"""
This module defines the ASE interface to SIESTA.
Written by Mads Engelund (see www.espeem.com)
Home of the SIESTA package:
http://www.uam.es/departamentos/ciencias/fismateriac/siesta
2017.04 - Pedro Brandimarte: changes for python 2-3 compatible
"""
import os
import re
import shutil
import tempfile
from dataclasses import dataclass
from pathlib import Path
from typing import Any, Dict, List
import numpy as np
from ase import Atoms
from ase.calculators.calculator import FileIOCalculator, Parameters, ReadError
from ase.calculators.siesta.import_ion_xml import get_ion
from ase.calculators.siesta.parameters import PAOBasisBlock, format_fdf
from ase.data import atomic_numbers
from ase.io.siesta import read_siesta_xv
from ase.io.siesta_input import SiestaInput
from ase.units import Ry, eV
from ase.utils import deprecated
meV = 0.001 * eV
def parse_siesta_version(output: bytes) -> str:
match = re.search(rb'Siesta Version\s*:\s*(\S+)', output)
if match is None:
raise RuntimeError('Could not get Siesta version info from output '
'{!r}'.format(output))
string = match.group(1).decode('ascii')
return string
def get_siesta_version(executable: str) -> str:
""" Return SIESTA version number.
Run the command, for instance 'siesta' and
then parse the output in order find the
version number.
"""
# XXX We need a test of this kind of function. But Siesta().command
# is not enough to tell us how to run Siesta, because it could contain
# all sorts of mpirun and other weird parts.
temp_dirname = tempfile.mkdtemp(prefix='siesta-version-check-')
try:
from subprocess import PIPE, Popen
proc = Popen([executable],
stdin=PIPE,
stdout=PIPE,
stderr=PIPE,
cwd=temp_dirname)
output, _ = proc.communicate()
# We are not providing any input, so Siesta will give us a failure
# saying that it has no Chemical_species_label and exit status 1
# (as of siesta-4.1-b4)
finally:
shutil.rmtree(temp_dirname)
return parse_siesta_version(output)
def format_block(name, block):
lines = [f'%block {name}']
for row in block:
data = ' '.join(str(obj) for obj in row)
lines.append(f' {data}')
lines.append(f'%endblock {name}')
return '\n'.join(lines)
def bandpath2bandpoints(path):
return '\n'.join([
'BandLinesScale ReciprocalLatticeVectors',
format_block('BandPoints', path.kpts)])
class SiestaParameters(Parameters):
def __init__(
self,
label='siesta',
mesh_cutoff=200 * Ry,
energy_shift=100 * meV,
kpts=None,
xc='LDA',
basis_set='DZP',
spin='non-polarized',
species=(),
pseudo_qualifier=None,
pseudo_path=None,
symlink_pseudos=None,
atoms=None,
restart=None,
fdf_arguments=None,
atomic_coord_format='xyz',
bandpath=None):
kwargs = locals()
kwargs.pop('self')
Parameters.__init__(self, **kwargs)
def _nonpolarized_alias(_: List, kwargs: Dict[str, Any]) -> bool:
if kwargs.get("spin", None) == "UNPOLARIZED":
kwargs["spin"] = "non-polarized"
return True
return False
[docs]class Siesta(FileIOCalculator):
"""Calculator interface to the SIESTA code.
"""
allowed_xc = {
'LDA': ['PZ', 'CA', 'PW92'],
'GGA': ['PW91', 'PBE', 'revPBE', 'RPBE',
'WC', 'AM05', 'PBEsol', 'PBEJsJrLO',
'PBEGcGxLO', 'PBEGcGxHEG', 'BLYP'],
'VDW': ['DRSLL', 'LMKLL', 'KBM', 'C09', 'BH', 'VV']}
name = 'siesta'
_legacy_default_command = 'siesta < PREFIX.fdf > PREFIX.out'
implemented_properties = [
'energy',
'free_energy',
'forces',
'stress',
'dipole',
'eigenvalues',
'density',
'fermi_energy']
# Dictionary of valid input vaiables.
default_parameters = SiestaParameters()
# XXX Not a ASE standard mechanism (yet). We need to communicate to
# ase.spectrum.band_structure.calculate_band_structure() that we expect
# it to use the bandpath keyword.
accepts_bandpath_keyword = True
fileio_rules = FileIOCalculator.ruleset(
stdin_name='{prefix}.fdf',
stdout_name='{prefix}.out')
def __init__(self, command=None, profile=None, directory='.', **kwargs):
"""ASE interface to the SIESTA code.
