Source code for ase.io.gamess_us

"""Module for GAMESS US IO."""

from __future__ import annotations

import os
import re
from copy import deepcopy
from io import TextIOBase
from subprocess import TimeoutExpired, call

import numpy as np

from ase import Atoms
from ase.calculators.singlepoint import SinglePointCalculator
from ase.units import Bohr, Debye, Hartree
from ase.utils import reader, workdir, writer


def _format_value(val: bool | str) -> str:
    if isinstance(val, bool):
        return '.t.' if val else '.f.'
    return str(val).upper()


def _write_block(name: str, args: dict[str, bool | str]) -> str:
    out = [f' ${name.upper()}']
    for key, val in args.items():
        out.append(f'  {key.upper()}={_format_value(val)}')
    out.append(' $END')
    return '\n'.join(out)


def _write_geom(atoms: Atoms, basis_spec: dict[str | int, str] | None) -> str:
    out = [' $DATA', atoms.get_chemical_formula(), 'C1']
    for i, atom in enumerate(atoms):
        out.append(
            '{:<3} {:>3} {:20.13e} {:20.13e} {:20.13e}'.format(
                atom.symbol, atom.number, *atom.position
            )
        )
        if basis_spec is not None:
            basis = basis_spec.get(i)
            if basis is None:
                basis = basis_spec.get(atom.symbol)
            if basis is None:
                msg = f'Could not find an appropriate basis set for atom {i}'
                raise ValueError(msg)
            out += [basis, '']
    out.append(' $END')
    return '\n'.join(out)


def _write_ecp(atoms: Atoms, ecp: dict[int | str, str]) -> str:
    out = [' $ECP']
    for i, symbol in enumerate(atoms.symbols):
        if i in ecp:
            out.append(ecp[i])
        elif symbol in ecp:
            out.append(ecp[symbol])
        else:
            msg = f'Could not find an appropriate ECP for atom {i}'
            raise ValueError(msg)
    out.append(' $END')
    return '\n'.join(out)


_xc = dict(LDA='SVWN')


