# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at https://mozilla.org/MPL/2.0/.
from numbers import Integral
from itertools import product
from collections import deque, namedtuple
import numpy as np
from numpy import pi
try:
from . import _siesta
found_module = True
except Exception as e:
print(e)
found_module = False
from ..sile import add_sile, SileError, SileWarning
from .sile import SileBinSiesta
from sisl._internal import set_module
from sisl.messages import info, warn, SislError
from ._help import *
import sisl._array as _a
from sisl import Geometry, Atom, Atoms, SuperCell, Grid, SparseCSR
from sisl import AtomicOrbital
from sisl.sparse import _ncol_to_indptr
from sisl.unit.siesta import unit_convert
from sisl.physics.sparse import SparseOrbitalBZ
from sisl.physics import Hamiltonian, DensityMatrix, EnergyDensityMatrix
from sisl.physics import BrillouinZone
from sisl.physics.overlap import Overlap
from sisl.physics.electron import EigenstateElectron
__all__ = ['tshsSileSiesta', 'onlysSileSiesta', 'tsdeSileSiesta']
__all__ += ['hsxSileSiesta', 'dmSileSiesta']
__all__ += ['wfsxSileSiesta']
__all__ += ['gridSileSiesta']
__all__ += ['tsgfSileSiesta']
_Bohr2Ang = unit_convert('Bohr', 'Ang')
_Ry2eV = unit_convert('Ry', 'eV')
_eV2Ry = unit_convert('eV', 'Ry')
def _toF(array, dtype, scale=None):
if scale is None:
return array.astype(dtype, order='F', copy=False)
elif array.dtype == dtype and array.flags.f_contiguous:
# no need to copy since the order is correct
return array * scale
# We have to handle cases
out = np.empty_like(array, dtype, order='F')
np.multiply(array, scale, out=out)
return out
def _geometry_align(geom_b, geom_u, cls, method):
""" Routine used to align two geometries
There are a few twists in this since the fdf-reads will automatically
try and pass a geometry from the output files.
In cases where the *.ion* files are non-existing this will
result in a twist.
This routine will select and return a merged Geometry which
fulfills the correct number of atoms and orbitals.
However, if the input geometries have mis-matching number
of atoms a SislError will be raised.
Parameters
----------
geom_b : Geometry from binary file
geom_u : Geometry supplied by user
Raises
------
SislError
if the geometries have non-equal atom count
"""
if geom_b is None:
return geom_u
elif geom_u is None:
return geom_b
# Default to use the users geometry
geom = geom_u
is_copy = False
def get_copy(geom, is_copy):
if is_copy:
return geom, True
return geom.copy(), True
if geom_b.na != geom.na:
# we have no way of solving this issue...
raise SileError(f"{cls.__name__}.{method} could not use the passed geometry as the "
f"of atoms is not consistent, user-atoms={geom_u.na}, file-atoms={geom_b.na}.")
# Try and figure out what to do
if not np.allclose(geom_b.xyz, geom.xyz):
warn(f"{cls.__name__}.{method} has mismatched atomic coordinates, will copy geometry and use file XYZ.")
geom, is_copy = get_copy(geom, is_copy)
geom.xyz[:, :] = geom_b.xyz[:, :]
if not np.allclose(geom_b.sc.cell, geom.sc.cell):
warn(f"{cls.__name__}.{method} has non-equal lattice vectors, will copy geometry and use file lattice.")
geom, is_copy = get_copy(geom, is_copy)
geom.sc.cell[:, :] = geom_b.sc.cell[:, :]
if not np.array_equal(geom_b.nsc, geom.nsc):
warn(f"{cls.__name__}.{method} has non-equal number of supercells, will copy geometry and use file supercell count.")
geom, is_copy = get_copy(geom, is_copy)
geom.set_nsc(geom_b.nsc)
# Now for the difficult part.
# If there is a mismatch in the number of orbitals we will
# prefer to use the user-supplied atomic species, but fill with
# *random* orbitals
if not np.array_equal(geom_b.atoms.orbitals, geom.atoms.orbitals):
warn(f"{cls.__name__}.{method} has non-equal number of orbitals per atom, will correct with *empty* orbitals.")
geom, is_copy = get_copy(geom, is_copy)
# Now create a new atom specie with the correct number of orbitals
norbs = geom_b.atoms.orbitals[:]
atoms = Atoms([geom.atoms[i].copy(orbitals=[-1.] * norbs[i]) for i in range(geom.na)])
geom._atoms = atoms
return geom
@set_module("sisl.io.siesta")
class onlysSileSiesta(SileBinSiesta):
""" Geometry and overlap matrix """
[docs] def read_supercell(self):
""" Returns a SuperCell object from a TranSiesta file """
n_s = _siesta.read_tshs_sizes(self.file)[3]
self._fortran_check('read_supercell', 'could not read sizes.')
arr = _siesta.read_tshs_cell(self.file, n_s)
self._fortran_check('read_supercell', 'could not read cell.')
nsc = np.array(arr[0], np.int32)
# We have to transpose since the data is read *as-is*
# The cell in fortran files are (:, A1)
# after reading this is still obeyed (regardless of order)
# So we transpose to get it C-like
# Note that care must be taken for the different data-structures
# In particular not all data needs to be transposed (sparse H and S)
cell = arr[1].T * _Bohr2Ang
return SuperCell(cell, nsc=nsc)
[docs] def read_geometry(self, geometry=None):
""" Returns Geometry object from a TranSiesta file """
# Read supercell
sc = self.read_supercell()
na = _siesta.read_tshs_sizes(self.file)[1]
self._fortran_check('read_geometry', 'could not read sizes.')
arr = _siesta.read_tshs_geom(self.file, na)
self._fortran_check('read_geometry', 'could not read geometry.')
# see onlysSileSiesta.read_supercell for .T
xyz = arr[0].T * _Bohr2Ang
lasto = arr[1]
# Since the TSHS file does not contain species information
# and/or other stuff we *can* reuse an existing
# geometry which contains the correct atomic numbers etc.
orbs = np.diff(lasto)
if geometry is None:
# Create all different atoms...
# The TSHS file does not contain the
# atomic numbers, so we will just
# create them individually
# Get unique orbitals
uorb = np.unique(orbs)
# Create atoms
atoms = []
for Z, orb in enumerate(uorb):
atoms.append(Atom(Z+1, [-1] * orb))
def get_atom(atoms, orbs):
for atom in atoms:
if atom.no == orbs:
return atom
atom = []
for orb in orbs:
atom.append(get_atom(atoms, orb))
else:
# Create a new geometry with the correct atomic numbers
atom = []
for ia, no in zip(geometry, orbs):
a = geometry.atoms[ia]
if a.no == no:
atom.append(a)
else:
# correct atom
atom.append(a.__class__(a.Z, [-1. for io in range(no)], mass=a.mass, tag=a.tag))
# Create and return geometry object
return Geometry(xyz, atom, sc=sc)
[docs] def read_overlap(self, **kwargs):
""" Returns the overlap matrix from the TranSiesta file """
tshs_g = self.read_geometry()
geom = _geometry_align(tshs_g, kwargs.get('geometry', tshs_g), self.__class__, 'read_overlap')
# read the sizes used...
sizes = _siesta.read_tshs_sizes(self.file)
self._fortran_check('read_overlap', 'could not read sizes.')
# see onlysSileSiesta.read_supercell for .T
isc = _siesta.read_tshs_cell(self.file, sizes[3])[2].T
self._fortran_check('read_overlap', 'could not read cell.')
no = sizes[2]
nnz = sizes[4]
ncol, col, dS = _siesta.read_tshs_s(self.file, no, nnz)
self._fortran_check('read_overlap', 'could not read overlap matrix.')
# Create the Hamiltonian container
S = Overlap(geom, nnzpr=1)
# Create the new sparse matrix
S._csr.ncol = ncol.astype(np.int32, copy=False)
S._csr.ptr = _ncol_to_indptr(ncol)
# Correct fortran indices
S._csr.col = col.astype(np.int32, copy=False) - 1
S._csr._nnz = len(col)
S._csr._D = _a.emptyd([nnz, 1])
S._csr._D[:, 0] = dS[:]
# Convert to sisl supercell
# equivalent as _csr_from_siesta with explicit isc from file
_csr_from_sc_off(S.geometry, isc, S._csr)
# In siesta the matrix layout is written in CSC format
# due to fortran indexing, this means that we need to transpose
# to get it to correct layout.
return S.transpose(sort=kwargs.get("sort", True))
[docs] def read_fermi_level(self):
r""" Query the Fermi-level contained in the file
Returns
-------
Ef : fermi-level of the system
"""
Ef = _siesta.read_tshs_ef(self.file) * _Ry2eV
self._fortran_check('read_fermi_level', 'could not read fermi-level.')
return Ef
@set_module("sisl.io.siesta")
class tshsSileSiesta(onlysSileSiesta):
""" Geometry, Hamiltonian and overlap matrix file """
[docs] def read_hamiltonian(self, **kwargs):
""" Returns the electronic structure from the siesta.TSHS file """
tshs_g = self.read_geometry()
geom = _geometry_align(tshs_g, kwargs.get('geometry', tshs_g), self.__class__, 'read_hamiltonian')
# read the sizes used...
sizes = _siesta.read_tshs_sizes(self.file)
self._fortran_check('read_hamiltonian', 'could not read sizes.')
