Materials Science · LeMaterial
LeMat-Bulk-DFT-Hull
Convex hull reference data for materials stability analysis across DFT and machine learning interatomic potentials.
What LeMat-Bulk-DFT-Hull is
LeMat-Bulk-DFT-Hull is a materials science dataset published by LeMaterial. It compiles convex hull reference energies used to assess the thermodynamic stability of bulk materials across chemistry, physics, and materials engineering.
Evaluating whether a material is thermodynamically stable or metastable relies on calculating its energy distance to the convex hull. This dataset facilitates stability benchmarking and screening by providing baseline energy metrics computed using density functional theory (DFT) alongside multiple machine learning interatomic potential (MLIP) models.
How it was made
The data comprises bulk materials evaluated against convex hulls derived from DFT and machine learning interatomic potentials, specifically ORB, UMA, MACE-MP, and MACE-OMAT. The collection provides an unfiltered split containing all materials with hull energies across every energy type, as well as specialised splits filtered to structures lying within 0.001 eV/atom of the respective hull.
Data records are organised in Parquet files alongside compressed NumPy files (.npz). Tabular datasets provide hull energies, while companion NPZ files store sparse elemental composition matrices and associated metadata such as element mappings.
Key facts
- Category
- Materials Science
- File formats
- Parquet, NPZ
- Data modalities
- Tabular, text
- Energy types
- DFT, ORB, UMA, MACE-MP, MACE-OMAT
- Filtering threshold
- 0.001 eV/atom above hull
What you can use it for
- Thermodynamic Stability Screening
- Identify bulk compounds that reside on or near the convex hull using precalculated DFT and MLIP energy baselines.
- Machine Learning Potential Benchmarking
- Compare convex hull stability predictions from ORB, UMA, MACE-MP, and MACE-OMAT models against reference DFT calculations.
- Compositional Analysis of Stable Phases
- Examine chemical spaces of stable materials using the provided sparse composition matrices and element mappings.
Explore the data
No licence declared
The authors have not published a licence, which means no permission to redistribute the data has been granted to anyone. The rows are not served here; read them on the original dataset page.How the columns are distributed
Over 144,127 rows of default/dft
e_above_hull
float- min
- -0
- median
- 0
- mean
- 1.1e-4
- max
- 1.0e-3
energy
float- min
- -2260.06
- median
- -60.87
- mean
- -89.94
- max
- -0.03
formation_energy
float- min
- -904.18
- median
- -8.10
- mean
- -18.28
- max
- 0.03
formation_energy_per_atom
float- min
- -37.93
- median
- -0.77
- mean
- -1.00
- max
- 1.0e-3
immutable_id
string_text- min
- 5
- median
- 12
- mean
- 11.57
- max
- 13
mace_mp_energy
float- min
- -2252.80
- median
- -60.38
- mean
- -89.21
- max
- 0.08
mace_omat_energy
float- min
- -2252.80
- median
- -60.38
- mean
- -89.21
- max
- 0.08
nsites
int- min
- 1
- median
- 11
- mean
- 15.39
- max
- 320
orb_conserv_inf_energy
floatNo distribution reported for this column.
100.0% empty
orb_direct_20_energy
floatNo distribution reported for this column.
100.0% empty
species_at_sites
list- min
- 1
- median
- 11
- mean
- 15.39
- max
- 320
true_energy
float- min
- -2260.06
- median
- -60.87
- mean
- -89.94
- max
- -0.03
uma_energy
float- min
- -2254.99
- median
- -60.44
- mean
- -89.33
- max
- 0.07
Strengths and limits
Good at
- Provides aligned hull energy evaluations across DFT and four distinct machine learning interatomic potential frameworks.
- Offers pre-filtered subsets isolating structures within 0.001 eV/atom of thermodynamic convex hulls.
- Includes sparse composition matrices alongside tabular data for streamlined chemical composition analysis.
Watch out for
- Pre-filtered subset splits are limited to compounds strictly within the 0.001 eV/atom threshold.
- Machine learning potential coverage is restricted to ORB, UMA, MACE-MP, and MACE-OMAT model outputs.
Questions
Can I look inside LeMat-Bulk-DFT-Hull without downloading it?
Not here: it declares no licence, so no permission to redistribute its rows has been granted. Its description and statistics are still shown.
What licence is LeMat-Bulk-DFT-Hull under?
Its authors have not declared a licence, which means no permission to reuse or redistribute it has been granted.
Which energy calculation methods are included?
The dataset includes DFT reference energies alongside machine learning interatomic potential energies from ORB, UMA, MACE-MP, and MACE-OMAT.
What criteria define the filtered dataset splits?
The filtered splits for dft, orb, uma, mace_mp, and mace_omat contain materials falling within 0.001 eV/atom of their respective convex hulls.
What file types are available in the repository?
Tabular stability metrics are provided as Parquet files, while composition matrices and metadata element mappings are stored in sparse NPZ files.
Written from LeMat-Bulk-DFT-Hull's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.