July 8, 2021 Dataset Open Access

LDOS/SNAP data for MALA: Aluminium at 298K and 933K

Fiedler, Lenz; Modine, Normand A.; Thompson, Aidan P.; Cangi, Attila; Rajamanickam, Siva

# Aluminium data set for Machine Learning applications

This dataset contains DFT inputs, outputs, LDOS data and bispectrum descriptor vectors for an aluminium cell of 256 atoms at varying temperatures and ambient mass density. All simulations that include the LDOS have been performed at an LDOS converged k-grid of 8x8x8 k-points. Calculations which do not inlcude the LDOS have been performed at a total free energy converged k-grid of 4x4x4 k-points (i.e., the total free energy has been converged to 1 meV/atom accuracy).

For each temperature, a .zip file is included in this data set. All .zip files are structured in the same way. For the two largest temperatures, the zip files have been split into smaller portions for easier download; please note that you still have to download all parts of the zip file locally and re-assemble it via the zip command line utility.

Temperature here primarily refers to electronic temperature. However, in almost all cases, the ionic temperature is the same as the electronic temperature. The few cases where this does not hold true are detailed in the individual .zip files by a "different_ionic_temperatures.md" file. If no such file is present in .zip file, then all calculations have been performed at matching electronic and ionic temperatures.

## Authors:

- Fiedler, Lenz (HZDR / CASUS)
- Cangi, Attila (HZDR / CASUS)
- Modine, Normand A. (SNL)
- Thompson, Aidan P. (SNL)
- Rajamanickam, Siva (SNL)

Affiliations:

HZDR - Helmholtz-Zentrum Dresden-Rossendorf

CASUS - Center for Advanced Systems Understanding

SNL - Sandial National Laboratories

## Dataset description

- Total size: 1.1 TB
- System: Al256
- Temperature(s): 100K, 200K, 298K, 400K, 500K, 600K, 700K, 800K, 933K
- Mass density(ies): 2.699 gcc
- Crystal Structure: fcc (material mp-134 in the materials project)
- Number of atomic snapshots: 105
    - 30 (100K): 138 GB
    - 3  (200K): 42 GB
    - 10 (298K): 137 GB
    - 3  (400K): 41 GB
    - 20 (500K): 237 GB (zip file split in three portions)
    - 3  (600K): 42 GB
    - 3  (700K): 42 GB
    - 3  (800K): 42 GB
    - 30 (933K): 360 GB (zip file split in four portions)
- Contents:
    - ideal crystal structure: no
    - MD trajectory: no
    - Atomic positions: yno
    - DFT inputs: yes
    - DFT outputs (energies): yes
    - SNAP vectors: yes (partially, see below)
        - dimensions: 200x200x200x94 (last dimension: first three entries are x,y,z coordinates, data size is 91)
        - units: a.u.
    - LDOS vectors: yes (partially, see below)
        - dimensions: 200x200x200x250
        - units: 1/(Ry*Bohr^3)
        - note: LDOS parameters are the same for all sizes of the unit cell
    - trained networks: no

## Dataset structure

For each temperature, a .zip file is included which contains one folder per combination of mass density and number of atoms (only one folder in case of this dataset). Therein, one finds the following folders:

- ldos: holds the LDOS vectors
- bispectrum: holds the SNAP fingerprint vectors
- dft_outputs: holds the outputs from the DFT calculations, i.e. energies in the form of a QE output file
- dft_inputs: holds the inputs for the DFT calculations, in the form of a QE input file
- different_ionic_temperatures.md: If necessary, details which snapshots have an ionic temperature different from the given electronic temperature