Is there any suite of tests for LAMMPS other than the items in the examples and bench directories? For example, are there any tests used internally by the developers before releasing a new version of the code? I’m interested in something that’s automated, or possible to automate, to run a quick validation check on new or updated LAMMPS installations on medium to large clusters.
Failing that, is picking an arbitrary allowable margin of error for each example/benchmark and using a script to grep the output of the examples/benchmarks versus their reference results the only solution? My background is in HPC rather than thorough domain knowledge of LAMMPS and what margin of variation in the example/benchmark results is reasonable.