Dear vahid

I know in siesta, system when relaxed drop in local state not a global state. 
Moreover, total energy should not change a lot.

best regards

homa




________________________________
 From: Vahid Askarpour <[email protected]>
To: [email protected] 
Sent: Tuesday, November 6, 2012 10:08 PM
Subject: [SIESTA-L] Energy convergence in variable-cell relaxation
 
Dear Siesta Users,

Somehow my first email yesterday never made it to the forum so I am 
resubmitting my question in case no one has received it

I am using the following Zmatrix.data:

LatticeConstant 1.0 Ang
%block LatticeVectors
          6.280434196      -0.005586432       0.001779433
         -0.010035350      10.640250385      -0.140887455
          0.002995775      -0.139896936      10.441268506
%endblock LatticeVectors
ZM.UnitsLength Ang
ZM.UnitsAngle rad
%block Zmatrix
molecule_cartesian
1   0   0   0      3.26412488      0.18434288      8.34284764   1   1   1
2   1   0   0      1.65828859      2.90193293     -1.90908700   1   1   1
molecule_cartesian
1   0   0   0      3.00128916      5.56719236      3.03112031   1   1   1
2   1   0   0      1.66353716      2.91485149     -1.24973899   1   1   1
molecule_cartesian
1   0   0   0      3.60753463      0.02717187      2.80841061   1   1   1
2   1   0   0      1.66906300      2.58058615     -2.97842135   1   1   1
molecule_cartesian
1   0   0   0      2.65255564      5.27969113      7.97252916   1   1   1
2   1   0   0      1.66678629      2.61640917     -0.16745125   1   1   1
molecule_cartesian
1   0   0   0      1.14859333      9.24514323      3.66253191   1   1   1
2   1   0   0      1.67686751      1.59027708     -2.17167021   1   1   1
molecule_cartesian
1   0   0   0      5.06965174      1.28297685      3.79833399   1   1   1
2   1   0   0      1.67453554      1.66224662      0.98895492   1   1   1
molecule_cartesian
1   0   0   0      5.10773991      3.86224401      8.95287684   1   1   1
2   1   0   0      1.66720020      1.62595003     -1.01568338   1   1   1
molecule_cartesian
1   0   0   0      1.17918714      6.56017549      8.95228834   1   1   1
2   1   0   0      1.67350997      1.63380731      2.18415723   1   1   1

the following input.fdf:

SystemName HI
SystemLabel HI
NumberOfSpecies 2
NumberOfAtoms   16

%block ChemicalSpeciesLabel
1    1   H
2    53  I
%endblock ChemicalSpeciesLabel

PAO.BasisSize   DZ
kgrid_cutoff    12.0 Ang
MeshCutoff      120 Ry
PAO.EnergyShift 0.002 Ry

%include mydefaults.fdf

# DFT, Grid, SCF
XC.functional           GGA
XC.authors              PBE
MaxSCFIterations        1000
DM.MixingWeight         0.150
DM.Tolerance            1.d-5
DM.NumberPulay          5      # Number of SCF steps between pulay mixing

# Relaxation, smearing, etc.
ElectronicTemperature   300 K
MD.TypeOfRun            cg     # Type of dynamics
Optim.Broyden           .true.
MD.VariableCell         .true.
MD.ConstantVolume       .false.
MD.MaxForceTol          1d-5 eV/Ang
MD.NumCGsteps           200
Use-Save-CG             .true.
Use-Save-XV             .true.
MD.Broyden.Initial.Inverse.Jacobian 0.20
MD.RemoveIntramolecularPressure .true.
#MD.PreconditionVariableCell 10.0 Ang
MD.TargetPressure       0.001 GPa

# Treatment of the Z-matrix
ZM.ForceTolLen 0.04 eV/Ang
ZM.ForceTolAng 0.0001 eV/deg
ZM.MaxDisplLen 0.1 Ang
ZM.MaxDisplAng 20.0 deg
AtomicCoordinatesFormat NotScaledCartesianAng
%include Zmatrix.data

and the following mydefaults.fdf:

SpinPolarized           .false.     # Spin unpolarized calculation
SolutionMethod          diagon      # OrderN or Diagon

# Output options
WriteCoorInitial        .true.
WriteCoorStep           .true.
WriteForces             .false.
WriteKpoints            .false.
WriteEigenvalues        .false.
WriteKbands             .false.
WriteBands              .false.
WriteMullikenPop        0
WriteCoorXmol           .false.
WriteMDCoorXmol         .false.
WriteMDhistory          .false.

# Options for saving/reading information
DM.UseSaveDM               .false.   # Use DM Continuation files
MD.UseSaveXV               .false.   # Use stored positions and velocities
MD.UseSaveCG               .false.   # Use stored positions and velocities
SaveRho                    .false.   # Write valence pseudocharge at the mesh
SaveDeltaRho               .false.   # Write RHOscf-RHOatm at the mesh
SaveElectrostaticPotential .false.   # Write the total elect. pot. at the mesh
SaveTotalPotential         .false.   # Write the total pot. at the mesh
WriteSiestaDim             .false.   # Write minimum dim to siesta.h and stop
WriteDenchar               .false.   # Write information for DENCHAR

After finishing the run, the enthalpy only changed between -4719.41 and 
-4719.49 but the volume changed between 696 and 716. In other words, the volume 
is changing but the energy remains unchanged. Shouldn't  the energy reach a 
minimum as the structure is relaxed? Or perhaps I am missing something.

Thank you for your suggestions.

Vahid

Responder a