
Data repository for the ChIMES_D2 interatomic potential. The following files and folders are provided: The final parameter file for the ChIMES_D2 interatomic potential. This can be used in either the chimes_calculator, chimes_md, or LAMMPS+ChIMES_FF codes to help run molecular simulations of deuterium ("params.penalty.txt.reduced"). A table of equation of state values (density, temperature, pressure, total energy) calculated from NVT simulations of deuterium using ChIMES_D2 ("ChIMES_D2_EOS_Table.txt"). Input files needed to run the ChIMES Active Learning Driver (ALD) code to carry out the active learning procedure described in the manuscript. Please note the file paths, number of requested nodes/cores, and loaded modules specified in the config.py file will need to be modified for your specific HPC machine/ALD installation. This also contains the initial training data in .xyzf format for the five state conditions used to train the model. Other important files needed to run the ChIMES fitting are included here such as fm_setup.in ("ActiveLearningDriver.zip"). LAMMPS input files and atomic positions for running NVT and MSST calculations with the ChIMES_D2 potential as described in the manuscript. LAMMPS must be compiled with ChIMES_FF/pair chimes included to run simulations with ChIMES potentials ("LAMMPS.zip"). The results from MSST simulations with ChIMES_D2 for a deuterium system at an initial density of 0.173 g/cc and temperature 19 K for a wide range of shock velocities ("MSST_Results.zip"). A folder with a Mathematica notebook for analyzing the Hugoniot curves predicted by ChIMES_D2 and comparing them to ab initio and experimental predictions ("Hugoniot.zip").
We also include some additional ChIMES potentials ("EXTRA_Potentials.zip") WARNING: UNLIKE THE MAIN ChIMES_D2 POTENTIAL, NONE OF THESE EXTRA MODELS HAVE BEEN THOROUGHLY TESTED! USE AT YOUR OWN RISK. 3B_params.penalty.txt.reduced: this potential was trained on the same data as ChIMES_D2, but does not include four body interactions (Chebyshev order 12/7/0 instead of 12/7/4). This should help decrease the computational cost at the price of reduced accuracy compared with the four body model. FSE_params.penalty.txt.reduced: The training set of this model contains the same configurations as ChIMES_D2, but also includes stresses and energies in the fit. Weight factors of 25 and 0.02 were used for stresses and energies, respectively. All other hyperparameters are identical to ChIMES_D2. This potential is expected to give almost identical results, except the zero point of the energy is closer to the DFT-MD zero point (which is not physically meaningful, but may just be nice for comparison to DFT-MD). LARGE_params.penalty.txt.reduced: All hyperparameters are identical to ChIMES_D2, except this model was fit using an expanded training set, including higher density data along the 0.837338 g/cc isochore up to 15,625 K. It is expected to give better results than ChIMES_D2 in this density range, although the perfomance in the low density range (0.2-0.3 g/cc) is slightly reduced. The Hugoniot prediction is almost identical to ChIMES_D2.
| citations This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | 0 | |
| popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network. | Average | |
| influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | Average | |
| impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network. | Average |
