当前位置: 首页 > 第一原理 >too many bands are not converged

too many bands are not converged

作者 hedge80
来源: 小木虫 150 3 举报帖子
+关注

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

今日热帖
  • 精华评论
  • fyuewen

    Try improving the conv_thr (eg. 1E-10, 1E-8) in nscf calculations a little bit to avoid the electronic unconvergence,

  • 卡开发发

    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。

  • 漫天飘雪

    conv_thr = 1.0e-10

猜你喜欢
下载小木虫APP
与700万科研达人随时交流
  • 二维码
  • IOS
  • 安卓