Dear Siesta users,
I am adding (and subtracting) charges from a silicon QD and I am
experiancing the following problems.
With both siesta 3.1 and 3.0 i use:

SpinPolarized            .true.
FixSpin                      .true.
TotalSpin                   1.0
MeshCutoff                256 Ry
MeshSubDivisions     6
NetCharge                -1.00

 DM.MixingWeight      0.001
 DM.Tolerance            1.d-4
 DM.NumberPulay      7
 SolutionMethod         Diagon
 DM.NumberKick            12
 DM.KickMixingWeight    0.04
 ElectronicTemperature   300 K

With Siesta 3.1 I reach convergence but the program crashes when it has to
output the wavefunction (I use: COOP.write .true.)
With Siesta 3.0 I not always reach convergence because the flags
DM.NumberKick and DM.KickMixingWeight  are like ignored (they are read but
the scf cicles are like if they wheren't there) but when the convergence is
reached the wavefunctions are outputted succesfully.
Removing a charge seems to be more difficult than adding one (self
consitency more difficult to rach),
for this reason I say I not always reach convergence (an explanetion on how
siesta charges systems, or a referance would be much appreciated).
I would like to know is someone else has experianced my same problem with
version 3.1 and If version 3.0 allready has the flags
DM.NumberKick and DM.KickMixingWeight working.

Thank you very much in advance,

Luigi.

Responder a