import re
import warnings
import numpy as np
from ase.atoms import Atoms
from ase.calculators.lammps import Prism, convert
from ase.data import atomic_masses, atomic_numbers
from ase.utils import reader, writer
[docs]
@reader
def read_lammps_data(
fileobj,
Z_of_type: dict = None,
sort_by_id: bool = True,
read_image_flags: bool = True,
units: str = 'metal',
atom_style: str = None,
style: str = None,
):
"""Method which reads a LAMMPS data file.
Parameters
----------
fileobj : file | str
File from which data should be read.
Z_of_type : dict[int, int], optional
Mapping from LAMMPS atom types (typically starting from 1) to atomic
numbers. If None, if there is the "Masses" section, atomic numbers are
guessed from the atomic masses. Otherwise, atomic numbers of 1 (H), 2
(He), etc. are assigned to atom types of 1, 2, etc. Default is None.
sort_by_id : bool, optional
Order the particles according to their id. Might be faster to set it
False. Default is True.
read_image_flags: bool, default True
If True, the lattice translation vectors derived from image flags are
added to atomic positions.
units : str, optional
`LAMMPS units <https://docs.lammps.org/units.html>`__. Default is
'metal'.
atom_style : {'atomic', 'charge', 'full'} etc., optional
`LAMMPS atom style <https://docs.lammps.org/atom_style.html>`__.
If None, `atom_style` is guessed in the following priority (1) comment
after `Atoms` (2) length of fields (valid only `atomic` and `full`).
Default is None.
"""
if style is not None:
warnings.warn(
FutureWarning('"style" is deprecated; please use "atom_style".'),
)
atom_style = style
# begin read_lammps_data
file_comment = next(fileobj).rstrip()
# default values (https://docs.lammps.org/read_data.html)
# in most cases these will be updated below
natoms = 0
# N_types = 0
xlo, xhi = -0.5, 0.5
ylo, yhi = -0.5, 0.5
zlo, zhi = -0.5, 0.5
xy, xz, yz = 0.0, 0.0, 0.0
mass_in = {}
vel_in = {}
bonds_in = []
angles_in = []
dihedrals_in = []
sections = [
'Atoms',
'Velocities',
'Masses',
'Charges',
'Ellipsoids',
'Lines',
'Triangles',
'Bodies',
'Bonds',
'Angles',
'Dihedrals',
'Impropers',
'Impropers Pair Coeffs',
'PairIJ Coeffs',
'Pair Coeffs',
'Bond Coeffs',
'Angle Coeffs',
'Dihedral Coeffs',
'Improper Coeffs',
'BondBond Coeffs',
'BondAngle Coeffs',
'MiddleBondTorsion Coeffs',
'EndBondTorsion Coeffs',
'AngleTorsion Coeffs',
'AngleAngleTorsion Coeffs',
'BondBond13 Coeffs',
'AngleAngle Coeffs',
]
header_fields = [
'atoms',
'bonds',
'angles',
'dihedrals',
'impropers',
'atom types',
'bond types',
'angle types',
'dihedral types',
'improper types',
'extra bond per atom',
'extra angle per atom',
'extra dihedral per atom',
'extra improper per atom',
'extra special per atom',
'ellipsoids',
'lines',
'triangles',
'bodies',
'xlo xhi',
'ylo yhi',
'zlo zhi',
'xy xz yz',
]
sections_re = '(' + '|'.join(sections).replace(' ', '\\s+') + ')'
header_fields_re = '(' + '|'.join(header_fields).replace(' ', '\\s+') + ')'
section = None
header = True
for line in fileobj:
# get string after #; if # does not exist, return ''
line_comment = re.sub(r'^.*#|^.*$', '', line).strip()
line = re.sub('#.*', '', line).rstrip().lstrip()
if re.match('^\\s*$', line): # skip blank lines
continue
# check for known section names
match = re.match(sections_re, line)
if match is not None:
section = match.group(0).rstrip().lstrip()
header = False
if section == 'Atoms': # id *
# guess `atom_style` from the comment after `Atoms` if exists
if atom_style is None and line_comment != '':
atom_style = line_comment
mol_id_in, type_in, charge_in, pos_in, travel_in = \
_read_atoms_section(fileobj, natoms, atom_style)
continue
if header:
field = None
