too many bands are not converged
刚学qe,在做si的能带和声子计算时没有任何错误,但用si类似的输入文件做ge和gaas时,碰到问题了。
输入文件如下:
+++++++++++++++++++++++++++++++
&control
calculation = 'vc-relax'
prefix = 'gaas'
outdir = './outdir'
pseudo_dir = '/home/work/pseudos'
etot_conv_thr = 1e-5
forc_conv_thr = 1e-4
/
&system
ibrav = 2
celldm(1) = 10.9
nat = 2
ntyp = 2
ecutwfc = 50
ecutrho = 600
/
&electrons
conv_thr = 1.0e-14
/
&ions
/
&cell
cell_dofree = 'ibrav'
/
atomic_species
ga 69.723 ga.pbe-dn-rrkjus_psl.1.0.0.upf
as 74.9216 as.pbe-n-rrkjus_psl.1.0.0.upf
atomic_positions (alat)
ga 0.00 0.00 0.00 0 0 0
as 0.25 0.25 0.25 0 0 0
k_points (automatic)
12 12 12 0 0 0
+++++++++++++++++++++++++++++++
&control
calculation = 'scf'
prefix = 'gaas'
outdir = './outdir'
pseudo_dir = '/home/work/pseudos'
etot_conv_thr = 1e-5
forc_conv_thr = 1e-4
/
&system
ibrav = 2
celldm(1) = 10.8704558116
nat = 2
ntyp = 2
ecutwfc = 50
ecutrho = 600
/
&electrons
conv_thr = 1.0e-14
/
&ions
/
&cell
cell_dofree = 'ibrav'
/
atomic_species
ga 69.723 ga.pbe-dn-rrkjus_psl.1.0.0.upf
as 74.9216 as.pbe-n-rrkjus_psl.1.0.0.upf
atomic_positions (alat)
ga 0.00 0.00 0.00
as 0.25 0.25 0.25
k_points (automatic)
12 12 12 1 1 1
++++++++++++++++++++++++++++++++
&control
calculation = 'nscf'
prefix = 'gaas'
outdir = './outdir'
pseudo_dir = '/home/work/pseudos'
etot_conv_thr = 1e-5
forc_conv_thr = 1e-4
/
&system
ibrav = 2
celldm(1) = 10.8704558116
nat = 2
ntyp = 2
ecutwfc = 50
ecutrho = 600
nbnd = 8
occupations = "tetrahedra'
/
&electrons
conv_thr = 1.0e-14
/
&ions
/
&cell
cell_dofree = 'ibrav'
/
atomic_species
ga 69.723 ga.pbe-dn-rrkjus_psl.1.0.0.upf
as 74.9216 as.pbe-n-rrkjus_psl.1.0.0.upf
atomic_positions (alat)
ga 0.00 0.00 0.00
as 0.25 0.25 0.25
k_points (automatic)
12 12 12 1 1 1
++++++++++++++++++++++++++
前面两步计算好像没有问题,但到了nscf计算,就了出现不收敛的问题。"too many bands are not converged"
之前做ge的计算时,也出现一样的问题,把赝势从ge.pbe-dn-rrkjus_psl.1.0.0.upf改成ge.pbe-n-rrkjus_psl.1.0.0.upf就没有问题,
计算结果也可以和教科书一致。这是什么原因呢?
盼高手给以解答。
另外,虽然使用ge.pbe-dn-rrkjus_psl.1.0.0.upf计算vc-relax时没有出现问题,但迭代步数远多于使用ge.pbe-n-rrkjus_psl.1.0.0.upf。这是什么原因呢?
返回小木虫查看更多
京公网安备 11010802022153号
Try improving the conv_thr (eg. 1E-10, 1E-8) in nscf calculations a little bit to avoid the electronic unconvergence,
dn应该还是半芯态的赝势,你还是用nbnd=8可能求解的能带数目不足,以我自己经验,建议试着放大一些。
conv_thr = 1.0e-10