Franklalalala/EMolStudio-review-bundle

EMolStudio review bundle

Software and test materials for A Density-Matrix Framework for Electronic-Structure Analysis of Electrolytes for Lithium Batteries.

EMolAgent demo

Edited demonstration · 1 min 08 s

Open the edited video · Original English recording, unedited · 7 min 26 s

Both videos are public and require no account. They are also included in Supplementary Software 1 under EMolStudio/videos/ for offline viewing.

Download and run

Download Supplementary Software 1, extract it, and run from EMolStudio/ with Python 3.10+:

python benchmarks/reproduce_tables.py --out outputs/table_check

The script recalculates archived results and checks all 135 displayed table values, including 5,500 dipole-vector pairs. See the reproducibility guide for expected output and EMolES inference.

Reduced multi-Li⁺ EC clusters

example/multi_li_ec/ in the archive reproduces Supplementary Note 3 and Fig. 4: EMolConf builds the Li⁺–EC clusters, UMA relaxes them, samples the reduced state and gives the vacuum and harmonic free-energy terms, and EMolES provides the spin densities and the solvation energies of the reduction potentials. With the GPU runtime of docs/INSTALL.md, the supplied structures give Supplementary Table 4:

python example/multi_li_ec/emoles_postprocess.py potentials --out outputs/multi_li_ec

Included assets

Asset	Contents
Supplementary software archive	Source, reference runtime, example checkpoint/geometries, multi-Li⁺ cluster workflow, evaluation records and scripts
molecule_test.tar.gz	500 held-out molecular structures
cluster_test.tar.gz	600 held-out first-shell clusters
qh9_test.tar.gz	9,335 QH9 test structures
checkpoints.tar.gz	EMolES checkpoints for the three test sets

Table conventions and the current metric update are summarized in numerical notes. Validation results are in VALIDATION.md. Downloads are public and require no account. The full application libraries and training datasets are separate from this deposit.

EMolAgent is maintained by Huize Yu in its public GitHub repository.

Attribution

QH9 is the quantum-Hamiltonian benchmark of Haiyang Yu and colleagues, NeurIPS Datasets and Benchmarks 2023. Cite QH9 and the accompanying EMolStudio manuscript when using these materials. EMolES and OpenEquivariance retain their upstream LGPL-3.0 and BSD-3-Clause terms; see THIRD_PARTY.md in the archive. Multiwfn is obtained separately from its authors. Public access does not assign a new licence or archival DOI to the authors' code and data.