# see onlysSileSiesta.read_supercell for .T
isc = _siesta.read_tshs_cell(self.file, sizes[3])[2].T
self._fortran_check('read_hamiltonian', 'could not read cell.')
spin = sizes[0]
no = sizes[2]
nnz = sizes[4]
ncol, col, dH, dS = _siesta.read_tshs_hs(self.file, spin, no, nnz)
self._fortran_check('read_hamiltonian', 'could not read Hamiltonian and overlap matrix.')
# Check whether it is an orthogonal basis set
orthogonal = np.abs(dS).sum() == geom.no
# Create the Hamiltonian container
H = Hamiltonian(geom, spin, nnzpr=1, orthogonal=orthogonal)
# Create the new sparse matrix
H._csr.ncol = ncol.astype(np.int32, copy=False)
H._csr.ptr = _ncol_to_indptr(ncol)
# Correct fortran indices
H._csr.col = col.astype(np.int32, copy=False) - 1
H._csr._nnz = len(col)
if orthogonal:
H._csr._D = _a.emptyd([nnz, spin])
H._csr._D[:, :] = dH[:, :] * _Ry2eV
else:
H._csr._D = _a.emptyd([nnz, spin+1])
H._csr._D[:, :spin] = dH[:, :] * _Ry2eV
H._csr._D[:, spin] = dS[:]
_mat_spin_convert(H)
# Convert to sisl supercell
# equivalent as _csr_from_siesta with explicit isc from file
_csr_from_sc_off(H.geometry, isc, H._csr)
# Find all indices where dS == 1 (remember col is in fortran indices)
idx = col[np.isclose(dS, 1.).nonzero()[0]]
if np.any(idx > no):
print(f'Number of orbitals: {no}')
print(idx)
raise SileError(f'{self!s}.read_hamiltonian could not assert '
'the supercell connections in the primary unit-cell.')
# see onlysSileSiesta.read_overlap for .transpose()
# For H, DM and EDM we also need to Hermitian conjugate it.
return H.transpose(spin=False, sort=kwargs.get("sort", True))
[docs] def write_hamiltonian(self, H, **kwargs):
""" Writes the Hamiltonian to a siesta.TSHS file """
# we sort below, so no need to do it here
# see onlysSileSiesta.read_overlap for .transpose()
csr = H.transpose(spin=False, sort=False)._csr
if csr.nnz == 0:
raise SileError(f'{self!s}.write_hamiltonian cannot write '
'a zero element sparse matrix!')
# Convert to siesta CSR
_csr_to_siesta(H.geometry, csr)
csr.finalize(sort=kwargs.get("sort", True))
_mat_spin_convert(csr, H.spin)
# Extract the data to pass to the fortran routine
cell = H.geometry.cell
xyz = H.geometry.xyz
# Get H and S
if H.orthogonal:
h = csr._D
s = csr.diags(1., dim=1)
# Ensure all data is correctly formatted (i.e. have the same sparsity pattern)
s.align(csr)
s.finalize(sort=kwargs.get("sort", True))
if s.nnz != len(h):
raise SislError('The diagonal elements of your orthogonal Hamiltonian '
'have not been defined, this is a requirement.')
s = s._D[:, 0]
else:
h = csr._D[:, :H.S_idx]
s = csr._D[:, H.S_idx]
# Get shorter variants
nsc = H.geometry.nsc[:].astype(np.int32)
isc = _siesta.siesta_sc_off(*nsc)
# see onlysSileSiesta.read_supercell for .T
_siesta.write_tshs_hs(self.file, nsc[0], nsc[1], nsc[2],
cell.T / _Bohr2Ang, xyz.T / _Bohr2Ang, H.geometry.firsto,
csr.ncol, csr.col + 1,
_toF(h, np.float64, _eV2Ry), _toF(s, np.float64),
isc)
self._fortran_check('write_hamiltonian', 'could not write Hamiltonian and overlap matrix.')
@set_module("sisl.io.siesta")
class dmSileSiesta(SileBinSiesta):
""" Density matrix file """
[docs] def read_density_matrix(self, **kwargs):
""" Returns the density matrix from the siesta.DM file """
# Now read the sizes used...
spin, no, nsc, nnz = _siesta.read_dm_sizes(self.file)
self._fortran_check('read_density_matrix', 'could not read density matrix sizes.')
ncol, col, dDM = _siesta.read_dm(self.file, spin, no, nsc, nnz)
self._fortran_check('read_density_matrix', 'could not read density matrix.')
# Try and immediately attach a geometry
geom = kwargs.get('geometry', kwargs.get('geom', None))
if geom is None:
# We truly, have no clue,
# Just generate a boxed system
xyz = [[x, 0, 0] for x in range(no)]
sc = SuperCell([no, 1, 1], nsc=nsc)
geom = Geometry(xyz, Atom(1), sc=sc)
if nsc[0] != 0 and np.any(geom.nsc != nsc):
# We have to update the number of supercells!
geom.set_nsc(nsc)
if geom.no != no:
raise SileError(f'{self!s}.read_density_matrix could not use the '
'passed geometry as the number of atoms or orbitals is '
'inconsistent with DM file.')
# Create the density matrix container
DM = DensityMatrix(geom, spin, nnzpr=1, dtype=np.float64, orthogonal=False)
# Create the new sparse matrix
DM._csr.ncol = ncol.astype(np.int32, copy=False)
DM._csr.ptr = _ncol_to_indptr(ncol)
# Correct fortran indices
DM._csr.col = col.astype(np.int32, copy=False) - 1
DM._csr._nnz = len(col)
DM._csr._D = _a.emptyd([nnz, spin+1])
DM._csr._D[:, :spin] = dDM[:, :]
# DM file does not contain overlap matrix... so neglect it for now.
DM._csr._D[:, spin] = 0.
_mat_spin_convert(DM)
# Convert the supercells to sisl supercells
if nsc[0] != 0 or geom.no_s >= col.max():
_csr_from_siesta(geom, DM._csr)
else:
warn(f'{self!s}.read_density_matrix may result in a wrong sparse pattern!')
return DM.transpose(spin=False, sort=kwargs.get("sort", True))
[docs] def write_density_matrix(self, DM, **kwargs):
""" Writes the density matrix to a siesta.DM file """
csr = DM.transpose(spin=False, sort=False)._csr
# This ensures that we don't have any *empty* elements
if csr.nnz == 0:
raise SileError(f'{self!s}.write_density_matrix cannot write '
'a zero element sparse matrix!')
_csr_to_siesta(DM.geometry, csr)
# We do not really need to sort this one, but we do for consistency
# of the interface.
csr.finalize(sort=kwargs.get("sort", True))
_mat_spin_convert(csr, DM.spin)
# Get DM
if DM.orthogonal:
dm = csr._D
else:
dm = csr._D[:, :DM.S_idx]
# Ensure shapes (say if only 1 spin)
dm.shape = (-1, len(DM.spin))
nsc = DM.geometry.sc.nsc.astype(np.int32)
_siesta.write_dm(self.file, nsc, csr.ncol, csr.col + 1, _toF(dm, np.float64))
self._fortran_check('write_density_matrix', 'could not write density matrix.')
@set_module("sisl.io.siesta")
class tsdeSileSiesta(dmSileSiesta):
""" Non-equilibrium density matrix and energy density matrix file """
[docs] def read_energy_density_matrix(self, **kwargs):
""" Returns the energy density matrix from the siesta.DM file """
# Now read the sizes used...
spin, no, nsc, nnz = _siesta.read_tsde_sizes(self.file)
self._fortran_check('read_energy_density_matrix', 'could not read energy density matrix sizes.')
ncol, col, dEDM = _siesta.read_tsde_edm(self.file, spin, no, nsc, nnz)
self._fortran_check('read_energy_density_matrix', 'could not read energy density matrix.')
# Try and immediately attach a geometry
geom = kwargs.get('geometry', kwargs.get('geom', None))
if geom is None:
# We truly, have no clue,
# Just generate a boxed system
xyz = [[x, 0, 0] for x in range(no)]
sc = SuperCell([no, 1, 1], nsc=nsc)
geom = Geometry(xyz, Atom(1), sc=sc)
if nsc[0] != 0 and np.any(geom.nsc != nsc):
# We have to update the number of supercells!
geom.set_nsc(nsc)
if geom.no != no:
raise SileError(f'{self!s}.read_energy_density_matrix could '
'not use the passed geometry as the number of atoms or orbitals '
'is inconsistent with DM file.')
