SI of paper "Thermodynamics of Water and Ice from a Fast and Scalable First-Principles Neuroevolution Potential" .
This repository is structured as follow:
-
nep-pre-train-model:
- nep.txt: NEP potential for water/ice systems, as descriped in the manscript
- nep.in: Input file used to develop the NEP potential
-
nep-fitting:
- data.xyz: Data set used for potential-fitting, in extended xyz format
- nep.in: Input file used to develop the NEP potential
-
notebook-examples:
- example-1.ipynb: Fit a NEP potential and make single-point predictions with pre-train model.
- example-2.ipynb: Perform simulations for runtime profiling, computing RDFs and vDOS.
- example-3.ipynb: Perform simulations to compute orthobaric densities for ice ih.
- example-4.ipynb: Perform simulations to compute isothermal compressibility.
- example-5.ipynb: Perform simulations to compute enthalpy of melting.
-
water-pimd-input-files:
- input3.xml: Parameter file for the PIMD run.
- run.in: LAMMPS-PIMD input file.
- compile_lammps_nep.sh: Bash script to compile NEP potential with LAMMPS.
-
D2O-simulations: Essential input files to reproduce calcualtions shown in figure 5 to 8.
-
Disclaimers:
- More details about the data set can be found here and here.
- The colab notebook examples are compactable with GPUMD and TDMDpy.
- Parameters for simulations and potential fittings used in these examples are modified so that each example can be executed in a reasonable amount of time, please refer to the publications for all relevant parameters to reproduce the results.