val = None
match = re.match('(.*)\\s+' + header_fields_re, line)
if match is not None:
field = match.group(2).lstrip().rstrip()
val = match.group(1).lstrip().rstrip()
if field is not None and val is not None:
if field == 'atoms':
natoms = int(val)
elif field == 'xlo xhi':
(xlo, xhi) = (float(x) for x in val.split())
elif field == 'ylo yhi':
(ylo, yhi) = (float(x) for x in val.split())
elif field == 'zlo zhi':
(zlo, zhi) = (float(x) for x in val.split())
elif field == 'xy xz yz':
(xy, xz, yz) = (float(x) for x in val.split())
if section is not None:
fields = line.split()
if section == 'Velocities': # id vx vy vz
atom_id = int(fields[0])
vel_in[atom_id] = [float(fields[_]) for _ in (1, 2, 3)]
elif section == 'Masses':
mass_in[int(fields[0])] = float(fields[1])
elif section == 'Bonds': # id type atom1 atom2
bonds_in.append([int(fields[_]) for _ in (1, 2, 3)])
elif section == 'Angles': # id type atom1 atom2 atom3
angles_in.append([int(fields[_]) for _ in (1, 2, 3, 4)])
elif section == 'Dihedrals': # id type atom1 atom2 atom3 atom4
dihedrals_in.append([int(fields[_]) for _ in (1, 2, 3, 4, 5)])
# set cell
cell = np.zeros((3, 3))
cell[0, 0] = xhi - xlo
cell[1, 1] = yhi - ylo
cell[2, 2] = zhi - zlo
cell[1, 0] = xy
cell[2, 0] = xz
cell[2, 1] = yz
# initialize arrays for per-atom quantities
positions = np.zeros((natoms, 3))
numbers = np.zeros(natoms, int)
ids = np.zeros(natoms, int)
types = np.zeros(natoms, int)
velocities = np.zeros((natoms, 3)) if len(vel_in) > 0 else None
masses = np.zeros(natoms) if len(mass_in) > 0 else None
mol_id = np.zeros(natoms, int) if len(mol_id_in) > 0 else None
charge = np.zeros(natoms, float) if len(charge_in) > 0 else None
travel = np.zeros((natoms, 3), int) if len(travel_in) > 0 else None
bonds = [''] * natoms if len(bonds_in) > 0 else None
angles = [''] * natoms if len(angles_in) > 0 else None
dihedrals = [''] * natoms if len(dihedrals_in) > 0 else None
ind_of_id = {}
# copy per-atom quantities from read-in values
for (i, atom_id) in enumerate(pos_in.keys()):
# by id
if sort_by_id:
ind = atom_id - 1
else:
ind = i
ind_of_id[atom_id] = ind
atom_type = type_in[atom_id]
positions[ind, :] = pos_in[atom_id]
if velocities is not None:
velocities[ind, :] = vel_in[atom_id]
if travel is not None:
travel[ind] = travel_in[atom_id]
if mol_id is not None:
mol_id[ind] = mol_id_in[atom_id]
if charge is not None:
charge[ind] = charge_in[atom_id]
ids[ind] = atom_id
# by type
types[ind] = atom_type
if Z_of_type is None:
numbers[ind] = atom_type
else:
numbers[ind] = Z_of_type[atom_type]
if masses is not None:
masses[ind] = mass_in[atom_type]
# convert units
positions = convert(positions, 'distance', units, 'ASE')
cell = convert(cell, 'distance', units, 'ASE')
if masses is not None:
masses = convert(masses, 'mass', units, 'ASE')
if velocities is not None:
velocities = convert(velocities, 'velocity', units, 'ASE')
# guess atomic numbers from atomic masses
# this must be after the above mass-unit conversion
if Z_of_type is None and masses is not None:
numbers = _masses2numbers(masses)
# create ase.Atoms
atoms = Atoms(
positions=positions,
numbers=numbers,
masses=masses,
cell=cell,
pbc=[True, True, True],
)
# add lattice translation vectors
if read_image_flags and travel is not None:
scaled_positions = atoms.get_scaled_positions(wrap=False) + travel
atoms.set_scaled_positions(scaled_positions)
# set velocities (can't do it via constructor)
if velocities is not None:
atoms.set_velocities(velocities)
atoms.arrays['id'] = ids
atoms.arrays['type'] = types
if mol_id is not None:
atoms.arrays['mol-id'] = mol_id
if charge is not None:
atoms.arrays['initial_charges'] = charge