Parameters:
- label : The basename of all files created during
calculation.
- mesh_cutoff : Energy in eV.
The mesh cutoff energy for determining number of
grid points in the matrix-element calculation.
- energy_shift : Energy in eV
The confining energy of the basis set generation.
- kpts : Tuple of 3 integers, the k-points in different
directions.
- xc : The exchange-correlation potential. Can be set to
any allowed value for either the Siesta
XC.funtional or XC.authors keyword. Default "LDA"
- basis_set : "SZ"|"SZP"|"DZ"|"DZP"|"TZP", strings which specify
the type of functions basis set.
- spin : "non-polarized"|"collinear"|
"non-collinear|spin-orbit".
The level of spin description to be used.
- species : None|list of Species objects. The species objects
can be used to to specify the basis set,
pseudopotential and whether the species is ghost.
The tag on the atoms object and the element is used
together to identify the species.
- pseudo_path : None|path. This path is where
pseudopotentials are taken from.
If None is given, then then the path given
in $SIESTA_PP_PATH will be used.
- pseudo_qualifier: None|string. This string will be added to the
pseudopotential path that will be retrieved.
For hydrogen with qualifier "abc" the
pseudopotential "H.abc.psf" will be retrieved.
- symlink_pseudos: None|bool
If true, symlink pseudopotentials
into the calculation directory, else copy them.
Defaults to true on Unix and false on Windows.
- atoms : The Atoms object.
- restart : str. Prefix for restart file.
May contain a directory.
Default is None, don't restart.
- fdf_arguments: Explicitly given fdf arguments. Dictonary using
Siesta keywords as given in the manual. List values
are written as fdf blocks with each element on a
separate line, while tuples will write each element
in a single line. ASE units are assumed in the
input.
- atomic_coord_format: "xyz"|"zmatrix", strings to switch between
the default way of entering the system's geometry
(via the block AtomicCoordinatesAndAtomicSpecies)
and a recent method via the block Zmatrix. The
block Zmatrix allows to specify basic geometry
constrains such as realized through the ASE classes
FixAtom, FixedLine and FixedPlane.
"""
# Put in the default arguments.
parameters = self.default_parameters.__class__(**kwargs)
# Call the base class.
FileIOCalculator.__init__(
self,
command=command,
profile=profile,
directory=directory,
**parameters)
def __getitem__(self, key):
"""Convenience method to retrieve a parameter as
calculator[key] rather than calculator.parameters[key]
Parameters:
-key : str, the name of the parameters to get.
"""
return self.parameters[key]
def species(self, atoms):
"""Find all relevant species depending on the atoms object and
species input.
Parameters :
- atoms : An Atoms object.
"""
return SiestaInput.get_species(
atoms, list(self['species']), self['basis_set'])
@deprecated(
"The keyword 'UNPOLARIZED' has been deprecated,"
"and replaced by 'non-polarized'",
category=FutureWarning,
callback=_nonpolarized_alias,
)
def set(self, **kwargs):
"""Set all parameters.
Parameters:
-kwargs : Dictionary containing the keywords defined in
SiestaParameters.