# Create the energy density matrix container
EDM = EnergyDensityMatrix(geom, spin, nnzpr=1, dtype=np.float64, orthogonal=False)
# Create the new sparse matrix
EDM._csr.ncol = ncol.astype(np.int32, copy=False)
EDM._csr.ptr = _ncol_to_indptr(ncol)
# Correct fortran indices
EDM._csr.col = col.astype(np.int32, copy=False) - 1
EDM._csr._nnz = len(col)
EDM._csr._D = _a.emptyd([nnz, spin+1])
EDM._csr._D[:, :spin] = dEDM[:, :] * _Ry2eV
# EDM file does not contain overlap matrix... so neglect it for now.
EDM._csr._D[:, spin] = 0.
_mat_spin_convert(EDM)
# Convert the supercells to sisl supercells
if nsc[0] != 0 or geom.no_s >= col.max():
_csr_from_siesta(geom, EDM._csr)
else:
warn(f'{self!s}.read_energy_density_matrix may result in a wrong sparse pattern!')
return EDM.transpose(spin=False, sort=kwargs.get("sort", True))
[docs] def read_fermi_level(self):
r""" Query the Fermi-level contained in the file
Returns
-------
Ef : fermi-level of the system
"""
Ef = _siesta.read_tsde_ef(self.file) * _Ry2eV
self._fortran_check('read_fermi_level', 'could not read fermi-level.')
return Ef
[docs] def write_density_matrices(self, DM, EDM, Ef=0., **kwargs):
r""" Writes the density matrix to a siesta.DM file
Parameters
----------
DM : DensityMatrix
density matrix to write to the file
EDM : EnergyDensityMatrix
energy density matrix to write to the file
Ef : float, optional
fermi-level to be contained
"""
DMcsr = DM.transpose(spin=False, sort=False)._csr
EDMcsr = EDM.transpose(spin=False, sort=False)._csr
DMcsr.align(EDMcsr)
EDMcsr.align(DMcsr)
if DMcsr.nnz == 0:
raise SileError(f'{self!s}.write_density_matrices cannot write '
'a zero element sparse matrix!')
_csr_to_siesta(DM.geometry, DMcsr)
_csr_to_siesta(DM.geometry, EDMcsr)
sort = kwargs.get("sort", True)
DMcsr.finalize(sort=sort)
EDMcsr.finalize(sort=sort)
_mat_spin_convert(DMcsr, DM.spin)
_mat_spin_convert(EDMcsr, EDM.spin)
# Ensure everything is correct
if not (np.allclose(DMcsr.ncol, EDMcsr.ncol) and
np.allclose(DMcsr.col, EDMcsr.col)):
raise ValueError(f'{self!s}.write_density_matrices got non compatible '
'DM and EDM matrices.')
if DM.orthogonal:
dm = DMcsr._D
else:
dm = DMcsr._D[:, :DM.S_idx]
if EDM.orthogonal:
edm = EDMcsr._D
else:
edm = EDMcsr._D[:, :EDM.S_idx]
nsc = DM.geometry.sc.nsc.astype(np.int32)
_siesta.write_tsde_dm_edm(self.file, nsc, DMcsr.ncol, DMcsr.col + 1,
_toF(dm, np.float64),
_toF(edm, np.float64, _eV2Ry), Ef * _eV2Ry)
self._fortran_check('write_density_matrices', 'could not write DM + EDM matrices.')
@set_module("sisl.io.siesta")
class hsxSileSiesta(SileBinSiesta):
""" Hamiltonian and overlap matrix file
This file does not contain all information regarding the system.
To ensure no errors are being raised one should pass a `Geometry` with
correct number of atoms and correct number of supercells.
The number of orbitals will be updated in the returned matrices geometry.
>>> hsx = hsxSileSiesta("siesta.HSX")
>>> HS = hsx.read_hamiltonian() # may fail
>>> HS = hsx.read_hamiltonian(geometry=<>) # should run correctly if above satisfied
Users are adviced to use the `tshsSileSiesta` instead since that correctly contains
all information.
"""
def _xij2system(self, xij, geometry=None):
""" Create a new geometry with *correct* nsc and somewhat correct xyz
Parameters
----------
xij : SparseCSR
orbital distances
geometry : Geometry, optional
passed geometry
"""
def get_geom_handle(xij):
atoms = self._read_atoms()
if not atoms is None:
return Geometry(np.zeros([len(atoms), 3]), atoms)
N = len(xij)
# convert csr to dok format
row = (xij.ncol > 0).nonzero()[0]
# Now we have [0 0 0 0 1 1 1 1 2 2 ... no-1 no-1]
row = np.repeat(row, xij.ncol[row])
col = xij.col
# Parse xij to correct geometry
# first figure out all zeros (i.e. self-atom-orbitals)
idx0 = np.isclose(np.fabs(xij._D).sum(axis=1), 0.).nonzero()[0]
row0 = row[idx0]
# convert row0 and col0 to a first attempt of "atomization"
atoms = []
for r in range(N):
idx0r = (row0 == r).nonzero()[0]
row0r = row0[idx0r]
# although xij == 0, we just do % to ensure unit-cell orbs
col0r = col[idx0[idx0r]] % N
if np.all(col0r >= r):
# we have a new atom
atoms.append(set(col0r))
else:
atoms[-1].update(set(col0r))
# convert list of orbitals to lists
def conv(a):
a = list(a)
a.sort()
return a
atoms = [list(a) for a in atoms]
if sum(map(len, atoms)) != len(xij):
raise ValueError(f"{self.__class__.__name__} could not determine correct "
"number of orbitals.")
atms = Atoms(Atom('H', [-1. for _ in atoms[0]]))
for orbs in atoms[1:]:
atms.append(Atom('H', [-1. for _ in orbs]))
return Geometry(np.zeros([len(atoms), 3]), atms)
geom_handle = get_geom_handle(xij)
def convert_to_atom(geom, xij):
# o2a does not check for correct super-cell index
n_s = xij.shape[1] // xij.shape[0]
atm_s = geom.o2a(np.arange(xij.shape[1]))
# convert csr to dok format
row = (xij.ncol > 0).nonzero()[0]
row = np.repeat(row, xij.ncol[row])
col = xij.col
arow = atm_s[row]
acol = atm_s[col]
del atm_s, row, col
idx = np.lexsort((acol, arow))
arow = arow[idx]
acol = acol[idx]
xij = xij._D[idx]
del idx
# Now figure out if xij is consistent
duplicates = np.logical_and(np.diff(acol) == 0,
np.diff(arow) == 0).nonzero()[0]
if duplicates.size > 0:
if not np.allclose(xij[duplicates+1] - xij[duplicates], 0.):
raise ValueError(f"{self.__class__.__name__} converting xij(orb) -> xij(atom) went wrong. "
"This may happen if your coordinates are not inside the unitcell, please pass "
"a usable geometry.")
# remove duplicates to create new matrix
arow = np.delete(arow, duplicates)
acol = np.delete(acol, duplicates)
xij = np.delete(xij, duplicates, axis=0)
# Create a new sparse matrix
# Create the new index pointer
indptr = np.insert(np.array([0, len(xij)], np.int32), 1,
(np.diff(arow) != 0).nonzero()[0] + 1)
assert len(indptr) == geom.na + 1
return SparseCSR((xij, acol, indptr), shape=(geom.na, geom.na * n_s))
def coord_from_xij(xij):
# first atom is at 0, 0, 0
na = len(xij)
xyz = _a.zerosd([na, 3])
xyz[0] = [0, 0, 0]
mark = _a.zerosi(na)
mark[0] = 1
run_atoms = deque([0])
while len(run_atoms) > 0:
atm = run_atoms.popleft()
xyz_atm = xyz[atm].reshape(1, 3)
neighbours = xij.edges(atm, exclude=atm)
neighbours = neighbours[neighbours < na]
# update those that haven't been calculated
idx = mark[neighbours] == 0
neigh_idx = neighbours[idx]
if len(neigh_idx) == 0:
continue
xyz[neigh_idx, :] = xij[atm, neigh_idx] - xyz_atm
mark[neigh_idx] = 1
# add more atoms to be processed, since we have *mark*
# we will only run every atom once
run_atoms.extend(neigh_idx.tolist())
# check that everything is correct
if (~idx).sum() > 0:
neg_neighbours = neighbours[~idx]
if not np.allclose(xyz[neg_neighbours, :],
xij[atm, neg_neighbours] - xyz_atm):
raise ValueError(f"{self.__class__.__name__} xij(orb) -> xyz did not "
f"find same coordinates for different connections")
if mark.sum() != na:
raise ValueError(f"{self.__class__.__name__} xij(orb) -> Geometry does not "
f"have a fully connected geometry. It is impossible to create relative coordinates")
return xyz
def sc_from_xij(xij, xyz):
na = xij.shape[0]
n_s = xij.shape[1] // xij.shape[0]
if n_s == 1:
