Published April 14, 2026 | Version v1

EAC-Net Benchmark Dataset: Charge Density Files and VASP Inputs for Training and Evaluation

  • 1. ROR icon University of Science and Technology of China
  • 2. ROR icon Ningbo Institute of Industrial Technology

Description

This dataset accompanies the manuscript "EAC-Net: Predicting real-space charge density via equivariant atomic contributions" (arXiv:2508.04052). It contains DFT-computed charge density files in HDF5 format for four benchmark systems (amorphous Si, Al surface, Al-Mg alloy, and liquid water), with explicit train/validation/test splits corresponding to Figure 2 of the manuscript. Also included are VASP input files for the high-entropy alloy zero-shot evaluation in Figure 3e–f. A detailed description of the data format, file organization, and train/test splits is provided in README.md.

Files

readme.md

Files (6.0 GB)

Name Size
md5:7894f7bfffe2398227fd7af15341e2c1
1.5 GB Preview Download
md5:162bf9e3b0432f9030e91f6b78fce745
1.5 GB Preview Download
md5:b795877e583fa639937eacf457dcc5b4
1.5 GB Preview Download
md5:290034a2d34aedfbd843490544d52072
1.5 GB Preview Download
md5:86881747615bfacd355bb5ad4fc16c75
40.0 kB Preview Download
md5:1b62b8a16a0f8a7013893c95f679d26a
9.1 kB Preview Download