.. deprecated:: 3.18.2
The keyword 'UNPOLARIZED' has been deprecated and replaced by
'non-polarized'
"""
# XXX Inserted these next few lines because set() would otherwise
# discard all previously set keywords to their defaults! --askhl
current = self.parameters.copy()
current.update(kwargs)
kwargs = current
# Find not allowed keys.
default_keys = list(self.__class__.default_parameters)
offending_keys = set(kwargs) - set(default_keys)
if len(offending_keys) > 0:
mess = "'set' does not take the keywords: %s "
raise ValueError(mess % list(offending_keys))
# Use the default parameters.
parameters = self.__class__.default_parameters.copy()
parameters.update(kwargs)
kwargs = parameters
# Check energy inputs.
for arg in ['mesh_cutoff', 'energy_shift']:
value = kwargs.get(arg)
if value is None:
continue
if not (isinstance(value, (float, int)) and value > 0):
mess = "'{}' must be a positive number(in eV), \
got '{}'".format(arg, value)
raise ValueError(mess)
# Check the functional input.
xc = kwargs.get('xc', 'LDA')
if isinstance(xc, (tuple, list)) and len(xc) == 2:
functional, authors = xc
if functional.lower() not in [k.lower() for k in self.allowed_xc]:
mess = f"Unrecognized functional keyword: '{functional}'"
raise ValueError(mess)
lsauthorslower = [a.lower() for a in self.allowed_xc[functional]]
if authors.lower() not in lsauthorslower:
mess = "Unrecognized authors keyword for %s: '%s'"
raise ValueError(mess % (functional, authors))
elif xc in self.allowed_xc:
functional = xc
authors = self.allowed_xc[xc][0]
else:
found = False
for key, value in self.allowed_xc.items():
if xc in value:
found = True
functional = key
authors = xc
break
if not found:
raise ValueError(f"Unrecognized 'xc' keyword: '{xc}'")
kwargs['xc'] = (functional, authors)
# Check fdf_arguments.
if kwargs['fdf_arguments'] is None:
kwargs['fdf_arguments'] = {}
if not isinstance(kwargs['fdf_arguments'], dict):
raise TypeError("fdf_arguments must be a dictionary.")
# Call baseclass.
FileIOCalculator.set(self, **kwargs)
def set_fdf_arguments(self, fdf_arguments):
""" Set the fdf_arguments after the initialization of the
calculator.
"""
self.validate_fdf_arguments(fdf_arguments)
FileIOCalculator.set(self, fdf_arguments=fdf_arguments)
def validate_fdf_arguments(self, fdf_arguments):
""" Raises error if the fdf_argument input is not a
dictionary of allowed keys.
"""
# None is valid
if fdf_arguments is None:
return
# Type checking.
if not isinstance(fdf_arguments, dict):
raise TypeError("fdf_arguments must be a dictionary.")
def write_input(self, atoms, properties=None, system_changes=None):
"""Write input (fdf)-file.
See calculator.py for further details.
Parameters:
- atoms : The Atoms object to write.
- properties : The properties which should be calculated.
- system_changes : List of properties changed since last run.
"""
super().write_input(
atoms=atoms,
properties=properties,
system_changes=system_changes)
filename = self.getpath(ext='fdf')