# easy!!
return SuperCell(xyz.max(axis=0) - xyz.min(axis=0) + 10., nsc=[1] * 3)
sc_off = _a.zerosd([n_s, 3])
mark = _a.zerosi(n_s)
mark[0] = 1
for atm in range(na):
neighbours = xij.edges(atm, exclude=atm)
uneighbours = neighbours % na
neighbour_isc = neighbours // na
# get offset in terms of unit-cell
off = xij[atm, neighbours] - (xyz[uneighbours] - xyz[atm].reshape(1, 3))
idx = mark[neighbour_isc] == 0
if not np.allclose(off[~idx], sc_off[neighbour_isc[~idx]]):
raise ValueError(f"{self.__class__.__name__} xij(orb) -> xyz did not "
f"find same supercell offsets for different connections")
if idx.sum() == 0:
continue
for idx in idx.nonzero()[0]:
nidx = neighbour_isc[idx]
if mark[nidx] == 0:
mark[nidx] = 1
sc_off[nidx] = off[idx]
elif not np.allclose(sc_off[nidx], off[idx]):
raise ValueError(f"{self.__class__.__name__} xij(orb) -> xyz did not "
f"find same supercell offsets for different connections")
# We know that siesta returns isc
# for iz in [0, 1, 2, 3, -3, -2, -1]:
# for iy in [0, 1, 2, -2, -1]:
# for ix in [0, 1, -1]:
# every block we find a half monotonically increasing vector additions
# Note the first is always [0, 0, 0]
# So our best chance is to *guess* the first nsc
# then reshape, then guess, then reshape, then guess :)
sc_diff = np.diff(sc_off, axis=0)
def get_nsc(sc_off):
""" Determine nsc depending on the axis """
# correct the offsets
ndim = sc_off.ndim
if sc_off.shape[0] == 1:
return 1
# always select the 2nd one since that contains the offset
# for the first isc [1, 0, 0] or [0, 1, 0] or [0, 0, 1]
sc_dir = sc_off[(1, ) + np.index_exp[0] * (ndim - 2)].reshape(1, 3)
norm2_sc_dir = (sc_dir ** 2).sum()
# figure out the maximum integer part
# we select 0 indices for all already determined lattice
# vectors since we know the first one is [0, 0, 0]
idx = np.index_exp[:] + np.index_exp[0] * (ndim - 2)
projection = (sc_off[idx] * sc_dir).sum(-1) / norm2_sc_dir
iprojection = np.rint(projection)
# reduce, find 0
idx_zero = np.isclose(iprojection, 0, atol=1e-5).nonzero()[0]
if idx_zero.size <= 1:
return 1
# only take those values that are continuous
# we *must* have some supercell connections
idx_max = idx_zero[1]
# find where they are close
# since there may be *many* zeros (non-coupling elements)
# we first have to cut off anything that is not integer
if np.allclose(projection[:idx_max], iprojection[:idx_max], atol=1e-5):
return idx_max
raise ValueError(f"Could not determine nsc from coordinates")
nsc = _a.onesi(3)
nsc[0] = get_nsc(sc_off)
sc_off = sc_off.reshape(-1, nsc[0], 3)
nsc[1] = get_nsc(sc_off)
sc_off = sc_off.reshape(-1, nsc[1], nsc[0], 3)
nsc[2] = sc_off.shape[0]
# now determine cell parameters
if all(nsc > 1):
cell = _a.arrayd([sc_off[0, 0, 1],
sc_off[0, 1, 0],
sc_off[1, 0, 0]])
else:
# we will never have all(nsc == 1) since that is
# taken care of at the start
# this gets a bit tricky, since we don't know one of the
# lattice vectors
cell = _a.zerosd([3, 3])
i = 0
for idx, isc in enumerate(nsc):
if isc > 1:
sl = [0, 0, 0]
sl[2 - idx] = 1
cell[i] = sc_off[tuple(sl)]
i += 1
# figure out the last vectors
# We'll just use Cartesian coordinates
while i < 3:
# this means we don't have any supercell connections
# along at least 1 other lattice vector.
lcell = np.fabs(cell).sum(0)
# figure out which Cartesian direction we are *missing*
cart_dir = np.argmin(lcell)
cell[i, cart_dir] = xyz[:, cart_dir].max() - xyz[:, cart_dir].min() + 10.
i += 1
return SuperCell(cell, nsc)
# now we have all orbitals, ensure compatibility with passed geometry
if geometry is None:
atm_xij = convert_to_atom(geom_handle, xij)
xyz = coord_from_xij(atm_xij)
sc = sc_from_xij(atm_xij, xyz)
geometry = Geometry(xyz, geom_handle.atoms, sc)
# Move coordinates into unit-cell
geometry.xyz[:, :] = (geometry.fxyz % 1.) @ geometry.cell
else:
if geometry.n_s != xij.shape[1] // xij.shape[0]:
atm_xij = convert_to_atom(geom_handle, xij)
sc = sc_from_xij(atm_xij, coord_from_xij(atm_xij))
geometry.set_nsc(sc.nsc)
def conv(orbs, atm):
if len(orbs) == len(atm):
return atm
return atm.copy(orbitals=[-1. for _ in orbs])
atms = Atoms(list(map(conv, geom_handle.atoms, geometry.atoms)))
if len(atms) != len(geometry):
raise ValueError(f"{self.__class__.__name__} passed geometry for reading "
"sparse matrix does not contain same number of atoms!")
geometry = geometry.copy()
# TODO check that geometry and xyz are the same!
geometry._atoms = atms
return geometry
def _read_atoms(self, **kwargs):
""" Reads basis set and geometry information from the HSX file """
# Now read the sizes used...
no, na, nspecies = _siesta.read_hsx_specie_sizes(self.file)
self._fortran_check('read_geometry', 'could not read specie sizes.')
# Read specie information
labels, val_q, norbs, isa = _siesta.read_hsx_species(self.file, nspecies, no, na)
# convert to proper string
labels = labels.T.reshape(nspecies, -1)
labels = labels.view(f"S{labels.shape[1]}")
labels = list(map(lambda s: b''.join(s).decode('utf-8').strip(),
labels.tolist())
)
self._fortran_check('read_geometry', 'could not read species.')
# to python index
isa -= 1
from sisl.atom import _ptbl
# try and convert labels into symbols
# We do this by:
# 1. label -> symbol
# 2. label[:2] -> symbol
# 3. label[:1] -> symbol
symbols = []
lbls = []
for label in labels:
lbls.append(label)
try:
symbol = _ptbl.Z_label(label)
symbols.append(symbol)
continue
except:
pass
try:
symbol = _ptbl.Z_label(label[:2])
symbols.append(symbol)
continue
except:
pass
try:
symbol = _ptbl.Z_label(label[:1])
symbols.append(symbol)
continue
except:
# we have no clue, assign -1
symbols.append(-1)
# Read in orbital information
atoms = []
for ispecie in range(nspecies):
n_l_zeta = _siesta.read_hsx_specie(self.file, ispecie+1, norbs[ispecie])
self._fortran_check('read_geometry', f'could not read specie {ispecie}.')
# create orbital
# no shell will have l>5, so m=10 should be more than enough
m = 10
orbs = []
for n, l, zeta in zip(*n_l_zeta):
# manual loop on m quantum numbers
if m > l:
m = -l
orbs.append(AtomicOrbital(n=n, l=l, m=m, zeta=zeta, R=-1.))
m += 1
# now create atom
atoms.append(Atom(symbols[ispecie], orbs, tag=lbls[ispecie]))
# now read in xij to retrieve atomic positions
Gamma, spin, no, no_s, nnz = _siesta.read_hsx_sizes(self.file)
self._fortran_check('read_geometry', 'could not read matrix sizes.')
ncol, col, _, dxij = _siesta.read_hsx_sx(self.file, Gamma, spin, no, no_s, nnz)
dxij = dxij.T * _Bohr2Ang
col -= 1
self._fortran_check('read_geometry', 'could not read xij matrix.')
# now create atoms object
atoms = Atoms([atoms[ia] for ia in isa])
return atoms
[docs] def read_hamiltonian(self, **kwargs):
""" Returns the electronic structure from the siesta.TSHS file """
# Now read the sizes used...
Gamma, spin, no, no_s, nnz = _siesta.read_hsx_sizes(self.file)
self._fortran_check('read_hamiltonian', 'could not read Hamiltonian sizes.')
ncol, col, dH, dS, dxij = _siesta.read_hsx_hsx(self.file, Gamma, spin, no, no_s, nnz)
dxij = dxij.T * _Bohr2Ang
col -= 1
self._fortran_check('read_hamiltonian', 'could not read Hamiltonian.')
ptr = _ncol_to_indptr(ncol)
xij = SparseCSR((dxij, col, ptr), shape=(no, no_s))
geom = self._xij2system(xij, kwargs.get('geometry', kwargs.get('geom', None)))
if geom.no != no or geom.no_s != no_s:
raise SileError(f"{str(self)}.read_hamiltonian could not use the "
"passed geometry as the number of atoms or orbitals is "
"inconsistent with HSX file.")
