Methodology1 min read

Where DFT Codes Disagree: A Cross-Database Energy Map

Cross-source DFT disagreement analysis across 0 compositions. 0 show high disagreement where independent DFT databases predict substantially different thermodynamic stability.

LG
Lattice Graph

Where DFT Codes Disagree: A Cross-Database Energy Map

Overview

Lattice Graph cross-references thermodynamic predictions from up to 9 independent density functional theory databases — Materials Project, OQMD, AFLOW, JARVIS, GNoME, Alexandria, NOMAD, MC3D, and OMat24. Of the 0 compositions with multi-source coverage, 0 (0%) show high disagreement (composite score ≥ 0.6).

Disagreement is quantified as a weighted composite of three signals:

  • Energy-above-hull spread (50% weight): max(EAH) − min(EAH) across sources, normalized to 0.5 eV/atom
  • Band-gap spread (30% weight): max(BG) − min(BG) in meV, normalized to 3000 meV
  • Spacegroup entropy (20% weight): Shannon entropy across reported crystal structures — high entropy means different codes predict different symmetries

Methodology

The disagreement score is computed in the dbt mart model mart_dft_disagreement_map. For each composition with data in 2+ DFT sources, we select the lowest-EAH structure per source, then compute the spread across sources. Spacegroup entropy uses Shannon entropy over the modal spacegroup reported by each source. The composite score weights these three signals at 50/30/20.

Trust tiers: high = 3+ sources with spread < 25 meV/atom; medium = 2+ sources; low = single source.

Apply this methodology to your materials

Lattice Graph automates computational screening so you can focus on what matters — the lab.

Explore the Platform