query about band reordering

[Dean]

Dear all,
After GW calculations, I find that the band reordering happens. QP energies do not arrange from small to larhe values, please see the datas below.
" QP [eV] @ K [1] (iku): 0.000000 0.000000 0.000000
B=23 Eo= -3.67 E= -4.85 E-Eo= -1.19 Re(Z)=0.77 Im(Z)=-0.175143E-2 nlXC= -22.3558 lXC= -17.3087 So= 3.50469
B=24 Eo= -3.53 E= -4.95 E-Eo= -1.42 Re(Z)=0.78 Im(Z)=-0.180935E-2 nlXC= -18.9270 lXC= -14.4768 So= 2.62507
B=25 Eo= -3.39 E= -4.46 E-Eo= -1.08 Re(Z)=0.76 Im(Z)=-0.204829E-2 nlXC= -21.4974 lXC= -16.8472 So= 3.24276
B=26 Eo= -3.20 E= -3.98 E-Eo= -0.78 Re(Z)=0.76 Im(Z)=-0.128012E-2 nlXC= -25.7508 lXC= -20.6486 So= 4.08062
B=27 Eo= -3.16 E= -3.95 E-Eo= -0.79 Re(Z)=0.75 Im(Z)=-0.164479E-2 nlXC= -22.8949 lXC= -18.2940 So= 3.54895
B=28 Eo= -3.15 E= -4.36 E-Eo= -1.21 Re(Z)=0.77 Im(Z)=-0.228981E-2 nlXC= -20.3397 lXC= -15.8849 So= 2.88288
B=29 Eo= -3.03 E= -4.29 E-Eo= -1.25 Re(Z)=0.77 Im(Z)=-0.266109E-2 nlXC= -19.9514 lXC= -15.5141 So= 2.81835
B=30 Eo= -2.62 E= -3.61 E-Eo= -0.99 Re(Z)=0.77 Im(Z)=-0.154539E-2 nlXC= -21.6207 lXC= -17.4401 So= 2.89157
B=31 Eo= -2.58 E= -3.77 E-Eo= -1.19 Re(Z)=0.77 Im(Z)=-0.160203E-2 nlXC= -21.7075 lXC= -17.1022 So= 3.07256
B=32 Eo= -2.54 E= -4.06 E-Eo= -1.53 Re(Z)=0.78 Im(Z)=-0.189234E-2 nlXC= -18.9362 lXC= -14.4673 So= 2.51086
B=33 Eo= -0.60 E= -1.47 E-Eo= -0.86 Re(Z)=0.79 Im(Z)=-0.129513E-2 nlXC= -16.8764 lXC= -15.3108 So= 0.470181
B=34 Eo= 0.58 E= 1.85 E-Eo= 1.27 Re(Z)=0.79 Im(Z)=-0.647037E-3 nlXC= -10.2084 lXC= -15.7727 So= -3.96542
B=35 Eo= 0.68 E= 1.78 E-Eo= 1.10 Re(Z)=0.79 Im(Z)=-0.816401E-3 nlXC= -10.8965 lXC= -15.7636 So= -3.47564
B=36 Eo= 1.36 E= 2.57 E-Eo= 1.22 Re(Z)=0.79 Im(Z)=-0.103515E-2 nlXC= -10.7124 lXC= -16.2772 So= -4.01906
B=37 Eo= 1.37 E= 2.68 E-Eo= 1.32 Re(Z)=0.78 Im(Z)=-0.193752E-3 nlXC= -10.5697 lXC= -16.6548 So= -4.40572
B=38 Eo= 2.36 E= 3.64 E-Eo= 1.28 Re(Z)=0.78 Im(Z)=-0.862395E-3 nlXC= -12.4788 lXC= -17.1807 So= -3.06701
B=39 Eo= 2.94 E= 4.25 E-Eo= 1.31 Re(Z)=0.79 Im(Z)=-0.138356E-2 nlXC= -10.9602 lXC= -16.8434 So= -4.21473
B=40 Eo= 3.30 E= 4.52 E-Eo= 1.22 Re(Z)=0.81 Im(Z)=-0.104612E-2 nlXC= -9.25290 lXC= -14.4277 So= -3.67791
B=41 Eo= 3.33 E= 4.59 E-Eo= 1.26 Re(Z)=0.79 Im(Z)=-0.168195E-2 nlXC= -10.3271 lXC= -16.3379 So= -4.40285
B=42 Eo= 3.71 E= 5.13 E-Eo= 1.41 Re(Z)=0.80 Im(Z)=-0.940186E-3 nlXC= -10.9999 lXC= -16.7924 So= -4.02613
B=43 Eo= 4.93 E= 5.64 E-Eo= 0.70 Re(Z)=0.87 Im(Z)=-0.119524E-2 nlXC= -3.88729 lXC= -7.12407 So= -2.42761
B=44 Eo= 5.01 E= 5.43 E-Eo= 0.42 Re(Z)=0.91 Im(Z)=-0.774684E-3 nlXC= -1.09512 lXC= -3.02463 So= -1.46755"
What does it mean? Do I need to correct it?
And, the interpolated band stuctures by ypp shows mussy characteristic at some K points? Does it have relationship with the band reordering?
Thanks in advance.

[Daniele Varsano]

Dear Yimin Ding,

band ordering is maintained according to the KS energies. GW corrections can lead to different correction and hence a different ordering of the bands. This usually happens when bands have different character (localization).
About band interpolation, please post your ypp.in file and we will have a look.
Best,
Daniele

[Dean]

Dear Daniele,
Thanks for your reply.
Please see the attachment.
Beat,

[Daniele Varsano]

Dear Yimin,

you can try to change the value of INTERP_Shell_Fac and see if the interpolation improves.
Unfortunately, there is not a rule of thumb for this parameter. Just try to play with it and see if you get a less noisy band structure.

If this is not the case, you can resort to wannier interpolation, this is a bit more laborious, but accuracy should increase.
Here is a guide for that:
viewtopic.php?f=9&t=1358&p=6528&hilit=Wannier#p6494
in the post there are links to tutorial and examples.

Best,
Daniele