atoms.arrays['mmcharges'] = charge.copy()
if bonds is not None:
for (atom_type, at1, at2) in bonds_in:
i_a1 = ind_of_id[at1]
i_a2 = ind_of_id[at2]
if len(bonds[i_a1]) > 0:
bonds[i_a1] += ','
bonds[i_a1] += f'{i_a2:d}({atom_type:d})'
for i, bond in enumerate(bonds):
if len(bond) == 0:
bonds[i] = '_'
atoms.arrays['bonds'] = np.array(bonds)
if angles is not None:
for (atom_type, at1, at2, at3) in angles_in:
i_a1 = ind_of_id[at1]
i_a2 = ind_of_id[at2]
i_a3 = ind_of_id[at3]
if len(angles[i_a2]) > 0:
angles[i_a2] += ','
angles[i_a2] += f'{i_a1:d}-{i_a3:d}({atom_type:d})'
for i, angle in enumerate(angles):
if len(angle) == 0:
angles[i] = '_'
atoms.arrays['angles'] = np.array(angles)
if dihedrals is not None:
for (atom_type, at1, at2, at3, at4) in dihedrals_in:
i_a1 = ind_of_id[at1]
i_a2 = ind_of_id[at2]
i_a3 = ind_of_id[at3]
i_a4 = ind_of_id[at4]
if len(dihedrals[i_a1]) > 0:
dihedrals[i_a1] += ','
dihedrals[i_a1] += f'{i_a2:d}-{i_a3:d}-{i_a4:d}({atom_type:d})'
for i, dihedral in enumerate(dihedrals):
if len(dihedral) == 0:
dihedrals[i] = '_'
atoms.arrays['dihedrals'] = np.array(dihedrals)
atoms.info['comment'] = file_comment
return atoms
def _read_atoms_section(fileobj, natoms: int, style: str = None):
type_in = {}
mol_id_in = {}
charge_in = {}
pos_in = {}
travel_in = {}
next(fileobj) # skip blank line just after `Atoms`
for _ in range(natoms):
line = next(fileobj)
line = re.sub('#.*', '', line).rstrip().lstrip()
fields = line.split()
if style is None:
style = _guess_atom_style(fields)
atom_id = int(fields[0])
if style == 'full' and len(fields) in (7, 10):
# id mol-id type q x y z [tx ty tz]
type_in[atom_id] = int(fields[2])
pos_in[atom_id] = tuple(float(fields[_]) for _ in (4, 5, 6))
mol_id_in[atom_id] = int(fields[1])
charge_in[atom_id] = float(fields[3])
if len(fields) == 10:
travel_in[atom_id] = tuple(int(fields[_]) for _ in (7, 8, 9))
elif style == 'atomic' and len(fields) in (5, 8):
# id type x y z [tx ty tz]
type_in[atom_id] = int(fields[1])
pos_in[atom_id] = tuple(float(fields[_]) for _ in (2, 3, 4))
if len(fields) == 8:
travel_in[atom_id] = tuple(int(fields[_]) for _ in (5, 6, 7))
elif style in ('angle', 'bond', 'molecular') and len(fields) in (6, 9):
# id mol-id type x y z [tx ty tz]
type_in[atom_id] = int(fields[2])
pos_in[atom_id] = tuple(float(fields[_]) for _ in (3, 4, 5))
mol_id_in[atom_id] = int(fields[1])
if len(fields) == 9:
travel_in[atom_id] = tuple(int(fields[_]) for _ in (6, 7, 8))
elif style == 'charge' and len(fields) in (6, 9):
# id type q x y z [tx ty tz]
type_in[atom_id] = int(fields[1])
pos_in[atom_id] = tuple(float(fields[_]) for _ in (3, 4, 5))
charge_in[atom_id] = float(fields[2])
if len(fields) == 9:
travel_in[atom_id] = tuple(int(fields[_]) for _ in (6, 7, 8))
else:
raise RuntimeError(
f'Style "{style}" not supported or invalid. '
f'Number of fields: {len(fields)}'
)
return mol_id_in, type_in, charge_in, pos_in, travel_in
def _guess_atom_style(fields):
"""Guess `atom_sytle` from the length of fields."""
if len(fields) in (5, 8):
return 'atomic'
if len(fields) in (7, 10):
return 'full'
raise ValueError('atom_style cannot be guessed from len(fields)')
def _masses2numbers(masses):
"""Guess atomic numbers from atomic masses."""
return np.argmin(np.abs(atomic_masses - masses[:, None]), axis=1)
[docs]
@writer
def write_lammps_data(
fd,
atoms: Atoms,
*,
specorder: list = None,
reduce_cell: bool = False,
force_skew: bool = False,
prismobj: Prism = None,
write_image_flags: bool = False,
masses: bool = False,
velocities: bool = False,
units: str = 'metal',
bonds: bool = True,
atom_style: str = 'atomic',
):
"""Write atomic structure data to a LAMMPS data file.
