3.1   Number of bands and Electronic temperature

The method 'SetNumberOfBands(nbands)' can be used to define the number of bands for the calculatior. nbands should be a least equal to halv the number of valence electrons.

The occupation statitstics can be set using the method SetOccupationStatistics(method).

dacapo support two methods for occupation statistics:

FermiDirac:

Fermi-Dirac occupation statistics. The temperature is controlled by the method SetElectronicTemperature(temp). Default is 0.1 eV.

MethfesselPaxton:

Currently, the Methfessel-Paxton scheme (PRB 40,3616, 1989) is implemented to 1st order.

3.2   Planewave and Density-cutoff

The method SetPlaneWaveCutoff(planewavecutoff) defines the plavewave cutoff for the calculation. The unit for planewavecutoff is eV.

The method SetDensityCutoff(densitycutoff) sets the density cutoff. By default the two cutoffs are equal, corresponding to using just one grid. For elements like Cu using ultra-soft pseudo potentials the ultra-soft pseudo-wavefunction are very soft, while the density still requires a fine grid, in this case it can be advantagous to set a larger value for densitycutoff that for planewavecutoff.

3.3   Changing file names

The following two methods can be use to change the file names used by dacapo:

SetNetCDFFile(filename):

This will set the NetCDF filename, default is 'out.nc'.

SetTxtFile(filename)

This will set the Ascii text file name, default is 'out.txt'. For information on the most inportant data in the Ascii file see dacapo Ascii file.

The method:

SetScratch(scratchdir):

will set the scratch directory used for swapfiles to scratchdir. Default for this directory is /scratch/<username> or if this is not found just /scratch.

3.4   Exchange-correlation functional

You can use the method SetXCFunctional(exc) to select the exchange-correlation used for the calculation. exc is one of:

VWN:

Vosko Wilk Nusair LDA-parametrization.

PW91:

Perdew Wang 91 GGA-parametrization.

PBE:

Perdew Burke Ernzerhof GGA-parametrization.

RPBE:

Revised PBE GGA-parametrization, Hammer, Hansen and NŽÃŽ¸rskov.

The method GetXCFunctional tells which of the above defined exchange-correlation functionals are currently used.

3.5   Defining a spin-polarized calculation

One can use the method SetSpinPolarized(True) to tell dacapo that the calculation should be spin-polarized.

One also need to specify an initial magnetic moment for the atoms. You can do this in the construction of the atoms, by using:

>>> from ASE import Atom, ListOfAtoms
>>> atoms = ListOfAtoms([Atom('O', (0, 0, 0),magmom=1.0),
...                      Atom('O', (0, 0, 1.2),magmom=1.0)])

This can also be set using the ListOfAtoms method SetMagneticMoments((1,1)) or setting the SetMagneticMoment method on each atom.

This will define a initial total magnetic moment of 2 Bohr magneton, for the O2 molecule, this being also the self-consistent result. You could also use, SetMagneticMoments((1,-1)), this will define a 'anti-ferro' magnetic state with total magnetic moment of zero. If this case probably the end result is still the same (2 Bohr magneton) but the number of self-consistent iterations will increase.

3.6   K-points

The method SetBZKPoints(kpts) can be used to define the k-points in the BZ. In general kpts are a python list of k-points defined in units of the reciprocal cell. You can however specify a 10x10x10 Monkhorst-Pack 3dimensional set simple by using SetBZKpoint((10,10,10)).

Below is a list of the special k-points sets that can be used:

MonkhorstPack:

A 3-dimensional Monkhorst-Pack special k-point set. (see H.J.Monkhorst and J.D.Pack, Physical Review B, vol. 13, page 5188, 1976. You can defined this k-points set simply using, SetBZKPoints((n1,n2,n3)). This will define a n1 by n2 by n3, Monkhorst Pack special k-point set. The size along each reciprocal vector given by n1, n2 and n3, respectively.

ChadiCohen:

A Chadi-Cohen (see Chadi and Cohen, Phys. Review B., vol 8,page 5747) 1-dimensional k-point can be defined using the ASE Utilities module ChadiCohen:

>>> from ASE.Utilities.ChadiCohen import CC6_1x1
>>> calc.SetBZKPoints(CC6_1x1)

This will define a 6 k-point 1x1 Chadi-Cohen special k-point set. In general the Chadi-Cohen sets are named CC+'<Nkpoints>'+_+'shape'. For example an 18 k point sq(3)xsq(3) is named 'CC18_sq3xsq3'.

You can use the method GetBZKPoints, to access the k-points just defined. The method GetIBZKPoints returns the symmetry reduced set of k-points.

