Source code for ase.calculators.calculator

import os
import copy
import subprocess
from math import pi, sqrt

import numpy as np

from ase.utils import jsonable
from ase.dft.kpoints import monkhorst_pack
from ase.cell import Cell


class CalculatorError(RuntimeError):
    """Base class of error types related to ASE calculators."""


class CalculatorSetupError(CalculatorError):
    """Calculation cannot be performed with the given parameters.

    Reasons to raise this errors are:
      * The calculator is not properly configured
        (missing executable, environment variables, ...)
      * The given atoms object is not supported
      * Calculator parameters are unsupported

    Typically raised before a calculation."""


class EnvironmentError(CalculatorSetupError):
    """Raised if calculator is not properly set up with ASE.

    May be missing an executable or environment variables."""


class InputError(CalculatorSetupError):
    """Raised if inputs given to the calculator were incorrect.

    Bad input keywords or values, or missing pseudopotentials.

    This may be raised before or during calculation, depending on
    when the problem is detected."""


class CalculationFailed(CalculatorError):
    """Calculation failed unexpectedly.

    Reasons to raise this error are:
      * Calculation did not converge
      * Calculation ran out of memory
      * Segmentation fault or other abnormal termination
      * Arithmetic trouble (singular matrices, NaN, ...)

    Typically raised during calculation."""


class SCFError(CalculationFailed):
    """SCF loop did not converge."""


class ReadError(CalculatorError):
    """Unexpected irrecoverable error while reading calculation results."""


class PropertyNotImplementedError(NotImplementedError):
    """Raised if a calculator does not implement the requested property."""


class PropertyNotPresent(CalculatorError):
    """Requested property is missing.

    Maybe it was never calculated, or for some reason was not extracted
    with the rest of the results, without being a fatal ReadError."""


def compare_atoms(atoms1, atoms2, tol=1e-15, excluded_properties=None):
    """Check for system changes since last calculation.  Properties in
    ``excluded_properties`` are not checked."""
    if atoms1 is None:
        system_changes = all_changes[:]
    else:
        system_changes = []

        properties_to_check = set(all_changes)
        if excluded_properties:
             properties_to_check -= set(excluded_properties)

        # Check properties that aren't in Atoms.arrays but are attributes of
        # Atoms objects
        for prop in ['cell', 'pbc']:
            if prop in properties_to_check:
                properties_to_check.remove(prop)
                if not equal(getattr(atoms1, prop), getattr(atoms2, prop), tol):
                    system_changes.append(prop)

        arrays1 = set(atoms1.arrays)
        arrays2 = set(atoms2.arrays)

        # Add any properties that are only in atoms1.arrays or only in
        # atoms2.arrays (and aren't excluded).  Note that if, e.g. arrays1 has
        # `initial_charges` which is merely zeros and arrays2 does not have
        # this array, we'll still assume that the system has changed.  However,
        # this should only occur rarely.
        system_changes += properties_to_check & (arrays1 ^ arrays2)

        # Finally, check all of the non-excluded properties shared by the atoms
        # arrays
        for prop in properties_to_check & arrays1 & arrays2:
                if not equal(atoms1.arrays[prop], atoms2.arrays[prop], tol):
                    system_changes.append(prop)

    return system_changes


all_properties = ['energy', 'forces', 'stress', 'stresses', 'dipole',
                  'charges', 'magmom', 'magmoms', 'free_energy', 'energies']


all_changes = ['positions', 'numbers', 'cell', 'pbc',
               'initial_charges', 'initial_magmoms']


# Recognized names of calculators sorted alphabetically:
names = ['abinit', 'ace', 'aims', 'amber', 'asap', 'castep', 'cp2k',
         'crystal', 'demon', 'demonnano', 'dftb', 'dftd3', 'dmol', 'eam', 'elk',
         'emt', 'espresso', 'exciting', 'ff', 'fleur', 'gaussian',
         'gpaw', 'gromacs', 'gulp', 'hotbit', 'jacapo', 'kim',
         'lammpslib', 'lammpsrun', 'lj', 'mopac', 'morse', 'nwchem',
         'octopus', 'onetep', 'openmx', 'psi4', 'qchem', 'siesta',
         'tip3p', 'tip4p', 'turbomole', 'vasp']


