Published July 31, 2026 | Version v3
Dataset Open

Precision verification results data for DFTK@PW|PseudoDojo-v0.5.1-fix

  • 1. ROR icon École Polytechnique Fédérale de Lausanne

* Contact person

Description

The Density-Functional Toolkit (DFTK) is a recent plane-wave density-functional theory (DFT) code, written with a flexible architecture and in a high-level language to facilitate development of new methods and support fundamental mathematical research on DFT. DFTK has been instrumental in a good dozen of recent papers, and its core DFT routines are considered stable. So far a systematic precision benchmark against other codes and approaches had not yet been performed. In this work, we apply the verification protocol established by E. Bosoni et al. [Nat. Rev. Phys. 6, 45-58 (2024)] based on the AiiDA high-throughput engine, to validate the precision of DFTK against the existing all-electron and pseudopotential data. This record contains our results, as well as the scripts needed to reproduce them. Out of the 720 systems, DFTK achieves excellent agreement with the all-electron average (ε<0.06) on 223 systems and good-but-not-excellent agreement (0.06<=&epsilon;<0.2) on 371 systems. Comparing DFTK with ABINIT (ABINIT@PW|PseudoDojo-v0.5), the agreement is excellent on 697 systems and good-but-not-excellent on 14 systems; the agreement is bad (ε>=0.2) for Ne and I systems only, likely due to the corresponding pseudopotentials.

This agreement is made possible by a custom "v0.5.1-fix" of PseudoDojo:

  • The pseudopotential files are truncated at rcut=5.99 bohr.
  • The UPF files in PseudoDojo v0.5 for Bi, Pb, Po, Rn, Te, Tl were missing the third projector and had to be re-generated with a patched ONCVPSP.
  • The new pseudopotentials in PseudoDojo v0.5 do not have a recommended Ecut, so we used the recommended Ecut of PseudoDojo v0.4.1 + 20 Ha. 
We acknowledge and thank Yihan Wu for the initial versions of aiida-dftk and the DFTK support in aiida-common-workflows.
 
 

Files

File preview

All files

Files (2.8 GiB)

Name Apps Size
md5:c1c171bfb602d2d44796cac0f08b316b
1.4 GiB Download
md5:f5a90ef1bfe5f276f77f2d6c4b97dff7
1.3 GiB Download
md5:6cfafbcd0e6c884e9af6097dc20c1b67
49.5 MiB Preview Download
md5:094677c85f31aac31cb942cff735f29b
2.1 KiB Preview Download

Funding

NCCR MARVEL
MARVEL/P3 - Digital Infrastructure of Open Simulations and Data pillar3

References

Journal reference
E. Bosoni et al., Nat. Rev. Phys. 6, 45 (2024), doi: 10.1038/s42254-023-00655-3

Journal reference
M. F. Herbst, A. Levitt, E. Cancès, Proceedings of the JuliaCon conferences 3 (26), 69 (2021), doi: 10.21105/jcon.00069