I’ve been doing structure relaxations on several different bulk systems and from what I’ve found, the lattice parameters are underestimated compared to experiment and other theory levels. That said, everything I read suggests that PBE overestimates the lattice parameters.
I’m using the
Siesta-psml branch: siesta-psml-R1
Pseudos from PseudoDojo: nc-sr-pbesol-stringent
w/ the following relaxation parameters:
###############GEN. RUN INFO###############v
XC.functional GGA
XC.authors PBEsol
MeshCutoff 200 Ry
MaxSCFIterations 200
DM.MixingWeight 0.1
DM.NumberPulay 10
%block kgrid_Monkhorst_Pack
8 0 0 0.0
0 8 0 0.0
0 0 8 0.0
%endblock kgrid_Monkhorst_Pack
###############GEN. RUN PARAMETERS###############^
###############RELAX PARAMETERS###############v
MD.TypeOfRun CG
MD.NumCGsteps 200
MD.MaxForceTol 0.02 eV/Ang
MD.VariableCell .true.
%block Geometry.Constraints
stress 4 5 6
%endblock
###############RELAX PARAMETERS###############^
The underestimated lattice parameters do provide an accurate band structure as compared to other theory as well as an accurate bandgap compared to both theory and experiment.
Any assistance in understanding why I underestimate when the conjecture is overestimation would be greatly appreciated.