# Create the Hamiltonian container
H = Hamiltonian(geom, spin, nnzpr=1, dtype=np.float32, orthogonal=False)
# Create the new sparse matrix
H._csr.ncol = ncol.astype(np.int32, copy=False)
H._csr.ptr = ptr
# Correct fortran indices
H._csr.col = col.astype(np.int32, copy=False)
H._csr._nnz = len(col)
H._csr._D = _a.emptyf([nnz, spin+1])
H._csr._D[:, :spin] = dH[:, :] * _Ry2eV
H._csr._D[:, spin] = dS[:]
_mat_spin_convert(H)
# Convert the supercells to sisl supercells
if no_s // no == np.product(geom.nsc):
_csr_from_siesta(geom, H._csr)
return H.transpose(spin=False, sort=kwargs.get("sort", True))
[docs] def read_overlap(self, **kwargs):
""" Returns the overlap matrix from the siesta.HSX file """
# Now read the sizes used...
Gamma, spin, no, no_s, nnz = _siesta.read_hsx_sizes(self.file)
self._fortran_check('read_overlap', 'could not read overlap matrix sizes.')
ncol, col, dS, dxij = _siesta.read_hsx_sx(self.file, Gamma, spin, no, no_s, nnz)
dxij = dxij.T * _Bohr2Ang
col -= 1
self._fortran_check('read_overlap', 'could not read overlap matrix.')
ptr = _ncol_to_indptr(ncol)
xij = SparseCSR((dxij, col, ptr), shape=(no, no_s))
geom = self._xij2system(xij, kwargs.get('geometry', kwargs.get('geom', None)))
if geom.no != no or geom.no_s != no_s:
raise SileError(f"{str(self)}.read_overlap could not use the "
"passed geometry as the number of atoms or orbitals is "
"inconsistent with HSX file.")
# Create the Hamiltonian container
S = Overlap(geom, nnzpr=1)
# Create the new sparse matrix
S._csr.ncol = ncol.astype(np.int32, copy=False)
S._csr.ptr = _ncol_to_indptr(ncol)
# Correct fortran indices
S._csr.col = col.astype(np.int32, copy=False)
S._csr._nnz = len(col)
S._csr._D = _a.emptyf([nnz, 1])
S._csr._D[:, 0] = dS[:]
# Convert the supercells to sisl supercells
if no_s // no == np.product(geom.nsc):
_csr_from_siesta(geom, S._csr)
# not really necessary with Hermitian transposing, but for consistency
return S.transpose(sort=kwargs.get("sort", True))
@set_module("sisl.io.siesta")
class wfsxSileSiesta(SileBinSiesta):
r""" Binary WFSX file reader for Siesta
The WFSX file assumes that users initialize the object with
a `parent` argument (or one of the other geometry related objects as
shown below).
The `parent` argument is necessary to convert WFSX k-points from 1/Ang to
reduced coordinates.
When returning `EigenstateElectron` objects the parent of these objects
are the equivalent of the `parent` argument upon initialization.
Therefore please remember to pass a correct `parent`.
Parameters
----------
parent : obj, optional
a parent may contain a geometry, and/or a supercell
geometry : Geometry, optional
a geometry contains a cell with corresponding lattice vectors
used to convert k [1/Ang] -> [b]
sc : SuperCell, optional
a supercell contains the lattice vectors to convert k
"""
def _setup(self, *args, **kwargs):
""" Simple setup that needs to be overwritten
All _read_next_* methods expect the fortran file unit to be handled
and that the position in the file is correct.
"""
super()._setup(*args, **kwargs)
# default sc
sc = None
# In case the instantiation was called with wfsxSileSiesta("path", geometry=geometry)
parent = kwargs.get("parent")
if parent is None:
geometry = None
elif isinstance(parent, Geometry):
geometry = parent
elif isinstance(parent, SuperCell):
sc = parent
else:
geometry = parent.geometry
geometry = kwargs.get("geometry", geometry)
if geometry is not None and sc is None:
sc = geometry.sc
sc = kwargs.get("sc", sc)
if sc is None and geometry is not None:
raise ValueError(f"{self.__class__.__name__}(geometry=Geometry, sc=None) is not an allowed argument combination.")
if parent is None:
parent = geometry
if parent is None:
parent = sc
self._parent = parent
self._geometry = geometry
self._sc = sc
if self._parent is None and self._geometry is None and self._sc is None:
def conv(k):
if not np.allclose(k, 0.):
warn(f"{self.__class__.__name__} cannot convert stored k-points from 1/Ang to reduced coordinates. Please ensure 'parent=Hamiltonian', 'geometry=Geometry', or 'sc=SuperCell' to ensure reduced k")
return k / _Bohr2Ang
else:
def conv(k):
return (k @ sc.cell.T) / (2 * pi * _Bohr2Ang)
self._convert_k = conv
def _open_wfsx(self, mode, rewind=False):
"""Open the file unit for the WFSX file.
Here we also initialize some variables to keep track of the state of the read.
"""
self._fortran_open(mode, rewind=rewind)
# Here we initialize the variables that will keep track of the state of the read.
# The process for identification is done on this basis:
# _ik is the current (Python) index for the k-point to be read
# _ispin is the current (Python) index for the spin-index to be read (only has meaning for a spin-polarized
# WFSX files)
# _state is:
# -1 : the file-descriptor has just been opened (i.e. in front of header)
# 0 : it means that the file-descriptor is in front of basis information
# 1 : it means that the file-descriptor is in front of k point information
# 2 : it means that the file-descriptor is in front of k point WFSX values
#
self._state = -1
self._ispin = 0
self._ik = 0
def _close_wfsx(self):
"""Close the file unit for the WFSX file.
We clean the variables used to keep track of read state.
"""
self._fortran_close()
# Clean variables
del self._state
del self._ik
del self._ispin
try:
del self._sizes
except:
pass
try:
del self._basis
except:
pass
try:
del self._funcs
except:
pass
def _setup_parsing(self, close=True, skip_basis=True):
"""Gets all the things needed to parse the wfsx file.
Parameters
-----------
close: bool, optional
Whether the file unit should be closed afterwards.
skip_basis : bool, optional
whether to also read the basis or not
"""
self._open_wfsx('r')
# Read the sizes relevant to the file.
# We also read whether there's only gamma point information or there are multiple points
self._sizes = self._read_next_sizes(skip_basis=skip_basis)
if not skip_basis:
self._basis = self._read_next_basis()
# Get the functions that should be used to parse state values.
if self._sizes.nspin in (4, 8):
# We will have twice as many coefficients.
func_index = 4
elif self._sizes.Gamma:
# State values will be in double precision floats
func_index = 1
else:
# State values will be in double precision complex
func_index = 2
Funcs = namedtuple("WFSXReads", ["read_index", "read_next"])
self._funcs = Funcs(
getattr(_siesta, f"read_wfsx_index_{func_index}"),
getattr(_siesta, f"read_wfsx_next_{func_index}")
)
if close:
self._close_wfsx()
def _read_next_sizes(self, skip_basis=False):
"""Reads the sizes if they are the next thing to be read.
Parameters
-----------
skip_basis: boolean, optional
Whether this method should also skip over the basis information.
Returns
-------
namedtuple :
- 'nspin': int. Number of spin components.
- 'no_u': int. Number of orbitals in the unit cell.
- 'nk': int. Number of k points in the file.
- 'Gamma': bool. Whether the file contains only the gamma point.
"""
# Check that we are in the right position in the file
if self._state != -1:
raise SileError(f"We are not in a position to read the sizes. State is: {self._state}")
# Read the sizes that we can find in the WFSX file
Sizes = namedtuple("Sizes", ["nspin", "no_u", "nk", "Gamma"])
sizes = _siesta.read_wfsx_next_sizes(self._iu, skip_basis)
# Inform that we are now in front of k point information
self._state = 1 if skip_basis else 0
# Check that everything went fine
self._fortran_check('_read_sizes', "could not read sizes")
return Sizes(*sizes)
def _read_next_basis(self):
"""Reads the basis if it is the next thing to be read.
Returns
-------
Atoms:
the basis read.
"""
# Check that we are in the right position in the file
if self._state != 0:
raise SileError(f"We are not in a position to read the basis. State is: {self._state}")
# Read the basis information that we can find in the WFSX file
basis_info = _siesta.read_wfsx_next_basis(self._iu, self._sizes.no_u)
# Inform that we are now in front of k point information
self._state = 1
# Check that everything went fine
self._fortran_check('_read_basis', "could not read basis information")
# Convert the information to a dict so that code is easier to follow.
basis_info = dict(zip(('atom_indices', 'atom_labels', 'orb_index_atom', 'orb_n', 'orb_symmetry'), basis_info))
# Sanitize the string information
for char_key in ("atom_labels", "orb_symmetry"):
basis_info[char_key] = np.array(["".join(label).rstrip() for label in basis_info[char_key].astype(str)])
# Find out the unique atom indices
unique_ats = np.unique(basis_info["atom_indices"])
def _get_atom_object(at):
"""Given an atom index, generates an Atom object with all the information we have about it"""
atom_orbs = np.where(basis_info["atom_indices"] == at)[0]
at_label = basis_info["atom_labels"][atom_orbs[0]]
orbitals = [
AtomicOrbital(f"{n}{symmetry}")
for n, symmetry in zip(basis_info["orb_n"][atom_orbs], basis_info["orb_symmetry"][atom_orbs])
]
return Atom(at_label, orbitals=orbitals)
# Generate the Atoms oject.
return Atoms([_get_atom_object(at) for at in unique_ats])
def _read_next_info(self, ispin, ik):
"""Reads the next eigenstate information.
