Xinyue Xie

I notice the LATTICE_VECTORS in POSCAR and STRU are different. Are these files used as input or generated as output?  

@hongriTianqi , I change the kpoints to 10×10×10, and keep other parameters unchanged. Then I make calculations via the ABACUS(v3.5), 8 results are all converged. Before calculations, testing a suitable...

INPUT INPUT_PARAMETERS suffix Fe128 pseudo_dir ./ orbital_dir ./ ecutwfc 100 # Rydberg scf_nmax 100 scf_thr 1e-7 # Rydberg basis_type lcao calculation relax smearing_method gauss #type of smearing_method: gauss; fd; fixed;...

 Using regular unit cells as the STRU, 2 step relaxation can be accomplished, but the force is not large.

Failed to upload "Fe128_issue.zip" is shown here. Maybe I can send it to you in other ways.

I tired v.3.6.4 before. There was the same Segmentation.  @xdzhu

For 1 processor, it still has a Segmentation fault.  

Setting ks_solver= scalapack_gvx doesn't help. 

Additionally, if 'basis_type' is set to 'lcao', the calculation continues until it produces an incorrect result, rather than stopping and reporting an error.