Veröffentlicht 2. August 2023 | Version v1
Dataset Offen

Machine learned Force-fields for an ab-initio Quality Description of Metal-Organic Frameworks

  • 1. ROR icon Institute of Solid State Physics

Beschreibung

 

Supplementary data for "Machine learned  Force-Fields for an ab-initio Quality Description of Metal-Organic Frameworks" as published in npj computational materials.

 

For detailed descriptions and clearer results, please consult the main paper and its Supplementary Material.

 

The structure is as follows:

 

- There is one compressed .tar.gz file for each material

- these contain the general structure:

  - DFT_relax                         (energy optimizations of the structures using DFT)

  - DFT_elastic                       (computation of the elastic stiffness tensor using DFT)

  - DFT_phonon_band_structures        (evaluation of the phonon bands using DFT)

  - VASP_validation                   (creation of the validation set using DFT)

  - VASP_learning                     (active learning runs carried out using VASP to create VASP MLP. Also contains all refitting procedures. The VASP ML_FF files can be found in the subdirectories)

  - VASP_MLP_evaluation               (all evaluated properties using the VASP MLP)

  - MTP_evaluation                    (all evaluated properties using the MTP)

 

For the evaluation of the machine learned potentials there should be the following subdirectories:  

- train                        (for MTPs, training of the potential)

- band                         (for VASP MLPs, phonon band structures)

- elastic                      (for VASP MLPs, elastic tensor elements)

- band_and_elastic             (for MTPs, elastic tensor elements and phonon band structures)

- validation_set_comparison    (for both, comparison of force/stress/energies)

- NPT                          (for both if carried out, contains evaluation of thermal expansion)

- NPT_speedtest                (for both if carried out, speed tests on large supercells)

- NEMD                         (for MTPs if carried out, thermal conductivity calculation using the non-equilibrium molecular dynamics approach)

- AEMD                         (for MTPs if carried out, thermal conductivity calculation using the approach to equilibrium molecular dynamics approach)

 

In general, the input files used for the evaluation are provided

(INCAR and POSCAR for VASP, .in and .lmp files for lammps, .conf files for phonopy)

and the relevant output files should be present. However, some of the output was omitted (e.g. molecular dynamics trajectories) due to the sheer size. If you are interested in the full raw data, please contact the authors.

 

Additionally, the scripts directory contains some evaluation script used to evaluate certain properties like the elastic tensor elements or phonon band structures or the evaluation of thermal conductivity for the MTPs. Additionally, some plotting scripts for easy visualization are provided. Most of the useful scripts can be found in the ML_FF_aux subdirectory. Please read the help information for each script for details. If it is of interest to the reader to use these scripts, please contact Sandro Wieser for further instructions if they are difficult to understand.

Dateien

Dateien (37.1 GB)

Name Größe Download all
md5:4304c39b5343706a30ecdb14d6c9f148
6.8 GB Herunterladen
md5:5bc138e85455cb00dda3eded932e456e
4.6 GB Herunterladen
md5:15c24738051921d1fe474fde3fa7e5f9
4.3 GB Herunterladen
md5:20c202e107e4f842b591a69248632dd0
17.6 GB Herunterladen
md5:3b1ee26d06c97ec6b0c7ed27ba317d5d
3.0 kB Herunterladen
md5:14d1b8ba5a5194101a579695fdbd7eee
69.8 kB Herunterladen
md5:4b859edcbe072c9c5cc269ebca71d8d4
3.8 GB Herunterladen

Weitere Details

Verwandte Arbeiten

Is supplement to
Journal article: 10.1038/s41524-024-01205-w (DOI)