3.7   Pseudo-potentials

You can change or view the pseudopotentials use for an element using the methods:

GetPseudoPotential(Z=<elementnumber>):

View the pseudopotential path for element elementnumber.

SetPseudoPotential(Z=<elementnumber>,path=path):

Set the pseudopotential path for element elementnumber.

3.8   Charge-density mixing

The method SetChargeMixing(value) can be used to enable or disable charge density mixing.

Using SetChargeMixing(True) the program will use Pulay mixing for the density (default).

Using SetChargeMixing(False) correspond to running a calculation using the Harris-Foulkes functional using the input density. See also the Harris calculation example.

The method SetKerkerPreconditioning(value) can be used to set Kerker preconditioning for the density mixing. For value=True Kerker preconditiong is used, i.e. q is different from zero, see eq. 82 in Kresse/FurthmŽüller: Comp. Mat. Sci. 6 (1996). The value of q is fix to give a damping of 20 of the lowest q vector. For value=False q is zero and mixing is linear (default).

3.9   Eigenvalue solver

The eigenvalue solver can be set using the method SetEigenvalueSolver(method). Here method is one of:

1. eigsolve: Block Davidson algorithm (Claus Bendtsen et al). This is the default method.
2. rmm-diis: Residual minimization method (RMM), using DIIS (direct inversion in the iterate subspace). The implementation follows closely the algorithm outlined in Kresse and Furthmuller, Comp. Mat. Sci, III.G/III.H

3.10   Dipole-correction

DipoleCorrection(MixingParameter=0.2,InitialValue=0.0):

This methods helps setting up the Dipole Correction scheme, For more information see the dacapo netcdf manual.

3.11   Setting non-default executable

GetJobType:

SetJobType can be used to view the current executable for dacapo.

SetJobType(executable=newexecutable):

SetJobType can be used to change the default dacapo executable dacapo.run to newexecutable.

3.12   Symmetry

SetSymmetryOff:

Do not use symmtry then reducing the k-point set.

SetSymmetryOn:

Use symmtry then reducing the k-point set.

3.13   Electrostatic decoupling

SetElectrostaticDecoupling(numberofgaussians=3,ecutoff=100):

Add electrostatic decoupling (Jan Rossmeisl). Decoupling activates the three dimensional electrostatic decoupling. Based on paper by Peter E. Bloechl: JCP 103 page7422 (1995).

3.14   Stress calculation

By default dacapo will not calculate the stress acting on the unitcell, you can use the method:

CalculateStress:

to tell dacapo to calculate the stress.

3.15   Local density of states

Dacapo can calculate the density of states projected onto atomic orbitals.

Typically one would do:

>>> atoms = ListOfAtoms(..) 
>>>
>>> calc = Dacapo(..)
>>> calc.CalculateAtomicDOS() 
>>>
>>> atoms.SetCalculator(calc)
>>>
>>> energy = atoms.GetPotentialEnergy()
>>> dos = calc.GetLDOS(atoms=[1],angularchannels=["s"]) 
>>> dos.GetPlot()

This will return a density of states projected onto the s-orbital of the first atom (atom numbers starts from 1).

The method:

CalculateAtomicDOS(energywindow=None)

tells the fortran program to calculate the local density of states projected onto atomic orbitals, using energywindow for calculating the energy resolved LDOS.

The method:

GetLDOS(**keywords ):

returns an instance of AtomProjectedDOSTool (see below).

Keyword for the GetDOS method:

• atoms

  List of atoms numbers (default is all atoms)

• angularchannels

  List of angular channel names. The full list of names are: 's', 'p_x', 'p_y', 'p_z','d_zz','dxx-yy', 'd_xy', 'd_xz', 'd_yz'. 'p' and 'd' can be used as shorthand for all p and all d channels respectively. (default is all d channels)

• spin

  List of spins (default all spins)

• cutoffradius

  'short' or 'infinite'. For cutoffradius = 'short' the integrals are truncated at 1 Angstrom. For cutoffradius = 'infinite' the integrals are not truncated. (default 'infinite')

The resulting DOS are added over the members of the three list (atoms,angularchannels and spin).

GetDOS returns a instance of the class AtomProjectedDOSTool (Thanks to John Kitchen for providing grace plot and band moments methods).

Methods for AtomProjectedDOSTool:

GetPlot

returns a GnuPlotAvartar for the energy resolved DOS A parent can be given too combine plot.

