Need support for newton off requirement when using pair_style nequip

[ChengQIAN93]

Dear Calphy developers, thank you for making such a strong and easy-to-use tool.

I recently had a small problem when using pair_style nequip in LAMMPS through Calphy. The NequIP pair style requires "Newton off," but LAMMPS by default uses "Newton on." I'm not sure where in Calphy's workflow this setting should be changed to automatically require the Newton flag when nequip potentials are selected. What's the best way to handle this requirement?

[srmnitc]

@ChengQIAN93 thanks for the issue. You can use this option: https://calphy.org/en/latest/inputfile.html#init-commands

[ChengQIAN93]

Thanks for your help, after adding the init-command keyword calphy works fine. But I have two more questions and would like to get your guidance.

1. when using a LAMMPS data file specified through lattice: alpha.data in input.yaml, the repeat parameter appears to be completely ignored. Despite setting values like repeat: [2, 2, 2], the simulation box maintains the original dimensions from the data file without any replication.

2. when using script_mode: True to generate SLURM scripts for efficient GPU distribution, running the submitted job fails with an AttributeError: 'Calculation' object has no attribute '_ghost_element_count'. The traceback points to line 302 in solid.py during the run_minimal_averaging process.

here is the input.yaml I used.

here is the script generated by the Calphy

here is the error when I run the script

[srmnitc]

Thanks again!

• If a datafile is provided, the repeat parameter is ignored. This is the expected behaviour.
• I fixed the issue in #177 , updating should help.

[ChengQIAN93]

Thank you for the update on Calphy. The script mode is working very well, and LAMMPS can run on separate GPU cards. However, I didn't find the temperature_sweep.dat file in any of the folders. Do I need to run additional commands? Thank you in advance, and I look forward to your reply.

This is what Calyphy export in the folder.

[srmnitc]

@ChengQIAN93 Unfortunately the script mode is not yet available for temperature sweep. One way twould be to do calculations for an array of temperatures one by one. Further expansion of the scipt mode is planned, but would need about 2 months or so.

[ChengQIAN93]

@srmnitc Thanks for the reply, I'll keep an eye on Calphy for updates!