special = {'cp2k': 'CP2K',
           'demonnano': 'DemonNano',
           'dftd3': 'DFTD3',
           'dmol': 'DMol3',
           'eam': 'EAM',
           'elk': 'ELK',
           'emt': 'EMT',
           'crystal': 'CRYSTAL',
           'ff': 'ForceField',
           'fleur': 'FLEUR',
           'gulp': 'GULP',
           'kim': 'KIM',
           'lammpsrun': 'LAMMPS',
           'lammpslib': 'LAMMPSlib',
           'lj': 'LennardJones',
           'mopac': 'MOPAC',
           'morse': 'MorsePotential',
           'nwchem': 'NWChem',
           'openmx': 'OpenMX',
           'qchem': 'QChem',
           'tip3p': 'TIP3P',
           'tip4p': 'TIP4P'}


external_calculators = {}


def register_calculator_class(name, cls):
    """ Add the class into the database. """
    assert name not in external_calculators
    external_calculators[name] = cls
    names.append(name)
    names.sort()


def get_calculator_class(name):
    """Return calculator class."""
    if name == 'asap':
        from asap3 import EMT as Calculator
    elif name == 'gpaw':
        from gpaw import GPAW as Calculator
    elif name == 'hotbit':
        from hotbit import Calculator
    elif name == 'vasp2':
        from ase.calculators.vasp import Vasp2 as Calculator
    elif name == 'ace':
        from ase.calculators.acemolecule import ACE as Calculator
    elif name == 'Psi4':
        from ase.calculators.psi4 import Psi4 as Calculator
    elif name in external_calculators:
        Calculator = external_calculators[name]
    else:
        classname = special.get(name, name.title())
        module = __import__('ase.calculators.' + name, {}, None, [classname])
        Calculator = getattr(module, classname)
    return Calculator


def equal(a, b, tol=None):
    """ndarray-enabled comparison function."""
    # XXX Known bugs:
    #  * Comparing cell objects (pbc not part of array representation)
    #  * Infinite recursion for cyclic dicts
    #  * Can of worms is open
    if tol is None:
        return np.array_equal(a, b)

    shape = np.shape(a)
    if shape != np.shape(b):
        return False

    if not shape:
        if isinstance(a, dict) and isinstance(b, dict):
            if a.keys() != b.keys():
                return False
            return all(equal(a[key], b[key], tol) for key in a.keys())
        return abs(a - b) < tol * abs(b) + tol

    return np.allclose(a, b, rtol=tol, atol=tol)


def kptdensity2monkhorstpack(atoms, kptdensity=3.5, even=True):
    """Convert k-point density to Monkhorst-Pack grid size.

    atoms: Atoms object
        Contains unit cell and information about boundary conditions.
    kptdensity: float
        Required k-point density.  Default value is 3.5 point per Ang^-1.
    even: bool
        Round up to even numbers.
    """

    recipcell = atoms.get_reciprocal_cell()
    kpts = []
    for i in range(3):
        if atoms.pbc[i]:
            k = 2 * pi * sqrt((recipcell[i]**2).sum()) * kptdensity
            if even:
                kpts.append(2 * int(np.ceil(k / 2)))
            else:
                kpts.append(int(np.ceil(k)))
        else:
            kpts.append(1)
    return np.array(kpts)


def kpts2mp(atoms, kpts, even=False):
    if kpts is None:
        return np.array([1, 1, 1])
    if isinstance(kpts, (float, int)):
        return kptdensity2monkhorstpack(atoms, kpts, even)
    else:
        return kpts


def kpts2sizeandoffsets(size=None, density=None, gamma=None, even=None,
                        atoms=None):
    """Helper function for selecting k-points.

    Use either size or density.

    size: 3 ints
        Number of k-points.
    density: float
        K-point density in units of k-points per Ang^-1.
    gamma: None or bool
        Should the Gamma-point be included?  Yes / no / don't care:
        True / False / None.
    even: None or bool
        Should the number of k-points be even?  Yes / no / don't care:
        True / False / None.
    atoms: Atoms object
        Needed for calculating k-point density.

