Espresso¶
Quantum ESPRESSO (QE) is an integrated suite of Open-Source computer codes for electronic-structure calculations and materials modeling at the nanoscale. It is based on density-functional theory, plane waves, and pseudopotentials.
The ASE calculator is an interface to the pw.x
executable.
Setup¶
Set up the calculator like a standard FileIOCalculator
:
export ASE_ESPRESSO_COMMAND="/path/to/pw.x -in PREFIX.pwi > PREFIX.pwo"
Any calculation will need pseudopotentials for the elements involved. The
directory for the pseudopotential files can be set with the pseudo_dir
parameter, otherwise QE will look in $ESPRESSO_PSEUDO
if it is set
as an environment variable if set; otherwise $HOME/espresso/pseudo/
is
used. The pseudopotentils are assigned for each element as a dictionary:
pseudopotentials = {'Na': 'Na_pbe_v1.uspp.F.UPF',
'Cl': 'Cl.pbe-n-rrkjus_psl.1.0.0.UPF'}
A simple calculation can be set up:
from ase.build import bulk
from ase.calculators.espresso import Espresso
from ase.constraints import UnitCellFilter
from ase.optimize import LBFGS
rocksalt = bulk('NaCl', crystalstructure='rocksalt', a=6.0)
calc = Espresso(pseudopotentials=pseudopotentials,
tstress=True, tprnfor=True, kpts=(3, 3, 3))
rocksalt.calc = calc
ucf = UnitCellFilter(rocksalt)
opt = LBFGS(ucf)
opt.run(fmax=0.005)
# cubic lattic constant
print((8*rocksalt.get_volume()/len(rocksalt))**(1.0/3.0))
Parameters¶
The calculator will interpret any of the documented options for pw.x
:
http://www.quantum-espresso.org
All parameters must be given in QE units, usually Ry or atomic units in line with the documentation. ASE does not add any defaults over the defaults of QE.
Parameters can be given as keywords and the calculator will put them into
the correct section of the input file. The calculator also accepts a keyword
argument input_data
which is a dict, parameters may be put into sections
in input_data
, but it is not necessary:
input_data = {
'system': {
'ecutwfc': 64,
'ecutrho': 576}
'disk_io': 'low'} # automatically put into 'control'
calc = Espresso(pseudopotentials=pseudopotentials,
tstress=True, tprnfor=True, # kwargs added to parameters
input_data=input_data)
Some parameters are used by ASE, or have additional meaning:
kpts
, is used to specify the k-point sampling * Ifkpts
is a tuple (or list) of 3 integerskpts=(int, int, int)
, it is interpreted as the dimensions of a Monkhorst-Pack grid. * Ifkpts
is set toNone
, only the Γ-point will be included and QE will use routines optimized for Γ-point-only calculations. Compared to Γ-point-only calculations without this optimization (i.e. withkpts=(1, 1, 1)
), the memory and CPU requirements are typically reduced by half. * Ifkpts
is a dict, it will either be interpreted as a path in the Brillouin zone (see ase.dft.kpoints.bandpath) if it contains the ‘path’ keyword, otherwise it is converted to a Monkhorst-Pack grid (see ase.calculators.calculator.kpts2sizeandoffsets)
koffset=(0, 0, 0)
set to 0 or 1 to displace the kpoint grid by a half cell in that direction. Also acceptsTrue
andFalse
.
kspacing=0.1
sets the minimum distance between kpoints in reciprocal space.
nspin=2
if any atom has a magnetic moment spin is turned on automatically.
Any FixAtoms
or FixCartesian
constraints are converted to Espresso
constraints (for dynamic calculations).
Alternative Calculators¶
There are several other QE Calculator
implementations based on ase
that provide a number of extra features:
Espresso Calculator Class¶
- class ase.calculators.espresso.Espresso(restart=None, ignore_bad_restart_file=<object object>, label='espresso', atoms=None, **kwargs)[source]¶
All options for pw.x are copied verbatim to the input file, and put into the correct section. Use
input_data
for parameters that are already in a dict, all otherkwargs
are passed as parameters.Accepts all the options for pw.x as given in the QE docs, plus some additional options:
- input_data: dict
A flat or nested dictionary with input parameters for pw.x
- pseudopotentials: dict
A filename for each atomic species, e.g.
{'O': 'O.pbe-rrkjus.UPF', 'H': 'H.pbe-rrkjus.UPF'}
. A dummy name will be used if none are given.- kspacing: float
Generate a grid of k-points with this as the minimum distance, in A^-1 between them in reciprocal space. If set to None, kpts will be used instead.
- kpts: (int, int, int), dict, or BandPath
If kpts is a tuple (or list) of 3 integers, it is interpreted as the dimensions of a Monkhorst-Pack grid. If
kpts
is set toNone
, only the Γ-point will be included and QE will use routines optimized for Γ-point-only calculations. Compared to Γ-point-only calculations without this optimization (i.e. withkpts=(1, 1, 1)
), the memory and CPU requirements are typically reduced by half. If kpts is a dict, it will either be interpreted as a path in the Brillouin zone (*) if it contains the ‘path’ keyword, otherwise it is converted to a Monkhorst-Pack grid (**). (*) see ase.dft.kpoints.bandpath (**) see ase.calculators.calculator.kpts2sizeandoffsets- koffset: (int, int, int)
Offset of kpoints in each direction. Must be 0 (no offset) or 1 (half grid offset). Setting to True is equivalent to (1, 1, 1).
Note
Set
tprnfor=True
andtstress=True
to calculate forces and stresses.Note
Band structure plots can be made as follows:
Perform a regular self-consistent calculation, saving the wave functions at the end, as well as getting the Fermi energy:
>>> input_data = {<your input data>} >>> calc = Espresso(input_data=input_data, ...) >>> atoms.calc = calc >>> atoms.get_potential_energy() >>> fermi_level = calc.get_fermi_level()
Perform a non-self-consistent ‘band structure’ run after updating your input_data and kpts keywords:
>>> input_data['control'].update({'calculation':'bands', >>> 'restart_mode':'restart', >>> 'verbosity':'high'}) >>> calc.set(kpts={<your Brillouin zone path>}, >>> input_data=input_data) >>> calc.calculate(atoms)
Make the plot using the BandStructure functionality, after setting the Fermi level to that of the prior self-consistent calculation:
>>> bs = calc.band_structure() >>> bs.reference = fermi_energy >>> bs.plot()