3.1. Command-line options

At run time, FitSNAP recognizes command-line options which may be used in any order. Either the full word or a one or two letter abbreviation can be used:

For example, the FitSNAP executable might be launched as follows:

mpirun -np 4 python -m fitsnap3 input.in --overwrite

--keyword GROUP NAME VALUE

Replace or add input keyword group GROUP, key NAME, with value VALUE. Type carefully; a misspelled key name or value may be silently ignored.


--lammpslog

Writes a LAMMPS log file for descriptor calculations. Since LAMMPS log files get overwritten with configuration that we calculate descriptors for, this will give the last log file that occured. This is mainly useful if you want to see the LAMMPS outputs and settings that were used to calculate descriptors.


--log FILE

Write outputs and warnings from Python’s native logging facility to a file named FILE. We currently do not output any information with the logging facility yet, so the capability mainly exists for developers and debugging for now.


--nofit

This will run FitSNAP without performing a fit. Only descriptors will be calculated. This is useful when paired with dump_dataframe = 1 in the [EXTRAS] section


--nscreen

Print outputs from the Output class for each node.


--overwrite

Overwrite output files. Otherwise, output files will not be overwritten and FitSNAP will error. This protects existing fits from being overwritten.


--pscreen

Print outputs from the Output class for each processor.


--relative

Put output files in the directory of the FitSNAP input file. For example if the FitSNAP input file is located at foo/bar/input.in, then running FitSNAP like

mpirun -np 4 python -m fitsnap3 foo/bar/input.in --relative

will write output files in the directory foo/bar. Without the --relative option, the files will be written in the current directory.


--screen

Print outputs to screen.


--screen2file FILE

Print screen to a file named FILE.


--tarball

Package SNAP fit files into a hashed tarball named fit-<hash>.tar.gz. This is only useful when fitting with SNAP descriptors since multiple files are used by LAMMPS. The tarball contains a SNAP parameter file, a SNAP coefficient file, a *.mod file containing the proper pair style to use, and an example LAMMPS input file showing how to use this potential.

ACE descriptors, on the other hand, only use *.yace files in LAMMPS, therefore no need for a tarball.


--verbose

Show more detailed information about processing. Currently there are no verbose outputs, so this option exists purely for development and testing purposes.