Welcome

Welcome to ABACUS-Pseudopotential-Numerical Atomic Orbital Square (APNS). APNS strives to present accurate test results for widely used pseudopotentials and their corresponding ABACUS numerical atomic orbitals, offering users a reference for assessing calculation efficiency. Moreover, APNS is crafted to supply bundles of rigorously tested pseudopotentials and numerical atomic orbitals, which exhibit high precision and are optimized for maximal efficiency, enabling ABACUS users to conduct simulations and create training datasets for deep learning models suitable for both plane-wave ('pw') and linear combination of atomic orbitals ('lcao') calculations. We also provide a suite of workflows to evaluate the suitability of pseudopotentials for specific systems. For additional details, please consult the "About" section.
APNS is mainly developed and maintained by ABACUS-AISI team.

Frequently Asked Questions (Q&A)

Q: What kind of pseudopotentials have been tested? Will other pseudopotentials be tested in the future?

A: APNS collects and tests pseudopotentials from different sources, including but not limited to:

SG15, version from 1.0 to 1.2, both scalar and full relativistic (SR, FR) included
PseudoDojo, version from 0.3 to 0.5, both SR, FR included
PSLibrary 0.3.1 Norm-Conserving (NC)
PD03 and PD04 pseudopotentials
LnPP1 series Goedecker-Teter-Hutter (GTH) pseudopotentials

Pseudopotentials planned to be tested in the future:

PSLibrary 0.3.1 Ultrasoft (US) and PAW
Garrity-Bennett-Rabe-Vanderbilt (GBRV) ultrasoft
ATOMPAW pseudopotentials for Lanthanides
LnPP2, UZH series GTH pseudopotentials

Q: What properties will be tested?

A: For pseudopotential tests, basic electronic structure aspects (including band similarity, total energy), optical/electrical aspects (including bandgaps), mechanical aspects (including Equation of State, Birch-Murnaghan equation) are considered. We also refer to paper How to verify the precision of density-functional-theory implementations via reproducible and universal workflows to measure precision of DFT-implementation-pseudopotential by delta tests.

Q: I find the test result gives a higher ecutwfc than mine, if my previous calculation uses a lower value, does it mean my calculation is wrong or unreliable?

A: One should note that the threshold set as convergence criteria in APNS are those usually used in ABACUS development and in-place testing. However, for practical calculations, the threshold for convergence and accurancy always up to context, and the test result presentend here should be a reference for guiding to a guess of the optimal value. It is always recommended to perform additional convergence tests on system of interest.

Q: What systems are included in pseudopotential and pseudopot-nao bundle tests?

The APNS project greatly benefits from the Materials Project database (link), which offers user-friendly interfaces for high-throughput materials screening. Currently, for each element, only the most stable and experimentally observed entries will be selected (is_stable = True and theoretical = False). To learn how to utilize the Materials Project API for your purposes, please refer to the Materials Project API.

Q: How to download pseudopotentials tested in APNS?

A: Pseudopotentials tested in APNS are available in its Github repository. Users can download them from here.

Q: How to download pseudopotential-numerical atomic orbital bundles tested in APNS?

A: Presently ABACUS numerical atomic orbitals are not fully directly available for download. For numerical atomic orbitals bundled with SG15-1.0 (SR) pseudopotentials, users can download them from ABACUS official website: PSEUDOPOTENTIALS AND ORBITALS. Users can also generate numerical atomic orbitals by themselves using ABACUS orbital generation code, for more information, please refer to official Github repository (ABACUS-orbitals) and corresponding hands-on tutorial maintained by ABACUS-PKU team:

ABACUS-AISI team is also working on refactoring orbital generation codes for both higher efficiency and precision.

Q: How to make my calculation results reliable?

A: To make your calculation results reliable, we recommend you to:

Always use tested pseudopotentials and numerical atomic orbitals
Based on converged values presented here, do additional convergence test on ecutwfc, kpoints sampling, realspace cutoff radius for numerical orbitals, etc.

Q: How to contribute to APNS?

A: We welcome contributions from the community. You can contribute to APNS by submitting issues, pull requests, or by contacting us directly.