more_fdf_args = {}
# Use the saved density matrix if only 'cell' and 'positions'
# have changed.
if (system_changes is None or
('numbers' not in system_changes and
'initial_magmoms' not in system_changes and
'initial_charges' not in system_changes)):
more_fdf_args['DM.UseSaveDM'] = True
if 'density' in properties:
more_fdf_args['SaveRho'] = True
species, species_numbers = self.species(atoms)
if self['pseudo_path'] is not None:
pseudo_path = self['pseudo_path']
elif 'SIESTA_PP_PATH' in self.cfg:
pseudo_path = self.cfg['SIESTA_PP_PATH']
else:
mess = "Please set the environment variable 'SIESTA_PP_PATH'"
raise Exception(mess)
species_info = SpeciesInfo(
atoms=atoms,
pseudo_path=Path(pseudo_path),
pseudo_qualifier=self.pseudo_qualifier(),
species=species)
writer = FDFWriter(
name=self.prefix,
xc=self['xc'],
spin=self['spin'],
mesh_cutoff=self['mesh_cutoff'],
energy_shift=self['energy_shift'],
fdf_user_args=self['fdf_arguments'],
more_fdf_args=more_fdf_args,
species_numbers=species_numbers,
atomic_coord_format=self['atomic_coord_format'].lower(),
kpts=self['kpts'],
bandpath=self['bandpath'],
species_info=species_info,
)
with open(filename, 'w') as fd:
writer.write(fd)
writer.link_pseudos_into_directory(
symlink_pseudos=self['symlink_pseudos'],
directory=Path(self.directory))
def read(self, filename):
"""Read structural parameters from file .XV file
Read other results from other files
filename : siesta.XV
"""
fname = self.getpath(filename)
if not fname.exists():
raise ReadError(f"The restart file '{fname}' does not exist")
with fname.open() as fd:
self.atoms = read_siesta_xv(fd)
self.read_results()
def getpath(self, fname=None, ext=None):
""" Returns the directory/fname string """
if fname is None:
fname = self.prefix
if ext is not None:
fname = f'{fname}.{ext}'
return Path(self.directory) / fname
def pseudo_qualifier(self):
"""Get the extra string used in the middle of the pseudopotential.
The retrieved pseudopotential for a specific element will be
'H.xxx.psf' for the element 'H' with qualifier 'xxx'. If qualifier
is set to None then the qualifier is set to functional name.
"""
if self['pseudo_qualifier'] is None:
return self['xc'][0].lower()
else:
return self['pseudo_qualifier']
def read_results(self):
"""Read the results."""
from ase.io.siesta_output import OutputReader
reader = OutputReader(prefix=self.prefix,
directory=Path(self.directory),
bandpath=self['bandpath'])
results = reader.read_results()
self.results.update(results)
self.results['ion'] = self.read_ion(self.atoms)
def read_ion(self, atoms):
"""
Read the ion.xml file of each specie
"""
species, species_numbers = self.species(atoms)
ion_results = {}
for species_number, spec in enumerate(species, start=1):
symbol = spec['symbol']
atomic_number = atomic_numbers[symbol]
if spec['pseudopotential'] is None:
if self.pseudo_qualifier() == '':
label = symbol
else:
label = f"{symbol}.{self.pseudo_qualifier()}"
pseudopotential = self.getpath(label, 'psf')
else:
pseudopotential = Path(spec['pseudopotential'])
label = pseudopotential.stem
name = f"{label}.{species_number}"
if spec['ghost']:
name = f"{name}.ghost"
atomic_number = -atomic_number
label = name.rsplit('.', 2)[0]
if label not in ion_results:
fname = self.getpath(label, 'ion.xml')
fname = Path(fname)
if fname.is_file():
ion_results[label] = get_ion(fname)
return ion_results
def band_structure(self):
return self.results['bandstructure']
def get_fermi_level(self):
return self.results['fermi_energy']
def get_k_point_weights(self):
return self.results['kpoint_weights']
def get_ibz_k_points(self):
return self.results['kpoints']
def get_eigenvalues(self, kpt=0, spin=0):
return self.results['eigenvalues'][spin, kpt]
def get_number_of_spins(self):
return self.results['eigenvalues'].shape[0]
def generate_atomic_coordinates(atoms: Atoms, species_numbers,
atomic_coord_format: str):
"""Write atomic coordinates.
Parameters
----------
fd : IO
An open file object.
atoms : Atoms
An atoms object.
"""
if atomic_coord_format == 'xyz':
return generate_atomic_coordinates_xyz(atoms, species_numbers)
elif atomic_coord_format == 'zmatrix':
return generate_atomic_coordinates_zmatrix(atoms, species_numbers)
else:
raise RuntimeError(
f'Unknown atomic_coord_format: {atomic_coord_format}')
def generate_atomic_coordinates_zmatrix(atoms: Atoms, species_numbers):
"""Write atomic coordinates in Z-matrix format.