This function should only be called after reading the sizes
or reading the previous state's values.
Parameters
----------
ispin: integer
the (python) spin index of the next eigenstate.
ik: integer
the (python) k index of the next eigenstate.
Returns
-------
array of shape (3,):
The k point of the state.
float:
The weight of the k point.
int:
Number of wavefunctions that the state contains. It is needed by the
function that reads the eigenstates values.
"""
# Store the indices of the current state
self._ik = ik
self._ispin = ispin
# Check that we are in a position where we will read state information
if self._state != 1:
raise SileError(f"We are not in a position to read k point information. State is: {self._state}")
# Read the state information
file_ispin, file_ik, k, weight, nwf = _siesta.read_wfsx_next_info(self._iu)
# Inform that we are now in front of state values
self._state = 2
# Check that the read went fine
self._fortran_check('_read_next_info', f"could not read next eigenstate info [{ispin + 1}, {ik + 1}]")
# Check that the read indices match the indices that we were expecting.
if file_ispin != ispin + 1 or file_ik != ik + 1:
self._ik = file_ik - 1
self._ispin = file_ispin - 1
raise SileError(f"WFSX indices do not match the expected ones. Expected: [{ispin + 1}, {ik + 1}], found [{file_ispin}, {file_ik}]")
return k, weight, nwf
def _read_next_values(self, ispin, ik, nwf):
"""Reads the next eigenstate values.
This function should only be called after reading the state's information
Parameters
----------
ispin: integer
the (python) spin index of the next eigenstate.
ik: integer
the (python) k index of the next eigenstate.
nwf: integer
The number of wavefunctions that the next eigenstate contains.
Should have been obtained by reading the state's info with `_read_next_info`.
Returns
-------
array of shape (nwf,):
The indices of each wavefunction that the state contains.
array of shape (nwf,):
The eigenvalues (in eV) of each wavefunction that the state contains.
array of shape (norbitals, nwf):
The coefficients for each wavefunction that the state contains.
"""
# Check that we are in the right position in the file
if self._state != 2:
raise SileError(f"We are not in a position to read k point WFSX values. State is: {self._state}")
# Read the state values
idx, eig, state = self._funcs.read_next(self._iu, self._sizes.no_u, nwf)
# Inform that we are now in front of the next state info
self._state = 1
# Check that everything went fine
self._fortran_check('_read_next_values', f"could not read next eigenstate values [{ispin + 1}, {ik + 1}]")
return idx, eig, state
def _read_next_eigenstate(self, ispin, ik):
"""Reads the next eigenstate in the WFSX file.
Parameters
----------
ispin: integer
the (python) spin index of the next eigenstate.
ik: integer
the (python) k index of the next eigenstate.
Returns
--------
EigenstateElectron:
The next eigenstate.
"""
# Get the information of this eigenstate
k, weight, nwf = self._read_next_info(ispin, ik)
# Now that we have the information, we can read the values because
# we know the number of wavefunctions stored in the k point (`nwf`)
idx, eig, state = self._read_next_values(ispin, ik, nwf)
# Build the info dictionary for the eigenstate to know how it was calculated
# We include the spin index if needed.
info = dict(k=self._convert_k(k), weight=weight, gauge="r", index=idx - 1)
if self._sizes.nspin == 2:
info["spin"] = ispin
# `eig` is already in eV
# See onlysSileSiesta.read_supercell to understand why we transpose `state`
return EigenstateElectron(state.T, eig, parent=self._parent, **info)
[docs] def read_sizes(self):
"""Reads the sizes related to this WFSX file
Returns
-------
int : number of spin components
int : number of orbitals in the unit-cell
int : number of k-points
bool : True if the file only contains the Gamma-point
"""
self._open_wfsx("r")
sizes = self._read_next_sizes()
self._close_wfsx()
return sizes
[docs] def read_basis(self):
"""Reads the basis contained in the WFSX file.
The WFSX file only contains information about the atom labels, which atom
each orbital belongs to and the orbital quantum numbers. It is thus not
complete in every sense.
Returns
-------
Atoms:
the basis read.
"""
self._open_wfsx("r")
self._sizes = self._read_next_sizes(skip_basis=False)
basis = self._read_next_basis()
self._close_wfsx()
return basis
[docs] def yield_eigenstate(self):
r""" Iterates over the states in the WFSX file
Yields
------
EigenstateElectron
"""
# Open file and get parsing information
self._setup_parsing(close=False)
if self._sizes.nspin == 2:
itt_spin = range(2)
else:
itt_spin = range(1)
try:
# Iterate over all eigenstates in the WFSX file, yielding control to the caller at
# each iteration.
for ik, ispin in product(range(self._sizes.nk), itt_spin):
yield self._read_next_eigenstate(ispin, ik)
# We ran out of eigenstates
self._close_wfsx()
except GeneratorExit:
# The loop in which the generator was used has been broken.
self._close_wfsx()
[docs] def read_eigenstate(self, k=(0, 0, 0), spin=0, ktol=1e-4):
"""Reads a specific eigenstate from the file.
This method iterates over the states until it finds a match. Don't call
this method repeatedly. If you want to loop eigenstates, use `yield_eigenstate`.
Parameters
----------
k: array-like of shape (3,), optional
The k point of the state you want to find.
spin: integer, optional
The spin index of the state you want to find. Only meaningful for polarized
calculations.
ktol: float, optional
The threshold value for considering two k-points the same (i.e. to match
the query k point with the state's k point).
See Also
--------
yield_eigenstate
Returns
-------
EigenstateElectron or None:
If found, the state that was queried.
If not found, returns `None`. NOTE this may change to an exception in the future
"""
# Iterate over all eigenstates in the file
for state in self.yield_eigenstate():
if state.info.get("spin", 0) == spin and np.allclose(state.info['k'], k, atol=ktol):
# This is the state that the user requested
return state
return None
[docs] def read_info(self):
"""Reads the information for all the k points contained in the file
Returns
-------
k: array of shape (nk, 3)
k values of the k points contained in the file.
weight: array of shape (nk,)
weight of each k point
nwf: array of shape (nspin, nk)
number of wavefunctions that each kpoint(-spin) contains.
"""
# Open file and get parsing information
self._setup_parsing(close=False, skip_basis=True)
# Check if we are in the correct position in the file (we should be just after the header)
if self._state != 1:
raise ValueError(f"We are not in a position to read eigenstate info in the file. State: {self._state}")
# Read all the information. Parse here the k values obtained.
# Store the information that should be returned under `returns`.
if self._sizes.nspin == 2:
nspin = 2
else:
nspin = 1
k, kw, nwf = _siesta.read_wfsx_next_all_info(self._iu, 1, self._sizes.nk)
# Close the file unit
self._close_wfsx()
# Check for errors in the read.
self._fortran_check("read_info", "could not read file information.")
return self._convert_k(k), kw, nwf
[docs] def read_brillouinzone(self):
""" Read the brillouin zone object """
k, weight, _ = self.read_info()
return BrillouinZone(self._parent, k=k, weight=weight)
@set_module("sisl.io.siesta")
class _gridSileSiesta(SileBinSiesta):
r""" Binary real-space grid file
The Siesta binary grid sile will automatically convert the units from Siesta
units (Bohr, Ry) to sisl units (Ang, eV) provided the correct extension is present.
"""
def read_supercell(self, *args, **kwargs):
r""" Return the cell contained in the file """
cell = _siesta.read_grid_cell(self.file).T * _Bohr2Ang
self._fortran_check('read_supercell', 'could not read cell.')
return SuperCell(cell)
def read_grid_size(self):
r""" Query grid size information such as the grid size and number of spin components
Returns
-------
int : number of spin-components
mesh : 3 values for the number of mesh-elements
"""
# Read the sizes
nspin, mesh = _siesta.read_grid_sizes(self.file)
self._fortran_check('read_grid_size', 'could not read grid sizes.')
return nspin, mesh
def read_grid(self, index=0, dtype=np.float64, *args, **kwargs):
""" Read grid contained in the Grid file
Parameters
----------
index : int or array_like, optional
the spin-index for retrieving one of the components. If a vector
is passed it refers to the fraction per indexed component. I.e.
``[0.5, 0.5]`` will return sum of half the first two components.
Default to the first component.
dtype : numpy.float64, optional
default data-type precision
"""
index = kwargs.get('spin', index)
# Read the sizes and cell
nspin, mesh = self.read_grid_size()
sc = self.read_supercell()
grid = _siesta.read_grid(self.file, nspin, mesh[0], mesh[1], mesh[2])
self._fortran_check('read_grid', 'could not read grid.')
if isinstance(index, Integral):
grid = grid[:, :, :, index]
else:
if len(index) > grid.shape[0]:
raise ValueError(self.__class__.__name__ + '.read_grid requires spin to be an integer or '
'an array of length equal to the number of spin components.')
