Goldilocks Data
Setting up a DFT calculation means choosing numerical inputs that are hard to choose well. A k-point mesh that is too coarse gives the wrong energy; one that is too dense wastes compute.
Goldilocks learns those choices from systematic convergence calculations. This site is where those data are made and documented.
Just want recommended inputs?
Then you want Goldilocks Core. Give it a structure and a calculation intent; it selects suitable inputs.
Read on if you want to rerun these calculations, inspect how convergence was labelled, or use the published records.
What this repository does
- Generate — submit parameter sweeps through AiiDA with stable structure identifiers and calculation provenance.
- Analyse — apply explicit convergence criteria and find the smallest acceptable input for each structure.
- Publish — deposit documented snapshots as citable published records for research and model training.
Model training belongs in Goldilocks ML. End-user input generation belongs in Goldilocks Core.
What has been published
Three records, from two sets of calculations, all on MC3D structures with Quantum ESPRESSO:
| Calculations | Records |
|---|---|
| No-spin SCF k-point sweeps | the converged mesh per structure, and the full per-calculation output behind the paper |
| nscf band structures | eigenvalues along the high-symmetry path, with Fermi level, metallicity and band gap |
Each record says what its columns mean and which conventions it froze; each page under Calculations gives the settings, a script that runs the same calculations yourself, and what the published data can answer without running anything.
Where to go
Published records Install goldilocks-data How the data was made