Using different PP (LDA FHI, GGA FHI, HGH NC, GGA PBE) I optimized the geometry with (optcell=1 or optcell=2), but when

**I restart optimization from optimized acell and WFN**(with irdwfk 1 keyword) I've got different energy from previously optimized ( higher one) and optimization process continued again to the previously optimized geometry.

graph demonstrate relationship between the x-dim of AgN3 unit cell and iteration number for first and second run.

the PP http://www.abinit.org/downloads/psp-links/lda_fhi

the input:

Code: Select all

`# AgN3 : geometry optimization`

#

# Definition of the unit cell

acell 5.600 5.980 5.998 angstrom

spgroup 72

rprim 1.0 0.0 0.0

0.0 1.0 0.0

0.0 0.0 1.0

chkprim 0

# Definition of the atom types

ntypat 2

znucl 7 47

# Definition of the atoms

natom 16 # There are 16 atoms

typat 4*2 12*1

xred # This keyword indicate that the location of the atoms

# will follow, one triplet of number for each atom

0.0 0.0 1/4 # Note the use of fractions (remember the limited.

1/2 1/2 1/4 # interpreter capabilities of ABINIT)

0.0 0.0 3/4 # Note the use of fractions (remember the limited.

1/2 1/2 3/4 # interpreter capabilities of ABINIT)

0.3422 0.1327 0.0 # 1-st N3 group (N1)

1/2 0.0 0.0 # 1-st N3 group (N2)

0.6578 -0.1327 0.0 # 1-st N3 group (N3)

0.1578 0.6327 0.0 # 2-nd N3 group (N1)

0.0 1/2 0.0 # 2-nd N3 group (N2)

-0.1578 0.3663 0.0 # 2-nd N3 group (N3)

0.1578 0.3663 1/2 # 3-rd N3 group (N1)

0.0 1/2 1/2 # 3-rd N3 group (N2)

-0.1578 0.6327 1/2 # 3-rd N3 group (N3)

0.6578 0.1327 1/2 # 4-th N3 group (N1)

1/2 0.0 1/2 # 4-th N3 group (N2)

0.3422 -0.1327 1/2 # 4-th N3 group (N3)

# Definition of the planewave basis set

ecut 20.0 # Maximal kinetic energy cut-off, in Hartree

# Definition of the k-point grid

kptopt 1 # Option for the automatic generation of k points, taking

# into account the symmetry

ngkpt 1 1 1 # This is a 1x1x1 grid based on the primitive vectors

nshiftk 1 # of the reciprocal space (that form a BCC lattice !),

# repeated four times, with different shifts :

shiftk 0.0 0.0 0.0

# In cartesian coordinates, this grid is simple cubic, and

# actually corresponds to the

# so-called 4x4x4 Monkhorst-Pack grid

# Definition of the SCF procedure

nstep 30 # Maximal number of SCF cycles

toldfe 1.0d-10 # Will stop when, twice in a row, the difference.

# between two consecutive evaluations of total energy.

# differ by less than toldfe (in Hartree).

diemac 3.0 # Although this is not mandatory, it is worth to

# precondition the SCF cycle. The model dielectric

# function used as the standard preconditioner

# is described in the "dielng" input variable section.

prtden 0 # No output of electron density

# add to conserve old < 6.7.2 behavior for calculating forces at each SCF step

# optforces 1

optcell 1

ntime 30

dilatmx 1.2

ecutsm 0.5

ionmov 2

how to avoid this behavior?

Best.

Vladimir.