Data for: A Computationally Efficient Quasi-Harmonic Study of Ice Polymorphs Using the FFLUX Force Field

This repository provides additional data to accompany the paper:

"A Computationally Efficient Quasi-Harmonic Study of Ice Polymorphs Using the FFLUX Force Field"
A. Pák, M. L. Brown and P. L. A. Popelier
Acta Crystallographica A (2025)
DOI: https://doi.org/10.1107/S2053273324010921.

In this article the machine learning force field FFLUX is applied to ice polymorphs in geometry optimisations, calculation of phonon spectra, and free energies. In addition to Helmholtz free energies, Gibbs free energies were calculated for the first time using FFLUX under the quasi-harmonic approximation. Data from these calculations is available in this repository, including:

• The Gaussian process regression machine learning model used in the FFLUX calculations;
• Files used to test the electrostatic energy prediction of the model;
• Files used to generate the optimised structures;
• Files used to calculate the phonon density of states, phonon dispersions and Helmholtz free energies;
• Files used to calculate the Gibbs free energies, with data at each compressed and expanded volume.

Details are provided for how the data was generated in the published paper and the supporting information.

Input files for the Vienna Ab Initio Simulation Package (VASP) code are also included in the repository. Although FFLUX has not yet been made publicly available, the water Gaussian process regression models and input files are given for when it is made available.

本仓库提供额外数据以辅助伴随论文《基于FFLUX力场对冰多形体的计算高效近谐振研究》（A Computationally Efficient Quasi-Harmonic Study of Ice Polymorphs Using the FFLUX Force Field）的使用，该论文由A. Pák、M. L. Brown和P. L. A. Popelier所著，发表于Acta Crystallographica A（2025）期，DOI为https://doi.org/10.1107/S2053273324010921。在本文中，机器学习力场FFLUX被应用于冰多形体的几何优化、声子谱计算以及自由能计算。此外，首次在近谐振近似下，利用FFLUX计算了吉布斯自由能。本仓库中提供了这些计算的数据，包括：

• 在FFLUX计算中使用的Gaussian过程回归机器学习模型；
• 用于测试模型静电能量预测的文件；
• 用于生成优化结构的文件；
• 用于计算声子态密度、声子色散和亥姆霍兹自由能的文件；
• 用于计算吉布斯自由能的文件，包含每个压缩和膨胀体积的数据。

关于数据的生成细节已在发表的文章及其补充信息中提供。

此外，仓库还包含了用于维也纳从头算模拟包（Vienna Ab Initio Simulation Package，简称VASP）代码的输入文件。尽管FFLUX尚未公开发布，但提供了水Gaussian过程回归模型和输入文件，以便在FFLUX发布时使用。