Problems with Lammps and potential files

[sebastianecheverrir]

Dear forum,

I am encountering some problems when using eOn together with lammps potentials that require a "potential file".

Running eon on the example akmc-fe-lammps (provided with the software), gives the following output. Note that there is no information about the reason for failure:
State list path does not exist; Creating: .//states/
Superbasin path does not exist, creating .//superbasins/
Registering results
Processed 0 results
Queue contains 0 searches
Making 1 process searches
job failed: .//jobs/scratch/0_0: b''
Created 1 searches
Currently in state 0 with confidence 0.000000

On the contrary, running the test included with the software (/eon/client/tests/lammps/) works fine. Here are the first lines of the file client.log:
[Matter] Iter Step size ||Force|| Energy
[Matter] 0 0.00000e+00 3.28013e+00 -1773.71530
[Matter] 1 2.00000e-01 3.50609e+00 -1774.27751
[Matter] 2 2.00000e-01 2.83334e+00 -1774.80119
[Matter] 3 2.00000e-01 2.28152e+00 -1775.28080
[Matter] 4 2.00000e-01 2.36860e+00 -1775.54974

The problem seems to happen when the file in.lammps points towards a potential file. For instance, the following in.lammps does not work:
pair_style eam/alloy
pair_coeff * * Fe.eam.alloy Fe

while the following works well:
pair_style morse 9.5
pair_coeff * * 0.7102 1.6047 2.897
pair_modify shift yes

Any ideas about what could be wrong?

Regards,
Sebastián



Some info about the system:
Ubuntu 20.04
python3.5 (Same issue found with 3.7)
Eon version svn revision 2466
lammps stable_5Nov2016 (same issue found with stable_7Aug2019 and latest version)

[graeme]

Hi Sebastian,

I think I would need to see the Fe.eam.alloy file, or better a .tar.gz file of the entire calculation, to debug this.

But do check to see if this work in lammps directly. This reading of the lammps input file is done entirely in the lammps code, not in the eon code, and so I expect there is some problem related to reading the Fe.eam.alloy potential file in lammps.

Graeme

[sebastianecheverrir]

Dear Graeme,

Thanks for your prompt reply and for pointing me in the right direction. It was indeed a problem with my installation of lammps. During compilation I did not activate the flag make yes-manybody, so the pair style was not recognised.

Maybe a good idea to include this additional step in the documentation.

Regards,
Sebastián

[graeme]

Great to hear that the problem was resolved. We do indeed need to add quite a bit of documentation.