LIGGGHTS 2.1.2 (2.1.1 and 2.06 tested): fix insert/pack error

Submitted by qfhou on Thu, 09/27/2012 - 08:24

related input:
"fix ins12 all insert/pack seed 100001 distributiontemplate pdd7 vel constant 0. 0.1 -0. insert_every once overlapcheck yes all_in yes particles_in_region 3 region bc7

run 250000 upto

fix ins13 all insert/pack seed 100001 distributiontemplate pdd3 vel constant 0.1 0. -0. insert_every once overlapcheck yes all_in yes particles_in_region 3 region bc3

fix ins14 all insert/pack seed 100001 distributiontemplate pdd4 vel constant 0. 0.1 -0. insert_every once overlapcheck yes all_in yes particles_in_region 3 region bc4"

error message:

"ERROR: Fix insert/pack (id ins13): 'start' step can not be before current step (fix_insert.cpp:375)
"

If 'fix ins 13..." is commented out. It is ok. However, if the 'run 250000 upto' is changed, it may happen to another 'fix insert/pack"

I wonder if this is a bug or there is a limitation of the number of 'fix insert/pack". If there is a limitation, I do not know why the error is not happened to the last 'fix insert/pack'. Thanks

btw, liggghts 2.1.2 has greatly improved the mesh detection. Thanks.

AttachmentSize
Plain text icon cubic.stl_.txt1.06 MB
Plain text icon in.packing.txt7.2 KB
Plain text icon log.liggghts.txt17.55 KB

Forums:

ckloss's picture

ckloss | Thu, 09/27/2012 - 12:10

Hi qfhou,

at first glance this looks like a bug, but I am sure this can easily be fixed - can you post a full script please?

Christoph

qfhou | Fri, 09/28/2012 - 02:22

Hi Chris,
Thanks for your help.
The full script is attached to the original post where the "ins18" is commented out. If there is any other problem, please let me know.
Please be noted that liggghts 2.1.2 won't accept 'neigh_modify delay 0' any more.

Regards,
Qinfu

ckloss's picture

ckloss | Fri, 09/28/2012 - 14:09

Hi Qinfu,

I can't reproduce the issue on my side. However, you dind't attach the stl file. Can you please post the exact input script that causes the error, preferably a small case that runs only a few seconds?

>>Please be noted that liggghts 2.1.2 won't accept 'neigh_modify delay 0' any more.
for me it works perfectly fine, none of the code related to this has been changed recently. can you please comment on that in more detail?

Christoph

qfhou | Tue, 10/02/2012 - 01:31

Hi Chris,
The stl, log and input are attached to the first post.
For some unknown reasons, I cannot reproduce the 'neigh_modify delay 0' error. You can ignore it.
Kind regards,
Qinfu

ckloss's picture

ckloss | Tue, 10/02/2012 - 20:08

Qinfu,

I tried to reproduce the error, but it simply doesn't come up on my side. Can you comment on which compiler you used?

Christoph

qfhou | Wed, 10/31/2012 - 09:24

is there any problem with my gcc version? Thanks

qfhou | Wed, 11/14/2012 - 06:57

Hi Christ, what compiler do you use? thanks and regards, Qinfu
It seems that there are some other problems for running liggghts properly.

ckloss's picture

ckloss | Wed, 11/14/2012 - 08:39

I am using gcc 4.4.3 on Ubuntu 10.0.4
We have had troubles with icc earlier

Cheers, Christoph

qfhou | Thu, 11/15/2012 - 01:25

Thanks. I may try gcc 4.4.3.

Another question related to mesh 1) as the new liggghts 2.2.2 support tet mesh, how do you generate the vtk file or convert a STL mesh to vtk format? 2) if the stl mesh is a tet mesh, liggghts can still read it and run. Will it cause any problem known?
Thanks and regards, Qinfu

jtvanlew | Tue, 10/29/2013 - 03:52

Hi Christoph,
I'm running into a "'start' step can not be before current step" error that I can't sort out on my own. Google implies this is the only other discussion about it online. I'm curious if this might still be a bug or if it was resolved and my problem lies somewhere else in my code.

Unfortunately my error comes at the end of a pretty complicated script. Essentially, I'm using the python-wrapping to interface with LIGGGHTS every M steps and pull out pair/grain contact forces. When forces are larger than a critical value, I have the pebble 'break' and replace it with a fix insert/pack of smaller particles. Finally I tell LIGGGHTS to keep running another M steps to loop.

It all works fine but once in a while, after there have been a few insertions, i get this error. I never specify which time steps to begin or which time steps to run 'up to' -- i just run it. So I'm confused by the error. I randomly assign strength to pebbles so the amount of pebble breakage is somewhat random every time I run the same code. So the error doesn't occur at exactly the same time but does happen without fail each time I run my routine.

If it would help debugging, I can send a zip of the files in my routine.

jon

ckloss's picture

ckloss | Wed, 10/30/2013 - 10:21

Hi jon,

if you can break down your "pretty complicated script" to something simple that can reproduce the issue within a minute or so, I'm happy to have a look

Cheers
Christoph