Coarse-grained methanol trajectory

This dataset is a supplement to the paper "Thermodynamic Transferability in Coarse-Grained Force Fields using Graph Neural Networks," available at https://arxiv.org/abs/2406.12112. 

To create this dataset, an all-atom trajectory of liquid methanol was generated using the GROMOS 54A7 force field and the LAMMPS software package.  The coarse-grained mapping described in the aforementioned paper was applied to the trajectory; the resulting positions, velocities, and forces are provided here. Additionally, the periodic cell is reported, as well as values of the radial distribution function calculated using the coarse-grained positions. For completeness, mass and species arrays are also included.

The original all-atom trajectory was generated using a Nosé-Hoover thermostat at 700 K. After equilibration, 50,000 timesteps of 1 fs were computed and every 50th frame recorded. The resulting 1,000 frames were used to generate this dataset.

The data is stored in a single Numpy .npz file, which contains eight arrays: 

key
shape
size (bytes)

cells
(100, 3, 3)
7200

forces
(100, 1024, 3)
2457600

masses
(100, 1024)
819200

positions
(100, 1024, 3)
2457600

rdf_bins
(300,)
2400

rdf_values
(300,)
2400

species
(100, 1024)
819200

velocities
(100, 1024, 3)
2457600

Total size of file: 9.03 MB