Hi, When I relax Fe(110) surface ( 2 iron layers, the bottom layer is fixed),the caculation stoped with:convergence NOT achieved after 200 iterations.But when I add ‘nbnd=380' in the input file,the caculation end with :Total force = 0.001689 Total SCF correction = 0.001787 SCF correction compared to forces is large: reduce conv_thr to get better values.And bfgs converged in 1 scf cycles and 0 bfgs steps.So I reduce conv_thr from 1.0d-5 to 1.0d-10 and delete‘nbnd=380',I find it still does not converge.What is the role of nbnd and could I get a more reasonable result by adjusting some parameters? &control calculation = 'relax', prefix='relaxFe1012', pseudo_dir = '/public/home/duan1/guiwei/QE/q-e-qe-6.3/pseudo/', outdir='/public/home/duan1/guiwei/QE/q-e-qe-6.3/tempdir/' tstress = .true. tprnfor = .true. etot_conv_thr=1.0D-3 forc_conv_thr=1.0D-2 / &system ibrav= 0, nat=60, ntyp=1, nspin = 2 starting_magnetization(1) =0.4 occupations='smearing', smearing='mp', degauss=0.02, ecutwfc =37.0 ecutrho =240.0 vdw_corr='DFT-D' london_s6 = 0.75 / &electrons conv_thr = 1.0d-5 mixing_beta = 0.1 mixing_mode = 'local-TF' mixing_ndim=16 electron_maxstep=200 / &IONS ion_dynamics='bfgs' / CELL_PARAMETERS angstrom 19.67999 0.000000 0.000000 0.000000 8.521686 0.000000 0.000000 0.000000 17.00000 ATOMIC_SPECIES Fe 55.847 Fe.pbe-n-kjpaw_psl.0.2.4.UPF ATOMIC_POSITIONS crystal ...... K_POINTS automatic 2 2 1 0 0 0
_______________________________________________ users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users