    """

    if size is not None and density is not None:
        raise ValueError('Cannot specify k-point mesh size and '
                         'density simultaneously')
    elif density is not None and atoms is None:
        raise ValueError('Cannot set k-points from "density" unless '
                         'Atoms are provided (need BZ dimensions).')

    if size is None:
        if density is None:
            size = [1, 1, 1]
        else:
            size = kptdensity2monkhorstpack(atoms, density, None)

    # Not using the rounding from kptdensity2monkhorstpack as it doesn't do
    # rounding to odd numbers
    if even is not None:
        size = np.array(size)
        remainder = size % 2
        if even:
            size += remainder
        else:  # Round up to odd numbers
            size += (1 - remainder)

    offsets = [0, 0, 0]
    if atoms is None:
        pbc = [True, True, True]
    else:
        pbc = atoms.pbc

    if gamma is not None:
        for i, s in enumerate(size):
            if pbc[i] and s % 2 != bool(gamma):
                offsets[i] = 0.5 / s

    return size, offsets

@jsonable('kpoints')
class KPoints:
    def __init__(self, kpts=None):
        if kpts is None:
            kpts = np.zeros((1, 3))
        self.kpts = kpts

    def todict(self):
        return vars(self)


def kpts2kpts(kpts, atoms=None):
    if kpts is None:
        return KPoints()

    if hasattr(kpts, 'kpts'):
        return kpts

    if isinstance(kpts, dict):
        if 'kpts' in kpts:
            return KPoints(kpts['kpts'])
        if 'path' in kpts:
            cell = Cell.ascell(atoms.cell)
            return cell.bandpath(pbc=atoms.pbc, **kpts)
        size, offsets = kpts2sizeandoffsets(atoms=atoms, **kpts)
        return KPoints(monkhorst_pack(size) + offsets)

    if isinstance(kpts[0], int):
        return KPoints(monkhorst_pack(kpts))

    return KPoints(np.array(kpts))


def kpts2ndarray(kpts, atoms=None):
    """Convert kpts keyword to 2-d ndarray of scaled k-points."""
    return kpts2kpts(kpts, atoms=atoms).kpts


class EigenvalOccupationMixin:
    """Define 'eigenvalues' and 'occupations' properties on class.

    eigenvalues and occupations will be arrays of shape (spin, kpts, nbands).

    Classes must implement the old-fashioned get_eigenvalues and
    get_occupations methods."""

    @property
    def eigenvalues(self):
        return self.build_eig_occ_array(self.get_eigenvalues)

    @property
    def occupations(self):
        return self.build_eig_occ_array(self.get_occupation_numbers)

    def build_eig_occ_array(self, getter):
        nspins = self.get_number_of_spins()
        nkpts = len(self.get_ibz_k_points())
        nbands = self.get_number_of_bands()
        arr = np.zeros((nspins, nkpts, nbands))
        for s in range(nspins):
            for k in range(nkpts):
                arr[s, k, :] = getter(spin=s, kpt=k)
        return arr


class Parameters(dict):
    """Dictionary for parameters.

    Special feature: If param is a Parameters instance, then param.xc
    is a shorthand for param['xc'].
    """

    def __getattr__(self, key):
        if key not in self:
            return dict.__getattribute__(self, key)
        return self[key]

    def __setattr__(self, key, value):
        self[key] = value

    @classmethod
    def read(cls, filename):
        """Read parameters from file."""
        # We use ast to evaluate literals, avoiding eval()
        # for security reasons.
        import ast
        with open(filename) as fd:
            txt = fd.read().strip()
        assert txt.startswith('dict(')
        assert txt.endswith(')')
        txt = txt[5:-1]

        # The tostring() representation "dict(...)" is not actually
        # a literal, so we manually parse that along with the other
        # formatting that we did manually:
        dct = {}
        for line in txt.splitlines():
            key, val = line.split('=', 1)
            key = key.strip()
            val = val.strip()
            if val[-1] == ',':
                val = val[:-1]
            dct[key] = ast.literal_eval(val)

        parameters = cls(dct)
        return parameters

    def tostring(self):
        keys = sorted(self)
        return 'dict(' + ',\n     '.join(
            '{}={!r}'.format(key, self[key]) for key in keys) + ')\n'

    def write(self, filename):
        file = open(filename, 'w')
        file.write(self.tostring())
        file.close()


