[Pw_forum] what's wrong with 'nscf' when doing bands?
degironc at sissa.it
Wed Dec 13 09:10:11 CET 2006
W. YU wrote:
>Thank you very much for answering my question.
>>you should not specify calculation='nscf' but
>>band structure run.
>The above suggestion really works for my case. But I
>am at a lost. I have been using 'nscf' for band
>structure calculation for a very long time and found
>no error at all. BTW, I found 'nscf' and 'bands' key
>words mixed in the examples of espresso 3.1.1 for band
>structure calculation. Examples05 use 'nscf' while
>example01 use 'bands'. Both works. Then what is the
>difference between these? If we must use 'bands' for
>band structure, when do we need 'nscf' or it is
I agree that is it confusing but the problem is that in the case you
declare a metallic system 'nscf'
try to locate the Fermi energy which is meaningless for a band
structure calculation (in particular
when kpoints have zero weight as in your input...) while 'bands' does not.
'nscf' was in the past the only option for a non-scf calculation and
this has been in the past a source of
problems because it was used for 1) band structure calculation, 2) BZ
refining in order to obtain nicer
dos etc, and 3) preparation-run for phonon calculations and maybe other
uses as well...
The needs of these different types of calculation are different... 1)
does not need any occupation
number, k-poit sampling and weight should be totally arbitrary and not
changed internally while 2) and
3) reguire and even sampling of the BZ which takes into consideration
reduction of symmetry w.r.t.
the declared bravais lattice , and 3) should be made, to be consistent,
with exactly the same sampling
as the original scf calculation... These competing needs call for
NOW calculation variable has specific keys-values for 'bands' and
'phonon' but still a generic 'nscf'' value
which, as I said, in metallic calculations try to locate the Fermi
energy and create the problem you
reported. For non metallic systems all these options are probably
equivalent or at least nscf and bands are
hence the (lousy) use of 'nscf' in examples as 'bands' replacement.
Hope this clarify but I'm sure this issue will reappear in the forum in
>Need a quick answer? Get one in minutes from people who know.
>Ask your question on www.Answers.yahoo.com
>Pw_forum mailing list
>Pw_forum at pwscf.org
More information about the users