Dear Camps, Is it possible that your first structure meets the CG convergence criteria on the first optimization step? You could try setting MD.MaxForceTol to a value smaller than the default, which I think is something like 0.05d0 eV/Ang. Also, just anecdotally, I find that setting MD.TypeOfRun to 'Broyden' instead of 'CG' results in an optimized structure in fewer steps. (The default values for all of the Broyden options seem to work fine).
Best of luck, Mike On Nov 16, 2011, at 5:01 PM, I. Camps wrote: > Hello SIESTA users, > > I am doing some structural relaxation test for III-V semiconductors. I > decided to start with zinc blend GaAs. The pseudo potentials were downloaded > from the SIESTA site for both LDA and GGA. > > The problem is that SIESTA is not doing structural relaxation for my input > file below. It stops after the first step without any error and with the > resulting atomic position identical to the input. > > If I change the second atom type (As) by, for example, oxigen or magnesium, > SIESTA run until the structural relaxation is done. These induce me to think > that the problem is not with the input file. > > I appreciate any help. > > Best regards, > > Camps > > > #*************** INPUT FILE ********************* > SystemName GaAs_zb > SystemLabel GaAs_zb > > NumberOfAtoms 8 > > NumberOfSpecies 2 > > %block ChemicalSpeciesLabel > 1 31 Ga > 2 33 As > %endblock ChemicalSpeciesLabel > > LatticeConstant 1.0 Ang > > %block LatticeParameters > 5.623300 5.623300 5.623300 90.000000 90.000000 90.000000 > %endblock LatticeParameters > > AtomicCoordinatesFormat Fractional > > %block AtomicCoordinatesAndAtomicSpecies > 0.000000000 0.000000000 0.000000000 1 > 0.250000000 0.250000000 0.250000000 2 > 0.750000000 0.750000000 0.250000000 2 > 0.750000000 0.250000000 0.750000000 2 > 0.250000000 0.750000000 0.750000000 2 > 0.000000000 0.500000000 0.500000000 1 > 0.500000000 0.000000000 0.500000000 1 > 0.500000000 0.500000000 0.000000000 1 > %endblock AtomicCoordinatesAndAtomicSpecies > > MD.TypeOfRun CG > MD.NumCGsteps 500 > #*************** INPUT FILE *********************