[docs]class Calculator(object): """Base-class for all ASE calculators. A calculator must raise PropertyNotImplementedError if asked for a property that it can't calculate. So, if calculation of the stress tensor has not been implemented, get_stress(atoms) should raise PropertyNotImplementedError. This can be achieved simply by not including the string 'stress' in the list implemented_properties which is a class member. These are the names of the standard properties: 'energy', 'forces', 'stress', 'dipole', 'charges', 'magmom' and 'magmoms'. """ implemented_properties = [] 'Properties calculator can handle (energy, forces, ...)' default_parameters = {} 'Default parameters' def __init__(self, restart=None, ignore_bad_restart_file=False, label=None, atoms=None, directory='.', **kwargs): """Basic calculator implementation. restart: str Prefix for restart file. May contain a directory. Default is None: don't restart. ignore_bad_restart_file: bool Ignore broken or missing restart file. By default, it is an error if the restart file is missing or broken. directory: str Working directory in which to read and write files and perform calculations. label: str Name used for all files. Not supported by all calculators. May contain a directory, but please use the directory parameter for that instead. atoms: Atoms object Optional Atoms object to which the calculator will be attached. When restarting, atoms will get its positions and unit-cell updated from file. """ self.atoms = None # copy of atoms object from last calculation self.results = {} # calculated properties (energy, forces, ...) self.parameters = None # calculational parameters if restart is not None: try: self.read(restart) # read parameters, atoms and results except ReadError: if ignore_bad_restart_file: self.reset() else: raise self.directory = directory self.prefix = None if label is not None: if directory != '.' and '/' in label: raise ValueError('Directory redundantly specified though ' 'directory="{}" and label="{}". ' 'Please omit "/" in label.' .format(directory, label)) self.set_label(label) if self.parameters is None: # Use default parameters if they were not read from file: self.parameters = self.get_default_parameters() if atoms is not None: atoms.calc = self if self.atoms is not None: # Atoms were read from file. Update atoms: if not (equal(atoms.numbers, self.atoms.numbers) and (atoms.pbc == self.atoms.pbc).all()): raise CalculatorError('Atoms not compatible with file') atoms.positions = self.atoms.positions atoms.cell = self.atoms.cell self.set(**kwargs) if not hasattr(self, 'name'): self.name = self.__class__.__name__.lower() @property def label(self): if self.directory == '.': return self.prefix # Generally, label ~ directory/prefix # # We use '/' rather than os.pathsep because # 1) directory/prefix does not represent any actual path # 2) We want the same string to work the same on all platforms if self.prefix is None: return self.directory + '/' return '{}/{}'.format(self.directory, self.prefix) @label.setter def label(self, label): if label is None: self.directory = '.' self.prefix = None return tokens = label.rsplit('/', 1) if len(tokens) == 2: directory, prefix = tokens else: assert len(tokens) == 1 directory = '.' prefix = tokens[0] if prefix == '': prefix = None self.directory = directory self.prefix = prefix
[docs] def set_label(self, label): """Set label and convert label to directory and prefix. Examples: * label='abc': (directory='.', prefix='abc') * label='dir1/abc': (directory='dir1', prefix='abc') * label=None: (directory='.', prefix=None) Calculators that must write results to files with fixed names can override this method so that the directory is set to all of label.""" self.label = label
def get_default_parameters(self): return Parameters(copy.deepcopy(self.default_parameters)) def todict(self, skip_default=True): defaults = self.get_default_parameters() dct = {} for key, value in self.parameters.items(): if hasattr(value, 'todict'): value = value.todict() if skip_default: default = defaults.get(key, '_no_default_') if default != '_no_default_' and equal(value, default): continue dct[key] = value return dct
[docs] def reset(self): """Clear all information from old calculation.""" self.atoms = None self.results = {}
