Boron data set for machine learning applications

<strong>Boron data set for machine learning applications</strong>

This dataset contains DFT inputs, outputs, LDOS data and bispectrum descriptor vectors for an α-rhombohedral boron cell of 144 atoms at room temperature and ambient mass density. All simulations have been performed at an LDOS converged k-grid of 4x4x4 k-points.

This dataset contains one .zip file for each of its five type of data (bispectrum descriptors, LDOS, DFT inputs, DFT outputs and trained models).

<em>Authors:</em>

- Fiedler, Lenz (HZDR / CASUS)<br>
- Cangi, Attila (HZDR / CASUS)

Affiliations<em>:</em>

HZDR - Helmholtz-Zentrum Dresden-Rossendorf<br>
CASUS - Center for Advanced Systems Understanding

<em>Dataset description</em>

- Total size: 26 GB<br>
- System: B144<br>
- Temperature(s): 298K<br>
- Mass density(ies): 2.483 gcc<br>
- Crystal Structure: amorphous (material mp-160 in the materials project)<br>
- Number of atomic snapshots: 15<br>
- Contents:<br>
    - ideal crystal structure: no<br>
    - MD trajectory: no<br>
    - Atomic positions: no<br>
    - DFT inputs: yes<br>
    - DFT outputs (energies): yes<br>
    - SNAP vectors: yes<br>
        - dimensions: 108x108x35x94 (last dimension: first three entries are x,y,z coordinates, data size is 91)<br>
        - units: a.u.<br>
    - LDOS vectors: yes<br>
        - dimensions: 108x108x35x241<br>
        - units: 1/(eV*Angstrom^3)<br>
    - trained networks: yes

<br>
<em>Dataset structure</em>

A .zip file is included for each for each of its five type of data:

- ldos.zip: holds the LDOS vectors (one HDF5 file per snapshot)<br>
- bispectrum.zip: holds the bispectrum fingerprint vectors  (one HDF5 file per snapshot)<br>
- dft_outputs: holds the outputs from the DFT calculations, i.e. energies and simulation parameters in a .json format (one per snapshot)<br>
- dft_inputs: holds the inputs for the DFT calculations, in the form of a QE input file (one per snapshot)<br>
- models: holds five trained NN models for the data set

<strong>面向机器学习应用的硼数据集</strong>

本数据集包含室温及常压质量密度下144原子α-菱面体硼（α-rhombohedral boron）晶胞的密度泛函理论（DFT）输入、输出、局域态密度（LDOS）数据及双谱描述符（bispectrum descriptor）向量。所有模拟均在LDOS收敛的4×4×4 k点（k-points）网格（k-grid）上执行。

本数据集针对其五种数据类型（双谱描述符、LDOS、DFT输入、DFT输出及训练模型）各包含一个.zip文件。

<em>作者：</em>

- Fiedler, Lenz（HZDR / CASUS）<br>
- Cangi, Attila（HZDR / CASUS）

<em>数据集描述：</em>

- 总大小：26 GB<br>
- 系统：B144<br>
- 温度：298K<br>
- 质量密度：2.483 g/cm³<br>
- 晶体结构：无定形（材料项目中的mp-160材料）<br>
- 原子快照数量：15<br>
- 内容：<br>
    - 理想晶体结构：否<br>
    - 分子动力学（MD）轨迹：否<br>
    - 原子位置：否<br>
    - DFT输入：是<br>
    - DFT输出（能量）：是<br>
    - SNAP向量（SNAP vectors）：是<br>
        - 维度：108×108×35×94（最后一维：前三项为x、y、z坐标，数据大小为91）<br>
        - 单位：原子单位（a.u.）<br>
    - LDOS向量：是<br>
        - 维度：108×108×35×241<br>
        - 单位：1/(eV·Å³)<br>
    - 训练模型：是

<em>数据集结构：</em>

针对五种数据类型各包含一个.zip文件：<br>
- ldos.zip：存储LDOS向量（每个快照一个HDF5文件）<br>
- bispectrum.zip：存储双谱指纹向量（每个快照一个HDF5文件）<br>
- dft_outputs：存储DFT计算输出，即JSON格式的能量及模拟参数（每个快照一个）<br>
- dft_inputs：存储DFT计算输入，格式为量子ESPRESSO（QE）输入文件（每个快照一个）<br>
- models：存储针对本数据集的五个训练好的神经网络（NN）模型