# It is F-contiguous, hence the last index
g = grid[:, :, :, 0] * index[0]
for i, scale in enumerate(index[1:]):
g += grid[:, :, :, 1+i] * scale
grid = g
# Simply create the grid (with no information)
# We will overwrite the actual grid
g = Grid([1, 1, 1], sc=sc)
# NOTE: there is no need to swap-axes since the returned array is in F ordering
# and thus the first axis is the fast (x, y, z) is retained
g.grid = grid * self.grid_unit
return g
@set_module("sisl.io.siesta")
class _gfSileSiesta(SileBinSiesta):
""" Surface Green function file containing, Hamiltonian, overlap matrix and self-energies
Do not mix read and write statements when using this code. Complete one or the other
before doing the other thing. Fortran does not allow the same file opened twice, if this
is needed you are recommended to make a symlink to the file and thus open two different
files.
This small snippet reads/writes the GF file
>>> with sisl.io._gfSileSiesta('hello.GF') as f:
... nspin, no, k, E = f.read_header()
... for ispin, new_k, k, E in f:
... if new_k:
... H, S = f.read_hamiltonian()
... SeHSE = f.read_self_energy()
To write a file do:
>>> with sisl.io._gfSileSiesta('hello.GF') as f:
... f.write_header(sisl.MonkhorstPack(...), E)
... for ispin, new_k, k, E in f:
... if new_k:
... f.write_hamiltonian(H, S)
... f.write_self_energy(SeHSE)
"""
def _setup(self, *args, **kwargs):
""" Simple setup that needs to be overwritten """
super()._setup(*args, **kwargs)
# The unit convention used for energy-points
# This is necessary until Siesta uses CODATA values
if kwargs.get("version", "old").lower() in ("old", "4.1"):
self._E_Ry2eV = 13.60580
else:
self._E_Ry2eV = _Ry2eV
def _open_gf(self, mode, rewind=False):
self._fortran_open(mode, rewind=rewind)
# They will at any given time
# correspond to the current Python indices that is to be read
# The process for identification is done on this basis:
# iE is the current (Python) index for the energy-point to be read
# ik is the current (Python) index for the k-point to be read
# ispin is the current (Python) index for the spin-index to be read (only has meaning for a spin-polarized
# GF files)
# state is:
# -1 : the file-descriptor has just been opened (i.e. in front of header)
# 0 : it means that the file-descriptor IS in front of H and S
# 1 : it means that the file-descriptor is NOT in front of H and S but somewhere in front of a self-energy
# is_read is:
# 0 : means that the current indices HAVE NOT been read
# 1 : means that the current indices HAVE been read
#
# All routines in the gf_read/write sources requires input in Python indices
self._state = -1
self._is_read = 0
self._ispin = 0
self._ik = 0
self._iE = 0
def _close_gf(self):
if not self._fortran_is_open():
return
self._fortran_close()
# Clean variables
del self._state
del self._iE
del self._ik
del self._ispin
try:
del self._no_u
except:
pass
try:
del self._nspin
except:
pass
def _step_counter(self, method, **kwargs):
""" Method for stepping values *must* be called before doing the actual read to check correct values """
opt = {'method': method}
if kwargs.get('header', False):
# The header only exists once, so check whether it is the correct place to read/write
if self._state != -1 or self._is_read == 1:
raise SileError(self.__class__.__name__ + '.{method} failed because the header has already '
'been read.'.format(**opt))
self._state = -1
self._ispin = 0
self._ik = 0
self._iE = 0
#print('HEADER: ', self._state, self._ispin, self._ik, self._iE)
elif kwargs.get('HS', False):
# Correct for the previous state and jump values
if self._state == -1:
# We have just read the header
if self._is_read != 1:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has not read the header.'.format(**opt))
# Reset values as though the header has just been read
self._state = 0
self._ispin = 0
self._ik = 0
self._iE = 0
elif self._state == 0:
if self._is_read == 1:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has already read the current HS for the given k-point.'.format(**opt))
elif self._state == 1:
# We have just read from the last energy-point
if self._iE + 1 != self._nE or self._is_read != 1:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has not read all energy-points for a given k-point.'.format(**opt))
self._state = 0
self._ik += 1
if self._ik >= self._nk:
# We need to step spin
self._ispin += 1
self._ik = 0
self._iE = 0
#print('HS: ', self._state, self._ispin, self._ik, self._iE)
if self._ispin >= self._nspin:
opt['spin'] = self._ispin + 1
opt['nspin'] = self._nspin
raise SileError(self.__class__.__name__ + '.{method} failed because of missing information, '
'a non-existing entry has been requested! spin={spin} max_spin={nspin}.'.format(**opt))
else:
# We are reading an energy-point
if self._state == -1:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has an unknown state.'.format(**opt))
elif self._state == 0:
if self._is_read == 1:
# Fine, we have just read the HS, ispin and ik are correct
self._state = 1
self._iE = 0
else:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has an unknown state.'.format(**opt))
elif self._state == 1:
if self._is_read == 0 and self._iE < self._nE:
# we haven't read the current energy-point.and self._iE + 1 < self._nE:
pass
elif self._is_read == 1 and self._iE + 1 < self._nE:
self._iE += 1
else:
raise SileError(self.__class__.__name__ + '.{method} failed because the file descriptor '
'has an unknown state.'.format(**opt))
if self._iE >= self._nE:
# You are trying to read beyond the entry
opt['iE'] = self._iE + 1
opt['NE'] = self._nE
raise SileError(self.__class__.__name__ + '.{method} failed because of missing information, '
'a non-existing energy-point has been requested! E_index={iE} max_E_index={NE}.'.format(**opt))
#print('SE: ', self._state, self._ispin, self._ik, self._iE)
# Always signal (when stepping) that we have not yet read the thing
if kwargs.get('read', False):
self._is_read = 1
else:
self._is_read = 0
def Eindex(self, E):
""" Return the closest energy index corresponding to the energy ``E``
Parameters
----------
E : float or int
if ``int``, return it-self, else return the energy index which is
closests to the energy.
"""
if isinstance(E, Integral):
return E
idxE = np.abs(self._E - E).argmin()
ret_E = self._E[idxE]
if abs(ret_E - E) > 5e-3:
warn(self.__class__.__name__ + " requesting energy " +
f"{E:.5f} eV, found {ret_E:.5f} eV as the closest energy!")
elif abs(ret_E - E) > 1e-3:
info(self.__class__.__name__ + " requesting energy " +
f"{E:.5f} eV, found {ret_E:.5f} eV as the closest energy!")
return idxE
def kindex(self, k):
""" Return the index of the k-point that is closests to the queried k-point (in reduced coordinates)
Parameters
----------
k : array_like of float or int
the queried k-point in reduced coordinates :math:`]-0.5;0.5]`. If ``int``
return it-self.
"""
if isinstance(k, Integral):
return k
ik = np.sum(np.abs(self._k - _a.asarrayd(k)[None, :]), axis=1).argmin()
ret_k = self._k[ik, :]
if not np.allclose(ret_k, k, atol=0.0001):
warn(SileWarning(self.__class__.__name__ + " requesting k-point " +
"[{:.3f}, {:.3f}, {:.3f}]".format(*k) +
" found " +
"[{:.3f}, {:.3f}, {:.3f}]".format(*ret_k)))
return ik
def read_header(self):
""" Read the header of the file and open it for reading subsequently
NOTES: this method may change in the future
Returns
-------
nspin : number of spin-components stored (1 or 2)
no_u : size of the matrices returned
k : k points in the GF file
E : energy points in the GF file
"""
# Ensure it is open (in read-mode)
if self._fortran_is_open():
_siesta.io_m.rewind_file(self._iu)
else:
self._open_gf('r')
nspin, no_u, nkpt, NE = _siesta.read_gf_sizes(self._iu)
self._fortran_check('read_header', 'could not read sizes.')
self._nspin = nspin
self._nk = nkpt
self._nE = NE
# We need to rewind (because of k and energy -points)
_siesta.io_m.rewind_file(self._iu)
self._step_counter('read_header', header=True, read=True)
k, E = _siesta.read_gf_header(self._iu, nkpt, NE)
self._fortran_check('read_header', 'could not read header information.')
if self._nspin > 2: # non-colinear
self._no_u = no_u * 2
else:
self._no_u = no_u
self._E = E * self._E_Ry2eV
self._k = k.T
return nspin, no_u, self._k, self._E
def disk_usage(self):
""" Calculate the estimated size of the resulting file
Returns
-------
estimated disk-space used in GB
"""
is_open = self._fortran_is_open()
if not is_open:
self.read_header()
# HS are only stored per k-point
HS = 2 * self._nspin * self._nk
SE = HS / 2 * self._nE
# Now calculate the full size
# no_u ** 2 = matrix size
# 16 = bytes in double complex
# 1024 ** 3 = B -> GB
mem = (HS + SE) * self._no_u ** 2 * 16 / 1024 ** 3
if not is_open:
self._close_gf()
return mem
def read_hamiltonian(self):
""" Return current Hamiltonian and overlap matrix from the GF file
Returns
-------
complex128 : Hamiltonian matrix
complex128 : Overlap matrix
"""
self._step_counter('read_hamiltonian', HS=True, read=True)
H, S = _siesta.read_gf_hs(self._iu, self._no_u)
self._fortran_check('read_hamiltonian', 'could not read Hamiltonian and overlap matrices.')
