Page 1 of 1

wrong band orders (eigenvalue or orbital) of GW calculations

Posted: Sat Apr 11, 2020 4:41 am
by xiaoming_wang
Hello,

I'm trying to do GW calculations of formaldehyde using vasp6. I tried single shot (NELM=1) calculations for EVGW0 and QPGW0R.

For EVGW0, the band ordering in the EIGENVAL file is wrong (see #7 and #8).
-----
1 -32.413511 1.000000
2 -19.809472 1.000000
3 -15.832828 1.000000
4 -14.767805 1.000000
5 -14.269542 1.000000
6 -10.105790 1.000000
7 1.421239 0.000000
8 0.667624 0.000000
9 2.740717 0.000000
10 2.992458 0.000000
11 2.707432 0.000000
12 2.943946 0.000000
13 2.769052 0.000000
-------
Looking at the orbitals (LPARD), the orbital character ordering seems correct, i.e., pi, n, pi*, sigma* for #5 #6 #7 #8, respectively.

My INCAR for EVGW0 is
---
SYSTEM = H2CO
ENCUT = 435 eV
PREC = A
LREAL = F
GGA = PE

ISMEAR = 0
SIGMA = 0.05
EDIFF = 1.e-8

NBANDS = 10620
ALGO = EVGW0
NELM = 1
NOMEGA = 60
---


For QPGW0R, the band ordering in EIGENVAL is correct, however, the orbital character ordering is wrong. The orbital order turns out to be pi, n, sigma*, pi* for #5 #6 #7 #8.
The EIGENVAL file:
---
1 -32.111647 1.000000
2 -21.009890 1.000000
3 -17.081610 1.000000
4 -15.737043 1.000000
5 -14.445081 1.000000
6 -10.563273 1.000000
7 -0.019784 0.000000
8 0.380517 0.000000
9 1.825867 0.000000
10 2.079296 0.000000
11 2.118000 0.000000
12 2.542946 0.000000
13 2.625485 0.000000
---
My INCAR for QPGW0R is

---
SYSTEM = H2CO
ENCUT = 435 eV
PREC = A
LREAL = F
GGA = PE

ISMEAR = 0
SIGMA = 0.05
EDIFF = 1.e-8

NBANDS = 10620
ALGO = QPGW0R
NELM = 1
NOMEGA = 16
---

This is not specific for vasp6, vasp5.4.4 also has the same issue. I'm wondering if there is a quick fix for this?

Best,
Xiaoming

Re: wrong band orders (eigenvalue or orbital) of GW calculations

Posted: Mon Apr 27, 2020 7:55 am
by merzuk.kaltak
This post is related to this thread.