Parameters
----------
fd : IO
An open file object.
atoms : Atoms
An atoms object.
"""
yield '\n'
yield var('ZM.UnitsLength', 'Ang')
yield '%block Zmatrix\n'
yield ' cartesian\n'
fstr = "{:5d}" + "{:20.10f}" * 3 + "{:3d}" * 3 + "{:7d} {:s}\n"
a2constr = SiestaInput.make_xyz_constraints(atoms)
a2p, a2s = atoms.get_positions(), atoms.symbols
for ia, (sp, xyz, ccc, sym) in enumerate(
zip(species_numbers, a2p, a2constr, a2s)):
yield fstr.format(
sp, xyz[0], xyz[1], xyz[2], ccc[0],
ccc[1], ccc[2], ia + 1, sym)
yield '%endblock Zmatrix\n'
# origin = tuple(-atoms.get_celldisp().flatten())
# yield block('AtomicCoordinatesOrigin', [origin])
def generate_atomic_coordinates_xyz(atoms: Atoms, species_numbers):
"""Write atomic coordinates.
Parameters
----------
fd : IO
An open file object.
atoms : Atoms
An atoms object.
"""
yield '\n'
yield var('AtomicCoordinatesFormat', 'Ang')
yield block('AtomicCoordinatesAndAtomicSpecies',
[[*atom.position, number]
for atom, number in zip(atoms, species_numbers)])
yield '\n'
# origin = tuple(-atoms.get_celldisp().flatten())
# yield block('AtomicCoordinatesOrigin', [origin])
@dataclass
class SpeciesInfo:
atoms: Atoms
pseudo_path: Path
pseudo_qualifier: str
species: dict # actually a kind of Parameters object, should refactor
def __post_init__(self):
pao_basis = []
chemical_labels = []
basis_sizes = []
file_instructions = []
for species_number, spec in enumerate(self.species, start=1):
symbol = spec['symbol']
atomic_number = atomic_numbers[symbol]
if spec['pseudopotential'] is None:
if self.pseudo_qualifier == '':
label = symbol
else:
label = f"{symbol}.{self.pseudo_qualifier}"
src_path = self.pseudo_path / f"{label}.psf"
else:
src_path = Path(spec['pseudopotential'])
label = src_path.stem
if not src_path.is_absolute():
src_path = self.pseudo_path / src_path
if not src_path.exists():
src_path = self.pseudo_path / f"{symbol}.psml"
name = src_path.name
name = name.split('.')
name.insert(-1, str(species_number))
if spec['ghost']:
name.insert(-1, 'ghost')
atomic_number = -atomic_number
name = '.'.join(name)
instr = FileInstruction(src_path, name)
file_instructions.append(instr)
label = '.'.join(np.array(name.split('.'))[:-1])
string = ' %d %d %s' % (species_number, atomic_number, label)
chemical_labels.append(string)
if isinstance(spec['basis_set'], PAOBasisBlock):
pao_basis.append(spec['basis_set'].script(label))
else:
basis_sizes.append((" " + label, spec['basis_set']))
self.file_instructions = file_instructions
self.chemical_labels = chemical_labels
self.pao_basis = pao_basis
self.basis_sizes = basis_sizes
def generate_text(self):
yield var('NumberOfSpecies', len(self.species))
yield var('NumberOfAtoms', len(self.atoms))
yield var('ChemicalSpecieslabel', self.chemical_labels)
yield '\n'
yield var('PAO.Basis', self.pao_basis)
yield var('PAO.BasisSizes', self.basis_sizes)
yield '\n'
@dataclass
class FileInstruction:
src_path: Path
targetname: str
def copy_to(self, directory):
self._link(shutil.copy, directory)
def symlink_to(self, directory):
self._link(os.symlink, directory)
def _link(self, file_operation, directory):
dst_path = directory / self.targetname
if self.src_path == dst_path:
return
dst_path.unlink(missing_ok=True)
file_operation(self.src_path, dst_path)
@dataclass
class FDFWriter:
name: str
xc: str
fdf_user_args: dict
more_fdf_args: dict
mesh_cutoff: float
energy_shift: float
spin: str
species_numbers: object # ?