[docs] def read(self, label): """Read atoms, parameters and calculated properties from output file. Read result from self.label file. Raise ReadError if the file is not there. If the file is corrupted or contains an error message from the calculation, a ReadError should also be raised. In case of succes, these attributes must set: atoms: Atoms object The state of the atoms from last calculation. parameters: Parameters object The parameter dictionary. results: dict Calculated properties like energy and forces. The FileIOCalculator.read() method will typically read atoms and parameters and get the results dict by calling the read_results() method.""" self.set_label(label)
def get_atoms(self): if self.atoms is None: raise ValueError('Calculator has no atoms') atoms = self.atoms.copy() atoms.calc = self return atoms @classmethod def read_atoms(cls, restart, **kwargs): return cls(restart=restart, label=restart, **kwargs).get_atoms()
[docs] def set(self, **kwargs): """Set parameters like set(key1=value1, key2=value2, ...). A dictionary containing the parameters that have been changed is returned. Subclasses must implement a set() method that will look at the chaneged parameters and decide if a call to reset() is needed. If the changed parameters are harmless, like a change in verbosity, then there is no need to call reset(). The special keyword 'parameters' can be used to read parameters from a file.""" if 'parameters' in kwargs: filename = kwargs.pop('parameters') parameters = Parameters.read(filename) parameters.update(kwargs) kwargs = parameters changed_parameters = {} for key, value in kwargs.items(): oldvalue = self.parameters.get(key) if key not in self.parameters or not equal(value, oldvalue): changed_parameters[key] = value self.parameters[key] = value return changed_parameters
[docs] def check_state(self, atoms, tol=1e-15): """Check for any system changes since last calculation.""" return compare_atoms(self.atoms, atoms, tol)
def get_potential_energy(self, atoms=None, force_consistent=False): energy = self.get_property('energy', atoms) if force_consistent: if 'free_energy' not in self.results: name = self.__class__.__name__ # XXX but we don't know why the energy is not there. # We should raise PropertyNotPresent. Discuss raise PropertyNotImplementedError( 'Force consistent/free energy ("free_energy") ' 'not provided by {0} calculator'.format(name)) return self.results['free_energy'] else: return energy def get_potential_energies(self, atoms=None): return self.get_property('energies', atoms) def get_forces(self, atoms=None): return self.get_property('forces', atoms) def get_stress(self, atoms=None): return self.get_property('stress', atoms) def get_stresses(self, atoms=None): return self.get_property('stresses', atoms) def get_dipole_moment(self, atoms=None): return self.get_property('dipole', atoms) def get_charges(self, atoms=None): return self.get_property('charges', atoms) def get_magnetic_moment(self, atoms=None): return self.get_property('magmom', atoms)
[docs] def get_magnetic_moments(self, atoms=None): """Calculate magnetic moments projected onto atoms.""" return self.get_property('magmoms', atoms)
def get_property(self, name, atoms=None, allow_calculation=True): if name not in self.implemented_properties: raise PropertyNotImplementedError('{} property not implemented' .format(name)) if atoms is None: atoms = self.atoms system_changes = [] else: system_changes = self.check_state(atoms) if system_changes: self.reset() if name not in self.results: if not allow_calculation: return None self.calculate(atoms, [name], system_changes) if name == 'magmom' and 'magmom' not in self.results: return 0.0 if name == 'magmoms' and 'magmoms' not in self.results: return np.zeros(len(atoms)) if name not in self.results: # For some reason the calculator was not able to do what we want, # and that is OK. raise PropertyNotImplementedError('{} not present in this ' 'calculation'.format(name)) result = self.results[name] if isinstance(result, np.ndarray): result = result.copy() return result def calculation_required(self, atoms, properties): assert not isinstance(properties, str) system_changes = self.check_state(atoms) if system_changes: return True for name in properties: if name not in self.results: return True return False
[docs] def calculate(self, atoms=None, properties=['energy'], system_changes=all_changes): """Do the calculation. properties: list of str List of what needs to be calculated. Can be any combination of 'energy', 'forces', 'stress', 'dipole', 'charges', 'magmom' and 'magmoms'. system_changes: list of str List of what has changed since last calculation. Can be any combination of these six: 'positions', 'numbers', 'cell', 'pbc', 'initial_charges' and 'initial_magmoms'. Subclasses need to implement this, but can ignore properties and system_changes if they want. Calculated properties should be inserted into results dictionary like shown in this dummy example:: self.results = {'energy': 0.0, 'forces': np.zeros((len(atoms), 3)), 'stress': np.zeros(6), 'dipole': np.zeros(3), 'charges': np.zeros(len(atoms)), 'magmom': 0.0, 'magmoms': np.zeros(len(atoms))} The subclass implementation should first call this implementation to set the atoms attribute. """ if atoms is not None: self.atoms = atoms.copy()
