Hello all,
I have a question about my code. I would like to simulate a shear test.
During the insertion of particles I got the following error
"ERROR on proc _: assertion failed (../region_neighbor_list_I.h:170)"
But if I increase the size of my domain following the z-axis the problem is solved.
You will find the file in the attachment.
The size of the box region :
region ri block 0.0 0.02 0.0 0.02 0.0 0.001 units box
(The problem is solved wiht region ri block 0.0 0.02 0.0 0.02 0.0 0.014 units box but I have too many particles)
The sizes of particles :
fix pts1 all particletemplate/sphere 32452843 atom_type 1 density constant 1000 radius constant 0.0002
fix pts2 all particletemplate/sphere 49979687 atom_type 1 density constant 1000 radius constant 0.00015
fix pts3 all particletemplate/sphere 67867979 atom_type 1 density constant 1000 radius constant 0.0001
fix pdd all particledistribution/discrete/massbased 15485863 3 pts1 0.42 pts2 0.36 pts3 0.22
fix ins all insert/pack seed 86028121 distributiontemplate pdd vel constant 0. 0. 0.0 insert_every once overlapcheck yes all_in yes particles_in_region 8750 region ri
The neighbor :
neighbor 0.0008 bin
neigh_modify delay 0
Thanks in advance for your answer
Alexandre
| Attachment | Size |
|---|---|
 My first test | 5.04 KB |
 My second test only with 1 particle and the mesh | 3.5 MB |
AlexSM | Tue, 04/13/2021 - 11:23
A factor scale 0.6 for rectangular mesh ?
Hello,
I have done some test to understand better the case.
I define a box 0 1 0 1 0 z (where z is changing).
I define a particle at the center of the box to run the code
I define a rectangular mesh at y=0 and the size of the mesh is 1 following the x-axis and z following the z-axis.
When the scale z/x (x =1) is larger than 0.6 I have not error but when it is smaller the error comes. And it is the same for other neighbor values or y position of the mesh. I have tried with x = 2 and again the same limit scale appears.
I have generated my mesh from Salome by the function "create a face".
Do you have any suggestion?
Thanks
Alexandre
mschramm | Tue, 04/13/2021 - 20:22
upto 8 processors
Hello,
Did you post the script that provides an error?
I was able to run the given script (after commenting out your stl file input...) with, upto, 8 cores.
AlexSM | Wed, 04/14/2021 - 09:56
I edited in the main
Hello,
I have just edited my post. You will find a folder ("My second test") with :
-my version of liggght
-the main (only 1 run, 1 particle and the mesh)
- 4 .stl
Face_1000_600 and Face_2000_1200 work on my computer. Face_1000_500 and Face_2000_1000 do not work on my computer. Files are built as Face_lenghtfollowingx_lenghtfollowingz.
I am running on 4 processors, is it the deal?
mschramm | Wed, 04/14/2021 - 16:49
using upto 8 cores
Hello,
Using the faces that you provided, I was not able to reproduce your error.
I also was not able to use your version of liggghts (error while loading shared libraries: libmpi_cxx.so.40).
AlexSM | Thu, 04/15/2021 - 15:09
What I found in the src
Hello,
I take a look into the src. It seems the "nall" parameter from SurfaceMesh::buildNeighbours() in surface_mesh_I.h changes. This parameter is a sum of "this->sizeLocal()" and "this->sizeGhost()" and it is this latter which changes in particular.
In the case of Face_1000_600, 4 cores and a domain 0 1 0 1 0 0.6 : It is working and I have nall = 12.
In the case of Face_1000_500, 4 cores and a domain 0 1 0 1 0 0.5 : It is not working and I have nall = 16. The error happens when the iteration ibin equals to 12 (the center of the bin seems too far away of the domain).
But in the case of Face_1000_500, 4 cores and a domain 0 1 0 1 0 0.6 : It is working and I have nall = 12.
I do not understand at all how this parameter sizeGhost() is computed and why the size of my domain modifies it.
How is it possible you are not able to use my version ? I get it from a "make auto" command in linux Ubuntu 20.04.
Alexandre
mschramm | Thu, 04/15/2021 - 16:04
different linked libraries
I am running ubuntu 20 on WSL2 that originally had ubuntu 18... It could be many possible differences.
What everything should come down to is what version of LIGGGHTS are you using (via the commit number).
AlexSM | Fri, 04/16/2021 - 09:34
Version LIGGGHTS-PUBLIC 3.8.0
You will find the asked information:
Version LIGGGHTS-PUBLIC 3.8.0 git commit 3b38a7dbe22018f905d4c5f792ab5ce644ef594e
AlexSM | Thu, 04/29/2021 - 16:35
Any news?
Hello,
I am coming back to this post. Does the problem come from the commit?
Alexandre
kardous | Thu, 04/29/2021 - 18:33
Mesh error
Hello,
While running my script with Liggghts I got this error that I don't undrestand. Can u help me with it please?
ERROR: Mesh error - root causes: (a) mesh empty or (b) dimensions too small - use different unit system (/build/liggghts-YO7u74/liggghts-3.8.0+repack1/src/multi_node_mesh_parallel_I.h:566
thank u for your help,
SK
SHUBHAM AGARWAL | Sun, 07/18/2021 - 22:18
I faced that too
Hello
Just to in inform m that I faced the similar problem. I have installed same version on two different PCs, one is with 6 core an other with 8 cores. In simulation I demand 4 cores, on 6 core PC I sometime face this error, while never on 8 core PC.
Shubham