Fix rigid/nvt giving NaN

LAMMPS LAMMPS General Discussion
1

Hello,

I have simulating a fiber system using fix rigid and fix rigid/nvt. The fix rigid part runs find but once it gets to the fix rigid/nvt, I get NaN and I have no clue what the problem is because I’ve used this input file and similar data files and those ran fine.

I’ve attached my input file.

LAMMPS_INPUT.data (38.4 KB)
input.in (856 Bytes)

Please let me know if more information is needed.

Thanks,
Frank

2

Your LAMMPS version and an example of a system that works alongside the system that does not work would help.

Simon

3

There is no problem running your input deck with the current version of LAMMPS 29 August 2024.

4

The version is: LAMMPS (2 Aug 2023 - Update 1)

LAMMPS_INPUT.data (9.5 KB)

The attached input works.

The input that does not work is of linear rigid bodies while the input that does work are square bodies, both made of point particles.

5

Hi Axel,

Has there been any updates to fix rigid/nvt on the way it handles linear rigid systems? I have found that by adding a small perturbation from linear structures, the simulation will run but if it’s strictly linear nvt will give NaN.

Thanks,
Frank

6

I have run simulations with systems like that before and had no issues either. I would suspect that your executables might be tainted by using a broken compiler that miscompiled parts of the code.

You would have to look for yourself. The source code of all versions is freely available and all details are recorded in the git history.

In recent years, I’ve been providing statically linked pre-compiled non-MPI binaries for all x86_64 Linux system, so you can test yourself if any of those are giving the unexpected results. There are also binaries for MacOS and Windows.