Trajectories for ordered and disordered crystalline NiAl

This record contains files connected to an example in the documentation of the dynasor package [1] illustrating the analysis of the dynamic properties of ordered and disordered samples of NiAl. The molecular dynamics (MD) trajectory files included here have been generated using the GPUMD package [2,3]. Please consult the GPUMD documentation for information on how to compile and run GPUMD and the run.in file. The input structures (GPUMD input file model-*.xyz) were generated using the prepare-structures.py script.

The runs use the general-purpose machine-learned potential based on the neuroevolution potential approach presented in [4]. The parameter file is available on Zenodo at https://zenodo.org/records/10081677/files/UNEP-v1-model-without-ZBL.txt

Please note that the runs employ the optional NetCDF extension of GPUMD. It is possible to use the standard extended-xyz output format (via the `dump_exyz` keyword) but both the execution time and the trajectory files will be multiple times longer. When using the NetCDF extension the runs should take about 17 to 19 minutes on an A100 or RTX3090 GPU.

[1] Computer Physics Communications 316, 109759 (2025)

[2] Journal of Chemical Physics 157, 114801 (2022)

[3] Materials Genome Engineering Advances, Early View, e70028 (2025)

[4] Nature Communications 15, 10208 (2024)