atomic_coord_format: str
kpts: object # ?
bandpath: object # ?
species_info: object
def write(self, fd):
for chunk in self.generate_text():
fd.write(chunk)
def generate_text(self):
yield var('SystemName', self.name)
yield var('SystemLabel', self.name)
yield "\n"
# Write explicitly given options first to
# allow the user to override anything.
fdf_arguments = self.fdf_user_args
for key in sorted(fdf_arguments):
yield var(key, fdf_arguments[key])
# Force siesta to return error on no convergence.
# as default consistent with ASE expectations.
if 'SCFMustConverge' not in fdf_arguments:
yield var('SCFMustConverge', True)
yield '\n'
yield var('Spin', self.spin)
# Spin backwards compatibility.
if self.spin == 'collinear':
key = 'SpinPolarized'
elif self.spin == 'non-collinear':
key = 'NonCollinearSpin'
else:
key = None
if key is not None:
yield var(key, (True, '# Backwards compatibility.'))
# Write functional.
functional, authors = self.xc
yield var('XC.functional', functional)
yield var('XC.authors', authors)
yield '\n'
# Write mesh cutoff and energy shift.
yield var('MeshCutoff', (self.mesh_cutoff, 'eV'))
yield var('PAO.EnergyShift', (self.energy_shift, 'eV'))
yield '\n'
yield from self.species_info.generate_text()
yield from self.generate_atoms_text(self.species_info.atoms)
for key, value in self.more_fdf_args.items():
yield var(key, value)
if self.kpts is not None:
kpts = np.array(self.kpts)
yield from SiestaInput.generate_kpts(kpts)
if self.bandpath is not None:
lines = bandpath2bandpoints(self.bandpath)
assert isinstance(lines, str) # rename this variable?
yield lines
yield '\n'
def generate_atoms_text(self, atoms: Atoms):
"""Translate the Atoms object to fdf-format."""
cell = atoms.cell
yield '\n'
if cell.rank in [1, 2]:
raise ValueError('Expected 3D unit cell or no unit cell. You may '
'wish to add vacuum along some directions.')
if np.any(cell):
yield var('LatticeConstant', '1.0 Ang')
yield block('LatticeVectors', cell)
yield from generate_atomic_coordinates(
atoms, self.species_numbers, self.atomic_coord_format)
# Write magnetic moments.
magmoms = atoms.get_initial_magnetic_moments()
# The DM.InitSpin block must be written to initialize to
# no spin. SIESTA default is FM initialization, if the
# block is not written, but we must conform to the
# atoms object.
if len(magmoms) == 0:
yield '#Empty block forces ASE initialization.\n'
yield '%block DM.InitSpin\n'
if len(magmoms) != 0 and isinstance(magmoms[0], np.ndarray):
for n, M in enumerate(magmoms):
if M[0] != 0:
yield (' %d %.14f %.14f %.14f \n'
% (n + 1, M[0], M[1], M[2]))
elif len(magmoms) != 0 and isinstance(magmoms[0], float):
for n, M in enumerate(magmoms):
if M != 0:
yield ' %d %.14f \n' % (n + 1, M)
yield '%endblock DM.InitSpin\n'
yield '\n'
def link_pseudos_into_directory(self, *, symlink_pseudos=None, directory):
if symlink_pseudos is None:
symlink_pseudos = os.name != 'nt'
for instruction in self.species_info.file_instructions:
if symlink_pseudos:
instruction.symlink_to(directory)
else:
instruction.copy_to(directory)
# Utilities for generating bits of strings.
#
# We are re-aliasing format_fdf and format_block in the anticipation
# that they may change, or we might move this onto a Formatter object
# which applies consistent spacings etc.
def var(key, value):
return format_fdf(key, value)
def block(name, data):
return format_block(name, data)