to the material mixing the
>>     outermost cells don't consist of lithoshperic material anymore,
>>     but only of, let's say, 80% lithospheric and 20% mantle. So the
>>     question is: is it possible to manually restrict the incoming
>>     material to be of the same properties as it was in the outer
>>     cells at the beginning of the simulation (as opposed to the
>>     previous time step, which as fast as I understand is defining the
>>     material right now -- correct me if I'm wrong). Also, I was
>>     wondering if parameters in section Boundary composition model are
>>     related to this? I tried setting Model name to "initial
>>     composition" but it didn't seem to change anything.
>>     Thanks in advance.
>>     Best regards,
>>     Lev Karatun.
>>
>>     _______________________________________________
>>     Aspect-devel mailing list
>>     Aspect-devel at geodynamics.org <mailto:Aspect-devel at geodynamics.org>
>>     http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel
>>     <http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel>
>     _______________________________________________ Aspect-devel
>     mailing list Aspect-devel at geodynamics.org
>     <mailto:Aspect-devel at geodynamics.org>
>     http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel
>     <http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel> 
>
> _______________________________________________
> Aspect-devel mailing list
> Aspect-devel at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/aspect-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/aspect-devel/attachments/20160929/6be4e160/attachment.html>