[docs] @writer def write_gamess_us_in(fd: TextIOBase, atoms: Atoms, properties=None, **params): params = deepcopy(params) if properties is None: properties = ['energy'] # set RUNTYP from properties iff value not provided by the user contrl = params.pop('contrl', {}) if 'runtyp' not in contrl: if 'forces' in properties: contrl['runtyp'] = 'gradient' else: contrl['runtyp'] = 'energy' # Support xc keyword for functional specification xc = params.pop('xc', None) if xc is not None and 'dfttyp' not in contrl: contrl['dfttyp'] = _xc.get(xc.upper(), xc.upper()) # Automatically determine multiplicity from magnetic moment magmom_tot = int(round(atoms.get_initial_magnetic_moments().sum())) if 'mult' not in contrl: contrl['mult'] = abs(magmom_tot) + 1 # Since we're automatically determining multiplicity, we also # need to automatically switch to UHF when the multiplicity # is not 1 if 'scftyp' not in contrl: contrl['scftyp'] = 'rhf' if contrl['mult'] == 1 else 'uhf' # effective core potentials ecp = params.pop('ecp', None) if ecp is not None and 'pp' not in contrl: contrl['pp'] = 'READ' # If no basis set is provided, use 3-21G by default. basis_spec = None if 'basis' not in params: params['basis'] = dict(gbasis='N21', ngauss=3) else: keys = set(params['basis']) # Check if the user is specifying a literal per-atom basis. # We assume they are passing a per-atom basis if the keys of the # basis dict are atom symbols, or if they are atom indices, or # a mixture of both. if keys.intersection(set(atoms.symbols)) or any( map(lambda x: isinstance(x, int), keys) ): basis_spec = params.pop('basis') out = [_write_block('contrl', contrl)] out += [_write_block(*item) for item in params.items()] out.append(_write_geom(atoms, basis_spec)) if ecp is not None: out.append(_write_ecp(atoms, ecp)) fd.write('\n\n'.join(out))
_geom_re = re.compile(r'^\s*ATOM\s+ATOMIC\s+COORDINATES') _atom_re = re.compile(r'^\s*(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s*\n') _energy_re = re.compile(r'^\s*FINAL [\S\s]+ ENERGY IS\s+(\S+) AFTER') _grad_re = re.compile(r'^\s*GRADIENT OF THE ENERGY\s*') _charges_re = re.compile(r'^\s*TOTAL MULLIKEN AND LOWDIN ATOMIC POPULATIONS\s*') _dipole_re = re.compile(r'^\s+DX\s+DY\s+DZ\s+\/D\/\s+\(DEBYE\)')
[docs] @reader def read_gamess_us_out(fd): atoms = None energy = None forces = None charges = None dipole = None for line in fd: # Geometry if _geom_re.match(line): fd.readline() symbols = [] pos = [] while True: atom = _atom_re.match(fd.readline()) if atom is None: break symbol, _, x, y, z = atom.groups() symbols.append(symbol.capitalize()) pos.append(list(map(float, [x, y, z]))) atoms = Atoms(symbols, np.array(pos) * Bohr) continue # Energy ematch = _energy_re.match(line) if ematch is not None: energy = float(ematch.group(1)) * Hartree # MPn energy. Supplants energy parsed above. elif line.strip().startswith('TOTAL ENERGY'): energy = float(line.strip().split()[-1]) * Hartree # Higher-level energy (e.g. coupled cluster) # Supplants energies parsed above. elif line.strip().startswith('THE FOLLOWING METHOD AND ENERGY'): energy = float(fd.readline().strip().split()[-1]) * Hartree elif _charges_re.match(line): fd.readline() charges = np.array( [float(fd.readline().split()[3]) for _ in symbols], ) # Gradients elif _grad_re.match(line): for _ in range(3): fd.readline() grad = [] while True: atom = _atom_re.match(fd.readline()) if atom is None: break grad.append(list(map(float, atom.groups()[2:]))) forces = -np.array(grad) * Hartree / Bohr elif _dipole_re.match(line): dipole = np.array(list(map(float, fd.readline().split()[:3]))) dipole *= Debye atoms.calc = SinglePointCalculator( atoms, energy=energy, forces=forces, charges=charges, dipole=dipole, ) return atoms
[docs] @reader def read_gamess_us_punch(fd): atoms = None energy = None forces = None charges = None dipole = None for line in fd: if line.strip() == '$DATA': symbols = [] pos = [] while line.strip() != '$END': line = fd.readline() atom = _atom_re.match(line) if atom is None: # The basis set specification is interlaced with the # molecular geometry. We don't care about the basis # set, so ignore lines that don't match the pattern. continue symbols.append(atom.group(1).capitalize()) pos.append(list(map(float, atom.group(3, 4, 5)))) atoms = Atoms(symbols, np.array(pos)) elif line.startswith('E('): energy = float(line.split()[1][:-1]) * Hartree elif line.strip().startswith('POPULATION ANALYSIS'): # Mulliken charges charges = np.array( [float(fd.readline().split()[2]) for _ in symbols], ) elif line.strip().startswith('DIPOLE'): dipole = np.array(list(map(float, line.split()[1:]))) * Debye elif line.strip() == '$GRAD': # The gradient block also contains the energy, which we prefer # over the energy obtained above because it is more likely to # be consistent with the gradients. It probably doesn't actually # make a difference though. energy = float(fd.readline().split()[1]) * Hartree grad = [] while line.strip() != '$END': line = fd.readline() atom = _atom_re.match(line) if atom is None: continue grad.append(list(map(float, atom.group(3, 4, 5)))) forces = -np.array(grad) * Hartree / Bohr atoms.calc = SinglePointCalculator( atoms, energy=energy, forces=forces, charges=charges, dipole=dipole, ) return atoms
def clean_userscr(userscr: str, prefix: str) -> None: for fname in os.listdir(userscr): tokens = fname.split('.') if tokens[0] == prefix and tokens[-1] != 'bak': fold = os.path.join(userscr, fname) os.rename(fold, fold + '.bak') def get_userscr(prefix: str, command: str) -> str | None: prefix_test = prefix + '_test' command = command.replace('PREFIX', prefix_test) with workdir(prefix_test, mkdir=True): try: call(command, shell=True, timeout=2) except TimeoutExpired: pass try: with open(f'{prefix_test}.log', encoding='utf-8') as fd: for line in fd: if line.startswith('GAMESS supplementary output files'): return ' '.join(line.split(' ')[8:]).strip() except FileNotFoundError: return None return None