Source code for iodata.formats.fcidump

# IODATA is an input and output module for quantum chemistry.
# Copyright (C) 2011-2019 The IODATA Development Team
#
# This file is part of IODATA.
#
# IODATA is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 3
# of the License, or (at your option) any later version.
#
# IODATA is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, see <http://www.gnu.org/licenses/>
# --
"""Molpro 2012 FCIDUMP file format.

Notes
-----
1. This function works only for restricted wave-functions.
2. One- and two-electron integrals are stored in chemists' notation in an FCIDUMP file,
   while IOData internally uses Physicist's notation.
3. Keep in mind that the FCIDUMP format changed in MOLPRO 2012, so files generated with
   older versions are not supported.

"""


from typing import TextIO

import numpy as np

from ..docstrings import document_load_one, document_dump_one
from ..iodata import IOData
from ..utils import set_four_index_element, LineIterator


__all__ = []


PATTERNS = ['*FCIDUMP*']


[docs]@document_load_one("Molpro 2012 FCIDUMP", ['core_energy', 'one_ints', 'nelec', 'spinpol', 'two_ints']) def load_one(lit: LineIterator) -> dict: """Do not edit this docstring. It will be overwritten.""" # check header line = next(lit) if not line.startswith(' &FCI NORB='): lit.error('Incorrect file header') # read info from header words = line[5:].split(',') header_info = {} for word in words: if word.count('=') == 1: key, value = word.split('=') header_info[key.strip()] = value.strip() nbasis = int(header_info['NORB']) nelec = int(header_info['NELEC']) spinpol = int(header_info['MS2']) # skip rest of header for line in lit: words = line.split() if words[0] == "&END" or words[0] == "/END" or words[0] == "/": break # read the integrals one_mo = np.zeros((nbasis, nbasis)) two_mo = np.zeros((nbasis, nbasis, nbasis, nbasis)) core_energy = 0.0 for line in lit: words = line.split() if len(words) != 5: lit.error('Expecting 5 fields on each data line in FCIDUMP') value = float(words[0]) if words[3] != '0': ii = int(words[1]) - 1 ij = int(words[2]) - 1 ik = int(words[3]) - 1 il = int(words[4]) - 1 # Uncomment the following line if you want to assert that the # FCIDUMP file does not contain duplicate 4-index entries. # assert two_mo.get_element(ii,ik,ij,il) == 0.0 set_four_index_element(two_mo, ii, ik, ij, il, value) elif words[1] != '0': ii = int(words[1]) - 1 ij = int(words[2]) - 1 one_mo[ii, ij] = value one_mo[ij, ii] = value else: core_energy = value return { 'nelec': nelec, 'spinpol': spinpol, 'one_ints': {'core_mo': one_mo}, 'two_ints': {'two_mo': two_mo}, 'core_energy': core_energy, }
LOAD_ONE_NOTES = """ The dictionary ``one_ints`` must contain a field ``core_mo``. Similarly, ``two_ints`` must contain ``two_mo``. """
[docs]@document_dump_one("Molpro 2012 FCIDUMP", ['one_ints', 'two_ints'], ['core_energy', 'nelec', 'spinpol'], {}, LOAD_ONE_NOTES) def dump_one(f: TextIO, data: IOData): """Do not edit this docstring. It will be overwritten.""" one_mo = data.one_ints['core_mo'] # Write header nactive = one_mo.shape[0] nelec = data.nelec or 0 spinpol = data.spinpol or 0 print(f' &FCI NORB={nactive:d},NELEC={nelec:d},MS2={spinpol:d},', file=f) print(f" ORBSYM= {','.join('1' for v in range(nactive))},", file=f) print(' ISYM=1', file=f) print(' &END', file=f) # Write integrals and core energy two_mo = data.two_ints['two_mo'] for i in range(nactive): # pylint: disable=too-many-nested-blocks for j in range(i + 1): for k in range(nactive): for l in range(k + 1): if (i * (i + 1)) / 2 + j >= (k * (k + 1)) / 2 + l: value = two_mo[i, k, j, l] if value != 0.0: print(f'{value:23.16e} {i+1:4d} {j+1:4d} {k+1:4d} {l+1:4d}', file=f) for i in range(nactive): for j in range(i + 1): value = one_mo[i, j] if value != 0.0: print(f'{value:23.16e} {i+1:4d} {j+1:4d} {0:4d} {0:4d}', file=f) if data.core_energy is not None: print(f'{data.core_energy:23.16e} {0:4d} {0:4d} {0:4d} {0:4d}', file=f)