SPARC AL Package: Inputs and Models for Molecular Dynamics Workflows

2.0 GB of input files, training data, and analysis scripts for the SPARC (Smart Potential with Atomistic Rare Events and Continuous Learning) workflow. The dataset covers two benchmark systems: 1,3-Butadiene conformational isomerization and ketohydroperoxide unimolecular decomposition. It was authored by Rahul Verma and last updated on May 23, 2026.

Use Cases

• Benchmarking active learning sampling strategies like plain AL-ML/MD, moving restraint, and umbrella sampling based on the described 1,3-Butadiene case.
• Training machine-learned interatomic potentials using the included DeepMD-kit inputs and DFT-labelled configurations.
• Reproducing research figures from the associated paper using the provided analysis notebook.
• Studying rare event dynamics in chemical systems like ketohydroperoxide decomposition using the supplied MD outputs and trained models.

Strengths

• Includes a complete workflow with configuration files, training inputs, labelled data, trained models, and outputs for two distinct chemical systems.
• Provides analysis scripts to reproduce paper figures, supporting reproducibility.
• Released under a permissive CC-BY-4.0 license, allowing for broad reuse and modification.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• The 2.0 GB size suggests a medium-scale dataset, but the exact number of molecular configurations is not specified.

Provenance

Source

Rahul Verma via figshare

Collection Method

Generated as part of the SPARC workflow for computational chemistry simulations.

Freshness

Last updated 2026-05-23 19:13:58; freshness should be verified.