Atoms lost using periodic boundary conditions in LIGGGHTS v3.8.0 with triclinic box

Submitted by rangr1 on
Attachment Size
LIGGGHTS input script 579 bytes
arnom

Triclinic is a feature we normally don't use at DCS (incl. our customers). Because of that we probably won't look into this issue any time soon. If anybody has a fix for the issue please post it here or submit a pull request and we'll be happy to include it in the next release.

Thanks for your understanding,
Arno

Wed, 8 Aug 2018 - 11:37 Permalink
Franzzzzzzzz

If anyone's still interested, this comes from a bug in the src/domain.cpp file, with the line initialising prd_lambda to zeros instead of one. So line 144 of domain.cpp in commit 1230dbd should read:
prd_lamda[0] = prd_lamda[1] = prd_lamda[2] = 1.0;

Wed, 22 Jan 2020 - 02:22 Permalink
jagan1mohan

Hello Franzzzz,

Greetings. I'm having an orthogonal periodic cube filled with particles. When I use nve/sphere in LIGGGHTS, particles stay inside and move across periodic boundary(s) as expected. When I convert box into triclinic, either by "0 0 0 XY XZ YZ" header or "change_box", I loose all particles within few time steps when I use nve/sphere. Do you know how to address this situation? I want to use this triclinic box to apply shear on XY plane but I'm unable to hold particles within even without any applied forces.

Thank you,
Jagan Mohan.

Sun, 8 Mar 2020 - 23:24 Permalink
mschramm

Hello,
A pull request contains the fixes.
Make sure you are using the latest version of liggghts.

Wed, 11 Mar 2020 - 00:10 Permalink
jagan1mohan

In reply to by mschramm

Hello Matt, Thanks for your reply. The output log when I run LIGGGHTS on my system gives following information.

LIGGGHTS (Version LIGGGHTS-PUBLIC 3.8.0, compiled 2019-12-06-20:21:52 by jb, git commit ce1931e377a2945e3fa25dae0eebc1f697009a32)

Is this the latest version? How to update it to latest version?

Thank you,
Jagan Mohan.

Wed, 11 Mar 2020 - 02:19 Permalink
mschramm

In reply to by jagan1mohan

The pull request happened in January.
In the src directory run,
git pull
Then run
make auto
To compile the program again.

Sat, 14 Mar 2020 - 16:20 Permalink
jagan1mohan

Hello Matt, Thanks for your reply. How do we come to know when LIGGGHTS is updated? Is there a reminder email / forum? Perhaps, using the latest version is always better.

Thank you,
Jagan Mohan.

Sat, 6 Jun 2020 - 22:07 Permalink
mschramm

In reply to by jagan1mohan

On github you can "watch" a repository. Then anytime an update happens, you will get an email.

As for updates on a schedule, I don't believe there is one. As far as I know, the public users just hope that an update will eventually happen and rejoice when one happens.

Sun, 7 Jun 2020 - 05:47 Permalink
mofazli

Dear Rangr1,
Did you want to impose the shear flow for that periodic box (Lees-Edwards Boundary Condition) using the triclinic box? If so, could you please tell us how did you that?

Thu, 6 Aug 2020 - 12:02 Permalink

I am setting up a single particle simulation with periodic boundary conditions (PBC) in LIGGGHTS-PUBLIC v3.8.0 and find that the atom is getting lost once it crosses the boundary for a triclinic box. This does not happen when I use LIGGGHTS-PUBLIC v3.5.0 or v3.4.1 for instance.

Please take a look at the input script attached that shows the behaviour described above. This atom is not lost if "block" is used instead of "prism" in my input script. I don't understand why this is happening, and perhaps it may be a bug.