Page 1 of 1

Number of bands in the non-linear response calc

Posted: Mon Sep 12, 2011 9:21 am
by ilukacevic
Dear all,

when running non-linear calc, I get an error -

In a non-linear response calculation, nband must be equal
to the number of valence bands.

I'm trying to calc. the Raman spectrum of boron (50% doping) doped TiO2. Ti pseudo has 12 valence electrons (2 Ti atoms in the unit cell), O has 6 (3 atoms) and B has 5 (1 atom). I count that would make 47 valence electrons, which occupy 24 bands. This is what I get in the output, when I set all nbands to 24:


occ1 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 1.000000
occ3 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 1.000000
occ4 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 1.000000
occ5 2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 2.000000
2.000000 2.000000 2.000000 2.000000 2.000000 1.000000


Setting nband to 23 or 25 also gave the same error. Could someone help me with this issue? I would appreciate any comment and lecture.

Thank you in advance!

Yours,
Igor Lukacevic

Re: Number of bands in the non-linear response calculations

Posted: Mon Apr 02, 2012 6:51 pm
by mverstra
Hi Igor,

this system is open shell, so "metallic" and the nonlinear stuff can not deal with it...

ciao

Re: Number of bands in the non-linear response calculations

Posted: Wed Apr 04, 2012 7:42 am
by ilukacevic
Acknowledged, sir!

Igor

Re: Number of bands in the non-linear response calculations

Posted: Fri Apr 06, 2012 10:08 am
by ilukacevic
ilukacevic wrote:Acknowledged, sir!

Igor


Correction - the system is not metallic, I use default occopt. it's open shell only because the B pseudo has an odd number of valence electrons. If I took the one with even number of valence electrons, it wouldn't be open shell, right? And the code would work, right?

Igor

Re: Number of bands in the non-linear response calc

Posted: Thu May 24, 2012 2:04 pm
by mverstra
B will always have an odd number of electrons... And in some cases you get metals with even numbers of electrons, so it's not a guarantee. Need to see the electronic structure and make sure you have a proper gap... But the code only stops if it sees that there are unfilled bands (occ < 2.0)