Parameters
----------
fd : file|str
File to which the output will be written.
atoms : Atoms
Atoms to be written.
specorder : list[str], optional
Chemical symbols in the order of LAMMPS atom types, by default None
force_skew : bool, optional
Force to write the cell as a
`triclinic <https://docs.lammps.org/Howto_triclinic.html>`__ box,
by default False
reduce_cell : bool, optional
Whether the cell shape is reduced or not, by default False
prismobj : Prism|None, optional
Prism, by default None
write_image_flags : bool, default False
If True, the image flags, i.e., in which images of the periodic
simulation box the atoms are, are written.
masses : bool, optional
Whether the atomic masses are written or not, by default False
velocities : bool, optional
Whether the atomic velocities are written or not, by default False
units : str, optional
`LAMMPS units <https://docs.lammps.org/units.html>`__,
by default 'metal'
bonds : bool, optional
Whether the bonds are written or not. Bonds can only be written
for atom_style='full', by default True
atom_style : {'atomic', 'charge', 'full'}, optional
`LAMMPS atom style <https://docs.lammps.org/atom_style.html>`__,
by default 'atomic'
"""
# FIXME: We should add a check here that the encoding of the file object
# is actually ascii once the 'encoding' attribute of IOFormat objects
# starts functioning in implementation (currently it doesn't do
# anything).
if isinstance(atoms, list):
if len(atoms) > 1:
raise ValueError(
'Can only write one configuration to a lammps data file!'
)
atoms = atoms[0]
fd.write('(written by ASE)\n\n')
symbols = atoms.get_chemical_symbols()
n_atoms = len(symbols)
fd.write(f'{n_atoms} atoms\n')
if specorder is None:
# This way it is assured that LAMMPS atom types are always
# assigned predictably according to the alphabetic order
species = sorted(set(symbols))
else:
# To index elements in the LAMMPS data file
# (indices must correspond to order in the potential file)
species = specorder
n_atom_types = len(species)
fd.write(f'{n_atom_types} atom types\n\n')
bonds_in = []
if (bonds and (atom_style == 'full') and
(atoms.arrays.get('bonds') is not None)):
n_bonds = 0
n_bond_types = 1
for i, bondsi in enumerate(atoms.arrays['bonds']):
if bondsi != '_':
for bond in bondsi.split(','):
dummy1, dummy2 = bond.split('(')
bond_type = int(dummy2.split(')')[0])
at1 = int(i) + 1
at2 = int(dummy1) + 1
bonds_in.append((bond_type, at1, at2))
n_bonds = n_bonds + 1
if bond_type > n_bond_types:
n_bond_types = bond_type
fd.write(f'{n_bonds} bonds\n')
fd.write(f'{n_bond_types} bond types\n\n')
if prismobj is None:
prismobj = Prism(atoms.get_cell(), reduce_cell=reduce_cell)
# Get cell parameters and convert from ASE units to LAMMPS units
xhi, yhi, zhi, xy, xz, yz = convert(
prismobj.get_lammps_prism(), 'distance', 'ASE', units)
fd.write(f'0.0 {xhi:23.17g} xlo xhi\n')
fd.write(f'0.0 {yhi:23.17g} ylo yhi\n')
fd.write(f'0.0 {zhi:23.17g} zlo zhi\n')
if force_skew or prismobj.is_skewed():
fd.write(f'{xy:23.17g} {xz:23.17g} {yz:23.17g} xy xz yz\n')
fd.write('\n')
if masses:
_write_masses(fd, atoms, species, units)
