gML-sTDA: Excited-State Electronic Structure Data for Nanotubes and Quantum Dots

A dataset from figshare containing results from generative machine-learning-accelerated excited-state calculations for large systems. The data, authored by Xiang-Yang Liu and updated in April 2026, includes benchmarks on carbon nanotube models up to 1114 atoms and applications to silicon quantum dots, black phosphorus, and rubrene clusters. The dataset is 326.5 KB in size and consists of XYZ files.

Use Cases

• Benchmarking machine-learning acceleration methods for excited-state calculations based on the described gML-sTDA protocol.
• Analyzing excitation energies and absorption spectra for carbon nanotubes based on the reported benchmarks up to 1114 atoms.
• Evaluating computational speedup factors for large-scale systems like silicon quantum dots and black phosphorus supercells.
• Training or validating generative models like Restricted Boltzmann Machines for configuration space sampling in quantum chemistry.

Strengths

• Benchmarks include systems up to 1114 atoms with 16,424 basis functions, indicating applicability to large-scale models.
• Reports specific performance metrics, including mean absolute errors of ~0.007–0.011 eV and speedup factors up to ~40x.
• Includes applications to multiple material systems (carbon nanotubes, silicon quantum dots, black phosphorus, rubrene).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 326.5 KB, indicating a limited scope of included data.

Provenance

Source

figshare

Collection Method

Results from computational quantum chemistry simulations using the gML-sTDA method.

Time Range

Benchmarks and applications date from the research publication period, likely circa 2026.

Freshness

Last updated 2026-04-11 13:10:15; freshness should be verified.

Geography

null