[docs] def calculate_numerical_forces(self, atoms, d=0.001): """Calculate numerical forces using finite difference. All atoms will be displaced by +d and -d in all directions.""" from ase.calculators.test import numeric_force return np.array([[numeric_force(atoms, a, i, d) for i in range(3)] for a in range(len(atoms))])
[docs] def calculate_numerical_stress(self, atoms, d=1e-6, voigt=True): """Calculate numerical stress using finite difference.""" stress = np.zeros((3, 3), dtype=float) cell = atoms.cell.copy() V = atoms.get_volume() for i in range(3): x = np.eye(3) x[i, i] += d atoms.set_cell(np.dot(cell, x), scale_atoms=True) eplus = atoms.get_potential_energy(force_consistent=True) x[i, i] -= 2 * d atoms.set_cell(np.dot(cell, x), scale_atoms=True) eminus = atoms.get_potential_energy(force_consistent=True) stress[i, i] = (eplus - eminus) / (2 * d * V) x[i, i] += d j = i - 2 x[i, j] = d x[j, i] = d atoms.set_cell(np.dot(cell, x), scale_atoms=True) eplus = atoms.get_potential_energy(force_consistent=True) x[i, j] = -d x[j, i] = -d atoms.set_cell(np.dot(cell, x), scale_atoms=True) eminus = atoms.get_potential_energy(force_consistent=True) stress[i, j] = (eplus - eminus) / (4 * d * V) stress[j, i] = stress[i, j] atoms.set_cell(cell, scale_atoms=True) if voigt: return stress.flat[[0, 4, 8, 5, 2, 1]] else: return stress
def get_spin_polarized(self): return False
[docs] def band_structure(self): """Create band-structure object for plotting.""" from ase.dft.band_structure import get_band_structure # XXX This calculator is supposed to just have done a band structure # calculation, but the calculator may not have the correct Fermi level # if it updated the Fermi level after changing k-points. # This will be a problem with some calculators (currently GPAW), and # the user would have to override this by providing the Fermi level # from the selfconsistent calculation. return get_band_structure(calc=self)
[docs]class FileIOCalculator(Calculator): """Base class for calculators that write/read input/output files.""" command = None # str 'Command used to start calculation' def __init__(self, restart=None, ignore_bad_restart_file=False, label=None, atoms=None, command=None, **kwargs): """File-IO calculator. command: str Command used to start calculation. """ Calculator.__init__(self, restart, ignore_bad_restart_file, label, atoms, **kwargs) if command is not None: self.command = command else: name = 'ASE_' + self.name.upper() + '_COMMAND' self.command = os.environ.get(name, self.command)
[docs] def calculate(self, atoms=None, properties=['energy'], system_changes=all_changes): Calculator.calculate(self, atoms, properties, system_changes) self.write_input(self.atoms, properties, system_changes) if self.command is None: raise CalculatorSetupError( 'Please set ${} environment variable ' .format('ASE_' + self.name.upper() + '_COMMAND') + 'or supply the command keyword') command = self.command if 'PREFIX' in command: command = command.replace('PREFIX', self.prefix) try: proc = subprocess.Popen(command, shell=True, cwd=self.directory) except OSError as err: # Actually this may never happen with shell=True, since # probably the shell launches successfully. But we soon want # to allow calling the subprocess directly, and then this # distinction (failed to launch vs failed to run) is useful. msg = 'Failed to execute "{}"'.format(command) raise EnvironmentError(msg) from err errorcode = proc.wait() if errorcode: path = os.path.abspath(self.directory) msg = ('Calculator "{}" failed with command "{}" failed in ' '{} with error code {}'.format(self.name, command, path, errorcode)) raise CalculationFailed(msg) self.read_results()
[docs] def write_input(self, atoms, properties=None, system_changes=None): """Write input file(s). Call this method first in subclasses so that directories are created automatically.""" absdir = os.path.abspath(self.directory) if absdir != os.curdir and not os.path.isdir(self.directory): os.makedirs(self.directory)
[docs] def read_results(self): """Read energy, forces, ... from output file(s).""" pass