Segment fault coupling LIGGGHTS and PARSCALE

Submitted by alberthappy on Mon, 09/21/2015 - 01:03

Hello,

My last post is about the miss of integrator file in running the parscale example provided with LIGGGHT 3.2.1 (patch) code. I just copied this file from other PARSCALE examples. But when I ran the example again, I got a new error this time and it seems much worse than the previous one. Here is the error:

*** Process received signal ***
["ID"-ThinkPad-T430:26148] Signal: Segmentation fault (11)
["ID"ThinkPad-T430:26148] Signal code: Address not mapped (1)
["ID"-ThinkPad-T430:26148] Failing at address: 0x1910
["ID"-ThinkPad-T430:26148] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10340) [0x7fb2f3a2e340]
["ID"-ThinkPad-T430:26148] [ 1] $HOME/LIGGGHTS_3_2_1p/src/lmp_ubuntuVTK_debug() [0x9255fc]
...

I tried to make a dynamic shared library intead of a static one and it gave the same error. But I could separately run the PARSCALE executable without any problem. I tried the compilation in both ubuntu14.04 and Redhat 6.5. This problem was occurring all the time. Any idea on what causes this problem?

Thank you all for the sharing of knowledge and experiences.

Feng

tforg | Wed, 11/04/2015 - 08:39

Hello Feng,

apologies for the late reply - I just saw this.

The problem is caused by the Id´s of the particles in ParScale.
In 'heatTransferBed_paScal/pascal/0/radius.json' the first particle Id should be 1 (instead of 0) and the last 800 (instead of 799). The same accounts for '/heatTransferBed_paScal/pascal/0/heat.json'.

You can do this quickly by hand or using the 'heatTransferBed_paScal/genInput.m' - script to generate them automatically.
Note, that in this script you should change 'index=iP-1' to 'index=iP'!

We will fix this very soon!

Hope that helps!
Thomas

alberthappy | Mon, 11/16/2015 - 20:21

Thomas,

Thanks for your reply. Can you tell us some main features of your new version code? I ask this because I have a plan to develop my own single particle code which couples the intraparticle mass transfer, heat transfer and momentum equation.

Best regards
Feng

Fenglei Qi

tforg | Tue, 12/01/2015 - 11:27

Dear all,

please add

"fix heatTransCoeff all property/atom heatTransCoeff scalar no yes no none #heat transfer coefficient"

to the LIGGGHTS input file in order to make the simulation run.

Cheers,
Thomas

vkoppejan | Wed, 01/06/2016 - 15:40

I have changed the geninput.m file, added the HT-coeff fix.

when I run the "liggghts < in.liggghts" I still get the following error

ERROR on process 0 (in file model_base.cpp, line 117):
[Process 0/1] Please supply this file:
[Process 0/1] ./pascal/settings/integrator.json

Would it be possible to get a working version of the the tutorial?

Thanks in advance

Victor

vkoppejan | Fri, 01/08/2016 - 14:34

I just copied a integrator.json file from another example case. This solved the problem.