GetIntegratedDOS

returns the integral up to the fermi energy

GetData

returns the energy resolved DOS

GetBandMoment(moments)

returns the moments of the projected DOS. GetBandMoment(0) return the integrated DOS. GetBandMoment(1,2) returns the first and second moment of the projected DOS (center and width)

SaveData(filename)

saves the data to the filename

GetGracePlot

makes a GracePlot (if GracePlot is installed)

>>> sim.ados.GetDOS(atoms=[1],angularchannels=['s'],  spin=[spin]))

3.16   Changing convergence parameters

The method

SetConvergenceParameters(**keywords)

can be used to set convergence parameters for the calculations. The following keywords are defined:

• energy:

  Absolute convergence in energy (default 1e-5)

• density:

  Convergence in density (default 0.0001)

• occupation:

  Convergence in occupation numbers (default 0.001)

SetConvergenceParameters(energy=0.00001,density=0.0001,occupation=0.001) correspond to the default settings.

The method

GetConvergenceParameters()

can be used to retrieve the convergence parameters.

3.17   NetCDF specific methods

Two methods are defined for general reading and writing of NetCDF entries. This provides a method for reading and writing any NetCDF variable defined in the dacapo netcdf manual, so that any settings for the dacapo fortran program, not defined by the standard sets of method from the DFT Calculator or from the extra Dacapo method described above, can be defined. The methods are listed below:

SetNetCDFEntry(variable,att=None,value=None):

Define a general netcdf entry for this calculation. If the netcdf entry is allready defined, the value are modified.

GetNetCDFEntry(variable,att=None):

Returns the value of NetCDF entry <variable> or NetCDF attribute att, if specified.

The effect of writing:

>>> calc = Dacapo() 
>>> calc.SetNumberOfBands(10) 

and

>>> calc = Dacapo() 
>>> calc.SetNetCDFEntry('ElectronicBands',att='NumberOfBands',value=10) 

is the same. Now you can get the information by either using

>>> print calc.GetNumberOfBands() 

or

>>> print calc.GetNetCDFEntry('ElectronicBands',att='NumberOfBands')

5.1   reading from an old calculation

If you want to restart a calculation or read specific information from an old calculation, one have to access the information via the atoms, using:

>>> atoms = Dacapo.ReadAtoms(filename='oldfile')

Now you can get the calculator, attached to the atoms, and get the information, using:

>>> calc = atoms.GetCalculator()
>>> print calc.GetEigenValues()

A short-hand version of this is:

>>> print
>>> Dacapo.ReadAtoms(filename='oldfile').GetCalculator.GetEigenValues()

Often ASE methods take the atoms as the argument, using the GetCalculator method to access the relevant information, i.e the VTK methods for 3D plotting:

>>> from ASE.Visualization.VTK import VTKPlotWaveFunction
>>> atoms = Dacapo.ReadAtoms(filename='oldfile')
>>> VTKPlotWaveFunction(atoms)

7.1   Setting up a constrain filter

In the simplest form a filter is just a wrapper around a ListOfAtoms object, eg. for a filters constraining some atoms in a ListOfAtoms, the filter is just a shortened version of the orginal ListOfAtoms. For a full description see the filter in the ASE manual.

Constraining one atom in a CO-molecule:

>>> from ASE import Atom, ListOfAtoms
>>> from ASE.Filters import Subset
>>> atoms = ListOfAtoms([Atom('C', (0,   0, 0), tag=117),
...                  Atom('O', (1.2, 0, 0))])
>>> mask=[atom.GetTag() == 1 for atom in atoms]
>>> subset = Subset(atoms,mask) 

Another often used constrain is to fix a bond-lenght. For a example of combining the subset filter and the fix bond length filter, see the h2o example.

7.2   Setting up a dynamics

Once the filter is defined on can setup a minimization or dynamics working that takes this filter as input.

To setup a conjugate-gradient minimizer using the filter subset defined above, one can use:

>>> from ASE.Dynamics import ConjugateGradient
>>> from ASE.Dynamics.LineMinimizers.LM1 import LM1
>>> dyn = ConjugateGradient(subset, fmax=0.05,lineMin=LM1(0.05))
>>> dyn.Converge()

For a full description of the minimization and dynamics algorithms available see dynamics in the ASE manual.

7.3   Generating a netcdf trajectory

To save the atomic configurations, energies, forces etc, produced from the mimimizer defined above, you need to add a trajectory object to the minimizer:

>>> path = NetCDFTrajectory('CO-trajectory.nc', atoms)
>>> dyn.Attach(path) 

See the h2o example. For more information on trajectories see trajectory in the ASE manual.