D-Tensor and AIMD Data for Three Metal Complexes via Machine Learning

Reference data for 3500 configurations of three metal complexes (CoSar, FeTPAPh, NiComplex) generated using the Molpro package. The dataset includes D-tensors (magnetic anisotropy) in cm⁻¹ and, for CoSar, a separate AIMD dataset with total energies and atomic forces from VASP simulations. The data was produced by Viktor Zaverkin at the University of Stuttgart and is associated with a 2021 publication.

Use Cases

• Training machine learning models to predict magnetic anisotropy tensors (D-tensors) based on atomic coordinates and nuclear charges.
• Developing surrogate models for ab initio molecular dynamics (AIMD) to predict total energies and atomic forces in periodic systems.
• Benchmarking computational methods for calculating thermally averaged magnetic properties of transition metal complexes.

Strengths

• Contains 3500 configurations for each of three distinct metal-organic complexes.
• Includes both static D-tensor data and AIMD data with energies and forces for one complex.
• Data is generated using established quantum chemistry (Molpro) and molecular dynamics (VASP) software packages.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown for the overall dataset, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Stuttgart

Collection Method

Generated using the Molpro quantum chemistry package for D-tensors and the VASP package for ab initio molecular dynamics.