# we don't convert to C order!
return H * _Ry2eV, S
def read_self_energy(self):
r""" Read the currently reached bulk self-energy
The returned self-energy is:
.. math::
\boldsymbol \Sigma_{\mathrm{bulk}}(E) = \mathbf S E - \mathbf H - \boldsymbol \Sigma(E)
Returns
-------
complex128 : Self-energy matrix
"""
self._step_counter('read_self_energy', read=True)
SE = _siesta.read_gf_se(self._iu, self._no_u, self._iE)
self._fortran_check('read_self_energy', 'could not read self-energy.')
# we don't convert to C order!
return SE * _Ry2eV
def HkSk(self, k=(0, 0, 0), spin=0):
""" Retrieve H and S for the given k-point
Parameters
----------
k : int or array_like of float, optional
k-point to read the corresponding Hamiltonian and overlap matrices
for. If a specific k-point is passed `kindex` will be used to find
the corresponding index.
spin : int, optional
spin-index for the Hamiltonian and overlap matrices
"""
if not self._fortran_is_open():
self.read_header()
# find k-index that is requested
ik = self.kindex(k)
_siesta.read_gf_find(self._iu, self._nspin, self._nk, self._nE,
self._state, self._ispin, self._ik, self._iE, self._is_read,
0, spin, ik, 0)
self._fortran_check('HkSk', 'could not find Hamiltonian and overlap matrix.')
self._state = 0
self._ispin = spin
self._ik = ik
self._iE = 0
self._is_read = 0 # signal this is to be read
return self.read_hamiltonian()
def self_energy(self, E, k=0, spin=0):
""" Retrieve self-energy for a given energy-point and k-point
Parameters
----------
E : int or float
energy to retrieve self-energy at
k : int or array_like of float, optional
k-point to retrieve k-point at
spin : int, optional
spin-index to retrieve self-energy at
"""
if not self._fortran_is_open():
self.read_header()
ik = self.kindex(k)
iE = self.Eindex(E)
_siesta.read_gf_find(self._iu, self._nspin, self._nk, self._nE,
self._state, self._ispin, self._ik, self._iE, self._is_read,
1, spin, ik, iE)
self._fortran_check('self_energy', 'could not find requested self-energy.')
self._state = 1
self._ispin = spin
self._ik = ik
self._iE = iE
self._is_read = 0 # signal this is to be read
return self.read_self_energy()
def write_header(self, bz, E, mu=0., obj=None):
""" Write to the binary file the header of the file
Parameters
----------
bz : BrillouinZone
contains the k-points, the weights and possibly the parent Hamiltonian (if `obj` is None)s
E : array_like of cmplx or float
the energy points. If `obj` is an instance of `SelfEnergy` where an
associated ``eta`` is defined then `E` may be float, otherwise
it *has* to be a complex array.
mu : float, optional
chemical potential in the file
obj : ..., optional
an object that contains the Hamiltonian definitions, defaults to ``bz.parent``
"""
if obj is None:
obj = bz.parent
nspin = len(obj.spin)
cell = obj.geometry.sc.cell
na_u = obj.geometry.na
no_u = obj.geometry.no
xa = obj.geometry.xyz
# The lasto in siesta requires lasto(0) == 0
# and secondly, the Python index to fortran
# index makes firsto behave like fortran lasto
lasto = obj.geometry.firsto
bloch = _a.onesi(3)
mu = mu
NE = len(E)
if E.dtype not in [np.complex64, np.complex128]:
E = E + 1j * obj.eta
Nk = len(bz)
k = bz.k
w = bz.weight
sizes = {
'na_used': na_u,
'nkpt': Nk,
'ne': NE,
}
self._nspin = nspin
self._E = E
self._k = np.copy(k)
self._nE = len(E)
self._nk = len(k)
if self._nspin > 2:
self._no_u = no_u * 2
else:
self._no_u = no_u
# Ensure it is open (in write mode)
self._close_gf()
self._open_gf('w')
# Now write to it...
self._step_counter('write_header', header=True, read=True)
# see onlysSileSiesta.read_supercell for .T
_siesta.write_gf_header(self._iu, nspin, _toF(cell.T, np.float64, 1. / _Bohr2Ang),
na_u, no_u, no_u, _toF(xa.T, np.float64, 1. / _Bohr2Ang),
lasto, bloch, 0, mu * _eV2Ry, _toF(k.T, np.float64),
w, self._E / self._E_Ry2eV,
**sizes)
self._fortran_check('write_header', 'could not write header information.')
def write_hamiltonian(self, H, S=None):
""" Write the current energy, k-point and H and S to the file
Parameters
----------
H : matrix
a square matrix corresponding to the Hamiltonian
S : matrix, optional
a square matrix corresponding to the overlap, for efficiency reasons
it may be advantageous to specify this argument for orthogonal cells.
"""
no = len(H)
if S is None:
S = np.eye(no, dtype=np.complex128, order='F')
self._step_counter('write_hamiltonian', HS=True, read=True)
_siesta.write_gf_hs(self._iu, self._ik, self._E[self._iE] / self._E_Ry2eV,
_toF(H, np.complex128, _eV2Ry),
_toF(S, np.complex128), no_u=no)
self._fortran_check('write_hamiltonian', 'could not write Hamiltonian and overlap matrices.')
def write_self_energy(self, SE):
r""" Write the current self energy, k-point and H and S to the file
The self-energy must correspond to the *bulk* self-energy
.. math::
\boldsymbol \Sigma_{\mathrm{bulk}}(E) = \mathbf S E - \mathbf H - \boldsymbol \Sigma(E)
Parameters
----------
SE : matrix
a square matrix corresponding to the self-energy (Green function)
"""
no = len(SE)
self._step_counter('write_self_energy', read=True)
_siesta.write_gf_se(self._iu, self._ik, self._iE, self._E[self._iE] / self._E_Ry2eV,
_toF(SE, np.complex128, _eV2Ry), no_u=no)
self._fortran_check('write_self_energy', 'could not write self-energy.')
def __len__(self):
return self._nE * self._nk * self._nspin
def __iter__(self):
""" Iterate through the energies and k-points that this GF file is associated with
Yields
------
bool, list of float, float
"""
# get everything
e = self._E
if self._nspin in [1, 2]:
GFStep = namedtuple("GFStep", ["spin", "do_HS", "k", "E"])
for ispin in range(self._nspin):
for k in self._k:
yield GFStep(ispin, True, k, e[0])
for E in e[1:]:
yield GFStep(ispin, False, k, E)
else:
GFStep = namedtuple("GFStep", ["do_HS", "k", "E"])
for k in self._k:
yield GFStep(True, k, e[0])
for E in e[1:]:
yield GFStep(False, k, E)
# We will automatically close once we hit the end
self._close_gf()
def _type(name, obj, dic=None):
if dic is None:
dic = {}
# Always pass the docstring
if not '__doc__' in dic:
try:
dic['__doc__'] = obj.__doc__.replace(obj.__name__, name)
except:
pass
return type(name, (obj, ), dic)
# Faster than class ... \ pass
tsgfSileSiesta = _type("tsgfSileSiesta", _gfSileSiesta)
gridSileSiesta = _type("gridSileSiesta", _gridSileSiesta, {'grid_unit': 1.})
if found_module:
add_sile('TSHS', tshsSileSiesta)
add_sile('onlyS', onlysSileSiesta)
add_sile('TSDE', tsdeSileSiesta)
add_sile('DM', dmSileSiesta)
add_sile('HSX', hsxSileSiesta)
add_sile('TSGF', tsgfSileSiesta)
add_sile('WFSX', wfsxSileSiesta)
# These have unit-conversions
add_sile('RHO', _type("rhoSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('LDOS', _type("ldosSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('RHOINIT', _type("rhoinitSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('RHOXC', _type("rhoxcSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('DRHO', _type("drhoSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('BADER', _type("baderSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('IOCH', _type("iorhoSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('TOCH', _type("totalrhoSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
# The following two files *require* that
# STM.DensityUnits Ele/bohr**3
# which I can't check!
# They are however the default
add_sile('STS', _type("stsSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('STM.LDOS', _type("stmldosSileSiesta", _gridSileSiesta, {'grid_unit': 1./_Bohr2Ang ** 3}))
add_sile('VH', _type("hartreeSileSiesta", _gridSileSiesta, {'grid_unit': _Ry2eV}))
add_sile('VNA', _type("neutralatomhartreeSileSiesta", _gridSileSiesta, {'grid_unit': _Ry2eV}))
add_sile('VT', _type("totalhartreeSileSiesta", _gridSileSiesta, {'grid_unit': _Ry2eV}))