Consultation on the Performance of ABACUS in Hybrid-GGA-based Spin-Polarized Calculations

[wxwth]

Dear ABACUS developers,

I am writing to inquire about the performance of ABACUS in hybrid-GGA-based spin-polarized calculations for a system under periodic boundary conditions (at the Gamma point). My system is a Fe(II)-bearing silicate melt containing approximately 160 atoms, and the spin multiplicity of this system needs to be optimized. At the GGA (i.e., PBE) level, the SCF for this system converged efficiently in VASP using Fermi smearing with σ=kBT. However, when attempting the simulation at the hybrid-GGA level using CP2K with the diagonalization method and smearing, the SCF failed to converge. I use CP2K at the hybrid-GGA level because it is struggling for VASP to perform such simulations for relatively large systems, but the ADMM method in CP2K significantly reduces the computational cost, especially for molecular dynamics simulations. I am wondering if ABACUS can efficiently handle such a system, particularly regarding the implementation of acceleration algorithms for hybrid-GGA and the stability of SCF convergence in spin-polarized calculations. Thank you!

Sincerely,
Xu