# Write (unwrapped) atomic positions. If wrapping of atoms back into the
# cell along periodic directions is desired, this should be done manually
# on the Atoms object itself beforehand.
fd.write(f'Atoms # {atom_style}\n\n')
if write_image_flags:
scaled_positions = atoms.get_scaled_positions(wrap=False)
image_flags = np.floor(scaled_positions).astype(int)
# when `write_image_flags` is True, the positions are wrapped while the
# unwrapped positions can be recovered from the image flags
pos = prismobj.vector_to_lammps(
atoms.get_positions(),
wrap=write_image_flags,
)
if atom_style == 'atomic':
# Convert position from ASE units to LAMMPS units
pos = convert(pos, 'distance', 'ASE', units)
for i, r in enumerate(pos):
s = species.index(symbols[i]) + 1
line = (
f'{i + 1:>6} {s:>3}'
f' {r[0]:23.17g} {r[1]:23.17g} {r[2]:23.17g}'
)
if write_image_flags:
img = image_flags[i]
line += f' {img[0]:6d} {img[1]:6d} {img[2]:6d}'
line += '\n'
fd.write(line)
elif atom_style == 'charge':
charges = atoms.get_initial_charges()
# Convert position and charge from ASE units to LAMMPS units
pos = convert(pos, 'distance', 'ASE', units)
charges = convert(charges, 'charge', 'ASE', units)
for i, (q, r) in enumerate(zip(charges, pos)):
s = species.index(symbols[i]) + 1
line = (
f'{i + 1:>6} {s:>3} {q:>5}'
f' {r[0]:23.17g} {r[1]:23.17g} {r[2]:23.17g}'
)
if write_image_flags:
img = image_flags[i]
line += f' {img[0]:6d} {img[1]:6d} {img[2]:6d}'
line += '\n'
fd.write(line)
elif atom_style == 'full':
charges = atoms.get_initial_charges()
# The label 'mol-id' has apparenlty been introduced in read earlier,
# but so far not implemented here. Wouldn't a 'underscored' label
# be better, i.e. 'mol_id' or 'molecule_id'?
if atoms.has('mol-id'):
molecules = atoms.get_array('mol-id')
if not np.issubdtype(molecules.dtype, np.integer):
raise TypeError(
f'If "atoms" object has "mol-id" array, then '
f'mol-id dtype must be subtype of np.integer, and '
f'not {molecules.dtype!s:s}.')
if (len(molecules) != len(atoms)) or (molecules.ndim != 1):
raise TypeError(
'If "atoms" object has "mol-id" array, then '
'each atom must have exactly one mol-id.')
else:
# Assigning each atom to a distinct molecule id would seem
# preferableabove assigning all atoms to a single molecule
# id per default, as done within ase <= v 3.19.1. I.e.,
# molecules = np.arange(start=1, stop=len(atoms)+1,
# step=1, dtype=int) However, according to LAMMPS default
# behavior,
molecules = np.zeros(len(atoms), dtype=int)
# which is what happens if one creates new atoms within LAMMPS
# without explicitly taking care of the molecule id.
# Quote from docs at https://lammps.sandia.gov/doc/read_data.html:
# The molecule ID is a 2nd identifier attached to an atom.
# Normally, it is a number from 1 to N, identifying which
# molecule the atom belongs to. It can be 0 if it is a
# non-bonded atom or if you don't care to keep track of molecule
# assignments.
# Convert position and charge from ASE units to LAMMPS units
pos = convert(pos, 'distance', 'ASE', units)
charges = convert(charges, 'charge', 'ASE', units)
for i, (m, q, r) in enumerate(zip(molecules, charges, pos)):
s = species.index(symbols[i]) + 1
line = (
f'{i + 1:>6} {m:>3} {s:>3} {q:>5}'
f' {r[0]:23.17g} {r[1]:23.17g} {r[2]:23.17g}'
)
if write_image_flags:
img = image_flags[i]
line += f' {img[0]:6d} {img[1]:6d} {img[2]:6d}'
line += '\n'
fd.write(line)
if bonds and (atoms.arrays.get('bonds') is not None):
fd.write('\nBonds\n\n')
for i in range(n_bonds):
bond_type = bonds_in[i][0]
at1 = bonds_in[i][1]
at2 = bonds_in[i][2]
fd.write(f'{i + 1:>3} {bond_type:>3} {at1:>3} {at2:>3}\n')
else:
raise ValueError(atom_style)
if velocities and atoms.get_velocities() is not None:
fd.write('\n\nVelocities\n\n')
vel = prismobj.vector_to_lammps(atoms.get_velocities())
# Convert velocity from ASE units to LAMMPS units
vel = convert(vel, 'velocity', 'ASE', units)
for i, v in enumerate(vel):
fd.write(f'{i + 1:>6} {v[0]:23.17g} {v[1]:23.17g} {v[2]:23.17g}\n')
fd.flush()
def _write_masses(fd, atoms: Atoms, species: list, units: str):
symbols_indices = atoms.symbols.indices()
fd.write('Masses\n\n')
for i, s in enumerate(species):
if s in symbols_indices:
# Find the first atom of the element `s` and extract its mass
# Cover by `float` to make a new object for safety
mass = float(atoms[symbols_indices[s][0]].mass)
else:
# Fetch from ASE data if the element `s` is not in the system
mass = atomic_masses[atomic_numbers[s]]
# Convert mass from ASE units to LAMMPS units
mass = convert(mass, 'mass', 'ASE', units)
atom_type = i + 1
fd.write(f'{atom_type:>6} {mass:23.17g} # {s}\n')
fd.write('\n')