Different results for single and double cells

[denis]

I'm trying to found equilibrium interatomic distance for infinite cumulene chain (...=C=C=C=...) and I found difference between two computations: if the case of 1 atom per cell I get 1.7 A distance, and in the case of 2 atoms per cell I get correct result of 1.3 A. I did both tabulating Etotal vs distance and "optcell 4" calculations. What is wrong in the former case?

Code: Select all

# 1 atom per cell - minimum at 1.7 A
ndtset 10
acell: 1.0 10.0 10.0 Angstrom
acell+ 0.1 0.0 0.0 Angstrom
natom 1  ntypat 1  typat 1  znucl  6
xred 0 0 0
ecut 40
nstep 1000
nband 10
tolvrs 1d-18
diemac 2.0  diemix 0.5
iscf 17

Code: Select all

# 2 atoms per cell - minimum at 2.6/2 = 1.3 A (correct value)
ndtset 10
acell: 2.0 10.0 10.0 Angstrom
acell+ 0.2 0.0 0.0 Angstrom
natom 2  ntypat 1  typat 2*1  znucl  6
xred 0 0 0  1/2 0 0
chkprim 0
ecut 40
nstep 1000
nband 10
tolvrs 1d-18
diemac 2.0  diemix 0.5
iscf 17

PS. ABINIT 7.10.4.

[amadon]

Dear Denis

First, could you give your name and affiliation, thanks.

Your calculations with two atoms per unit cell with the Gamma point only (0,0,0) should be
equivalent to a calculation with one atom per unit cell with two k-points, so your two calculations should logically differ.
Probably, you should first do a static calculation to check the energy, with possibly a converged k-point mesh and then in a second step relax.


Best regards
Bernard