[Pw_forum] Different band gap value from the 'relax' and 'nscf'calculation.

[#TAO JUNGUANG#]

Dear pwscf users,

When I set calculation='nscf'to calculate the band structure and further
the band gap , I got the valve of about 3.25eV, which agree TOO well
with the experiment value (3.2eV),so even myself double the result. In
order to figure out the reason, I run another one setting
calculation='relax' with the other parameters the same, and I got the
value of 1.64eV, which sounds quite reasonable for DFT. So I am wonder
is any chance that the 'nscf' and 'relax' different so much ?  And
should it right or not that when we use 'nscf' the system read some
information left by our former try? Thanks .

 My input file is shown below

 &CONTROL
  calculation='relax',
  restart_mode='from_scratch',
  pseudo_dir='/home/tao/pseudo/',
  prefix='band'
  tstress=.true.
  tprnfor=.true.
/
&SYSTEM
 ibrav=6, a=4.594,c=2.958, nat=6,ntyp=2,
 ecutwfc=128 ,nosym=.true.,nbnd=32



Tao Junguang