Skip to content

QRF95

Predicts how far apart k-points may sit and still give a converged answer — a k-distance in Å⁻¹. Goldilocks Core turns that one number into a mesh using the crystal's reciprocal lattice: N_i = ceil(|b_i| / k_distance).
Predicts k_distance, in Å⁻¹
Record q3bye-wep37, v2.0
Status historical — v2.0 is the last version
Notebook run it yourself

Use it

Download the record and nothing else — the files it needs travel with it.

from goldilocks_ml.inference import load_model

model = load_model("path/to/the/record")
prediction = model.predict(structure)

prediction.value  # the k-distance, in Å⁻¹
prediction.details["interval"]  # a low and a high estimate

How good it is

From the paper it was fitted for: R² 0.703, MAE 0.067 Å⁻¹, and 95.8% coverage at a mean interval width of 0.313 Å⁻¹.

Those are the paper's numbers, not this record's. The forest was fitted before this repository existed, so nothing here can rescore it, and it appears in no results table on this site.

When to be careful

  • The interval claims no coverage. The published record carries no calibration, so treat the low and high values as a spread, not a guarantee. The recommendation itself is unaffected.
  • It is not developed here any more. A successor will be a separate record with its own protocol and measured results, not a new version of this one.

Where it comes from

E. Patyukova, J. Yin, S. Basak, S. Pinilla Sanchez, A. Elena and G. Teobaldi, Automatic generation of input files with optimised k-point meshes for Quantum ESPRESSO self-consistent field single-point total energy calculations, Digital Discovery, 2026, 5, 2968–2982. doi:10.1039/d5dd00565e · preprint

20,178 structures from MC3D, reference calculations in Quantum ESPRESSO with SSSP 1.3 PBEsol pseudopotentials. Data at PSDI 75959-bwa52; training code at stfc/goldilocks_kpoints.