The ABINIT Users Mailing List ( CLOSED )

[Igor Lukacevic nomail <ilukacevic@fizika.unios.hr>]

Dear everyone,

I'm having concerns while trying to optimize a zinc blende phase of InAs. The
only free parameter is a lattice constant (a_exp = 11.44 Bohr). I'm using GGA
TM psps (no 4d electrons in In psp), ecut = 52 Ha and 12x12x12 k-point mesh
with 4 shifts. 

I want to calculate the equilibrium lattice constant in two ways: by detecting
a total energy minimum and by detecting the pressure zero. Energy minimum
calculation gave me a0 = 11.45 Bohr and zero pressure calculation gave me 
11.41
Bohr. Both calculations gave errors less than 0.5%. That worried me. Also, I
was worried because they differed mutualy by 0.04 Bohr. These calculations 
were
done with ecutsm = 0.

Then I remembered that one should include Pulay correction to get the pressure
right. If I understand correctly from the Input variables help, that can be
done by switching on the ecutsm. Putting the ecutsm to 0.5 Ha, energy minimum
calculation gave me a0 = 11.67 Bohr and zero pressure calculation gave me a0 =
11.65 Bohr. These results give the error of 1.7% and 1.5% respectively, which
looks more like a GGA result than those obtained with ecutsm = 0. But they
again differ mutualy, this time by 0.02 Bohr. This is my problem. Changing the
ecutsm value does nothing to get them to be the same.

Shouldn't they be exactly the same, like in Fig. 7 from Rignanese et al., PRB
52, 8160 (1995)?
How could I improve my calculations? Maybe I should follow the procedure of
Eqs. (36) and (37) from the same reference? (but isn't that the same as 
turning
on the ecutsm?) Any help would be welcome.
Could the lack of In 4d electrons influence these results?

Thank you all in advance!

Igor Lukacevic
Department of Physics, Osijek, Croatia