atomistic.py

# coding: utf-8
# Copyright (c) Max-Planck-Institut für Eisenforschung GmbH - Computational Materials Design (CM) Department
# Distributed under the terms of "New BSD License", see the LICENSE file.

from ase.io import write as ase_write
import copy

import numpy as np
import warnings

from pyiron.atomistics.structure.atoms import Atoms
from pyiron_base.generic.parameters import GenericParameters
from pyiron_base.job.generic import GenericJob as GenericJobCore
from pyiron_base.master.generic import GenericMaster

try:
    from pyiron.base.project import ProjectGUI
except (ImportError, TypeError, AttributeError):
    pass

__author__ = "Jan Janssen"
__copyright__ = (
    "Copyright 2020, Max-Planck-Institut für Eisenforschung GmbH - "
    "Computational Materials Design (CM) Department"
)
__version__ = "1.0"
__maintainer__ = "Jan Janssen"
__email__ = "janssen@mpie.de"
__status__ = "production"
__date__ = "Sep 1, 2017"


class AtomisticGenericJob(GenericJobCore):
    """
    Atomistic Generic Job class extends the Generic Job class with all the functionality to run jobs containing
    atomistic structures. From this class all specific atomistic Hamiltonians are derived. Therefore it should contain
    the properties/routines common to all atomistic jobs. The functions in this module should be as generic as possible.

    Args:
        project (ProjectHDFio): ProjectHDFio instance which points to the HDF5 file the job is stored in
        job_name (str): name of the job, which has to be unique within the project

    Attributes:

        .. attribute:: job_name

            name of the job, which has to be unique within the project

        .. attribute:: status

            execution status of the job, can be one of the following [initialized, appended, created, submitted, running,
                                                                      aborted, collect, suspended, refresh, busy, finished]

        .. attribute:: job_id

            unique id to identify the job in the pyiron database

        .. attribute:: parent_id

            job id of the predecessor job - the job which was executed before the current one in the current job series

        .. attribute:: master_id

            job id of the master job - a meta job which groups a series of jobs, which are executed either in parallel or in
            serial.

        .. attribute:: child_ids

            list of child job ids - only meta jobs have child jobs - jobs which list the meta job as their master

        .. attribute:: project

            Project instance the jobs is located in

        .. attribute:: project_hdf5

            ProjectHDFio instance which points to the HDF5 file the job is stored in

        .. attribute:: job_info_str

            short string to describe the job by it is job_name and job ID - mainly used for logging

        .. attribute:: working_directory

            working directory of the job is executed in - outside the HDF5 file

        .. attribute:: path

            path to the job as a combination of absolute file system path and path within the HDF5 file.

        .. attribute:: version

            Version of the hamiltonian, which is also the version of the executable unless a custom executable is used.

        .. attribute:: executable

            Executable used to run the job - usually the path to an external executable.

        .. attribute:: library_activated

            For job types which offer a Python library pyiron can use the python library instead of an external executable.

        .. attribute:: server

            Server object to handle the execution environment for the job.

        .. attribute:: queue_id

            the ID returned from the queuing system - it is most likely not the same as the job ID.

        .. attribute:: logger

            logger object to monitor the external execution and internal pyiron warnings.

        .. attribute:: restart_file_list

            list of files which are used to restart the calculation from these files.

        .. attribute:: job_type

            Job type object with all the available job types: ['ExampleJob', 'SerialMaster', 'ParallelMaster', 'ScriptJob',
                                                               'ListMaster']
    """

    def __init__(self, project, job_name):
        super(AtomisticGenericJob, self).__init__(project, job_name)
        self.__name__ = "AtomisticGenericJob"
        self.__version__ = "0.1"
        self._structure = None
        self._generic_input = GenericInput()
        self.output = GenericOutput(job=self)
        self.map_functions = MapFunctions()

    @property
    def structure(self):
        """

        Returns:

        """
        return self._structure

    @structure.setter
    def structure(self, basis):
        """

        Args:
            basis:

        Returns:

        """
        self._generic_input["structure"] = "atoms"
        self._structure = basis

    def set_input_to_read_only(self):
        """
        This function enforces read-only mode for the input classes, but it has to be implement in the individual
        classes.
        """
        self._generic_input.read_only = True

    def copy_to(
        self, project=None, new_job_name=None, input_only=False, new_database_entry=True
    ):
        """

        Args:
            destination:
            new_job_name:
            input_only:
            new_database_entry:

        Returns:

        """
        new_generic_job = super(AtomisticGenericJob, self).copy_to(
            project=project,
            new_job_name=new_job_name,
            input_only=input_only,
            new_database_entry=new_database_entry,
        )
        if not new_generic_job._structure:
            new_generic_job._structure = copy.copy(self._structure)
        return new_generic_job

    def calc_minimize(
        self, ionic_energy_tolerance=0, ionic_force_tolerance=1e-4, e_tol=None, f_tol=None, max_iter=1000, pressure=None, n_print=1
    ):
        """

        Args:
            ionic_energy_tolerance (float): Maximum energy difference between 2 steps
            ionic_force_tolerance (float): Maximum force magnitude that each of atoms is allowed to have
            e_tol (float): same as ionic_energy_tolerance (deprecated)
            f_tol (float): same as ionic_force_tolerance (deprecated)
            max_iter (int): Maximum number of force evluations
            pressure (float/list): Targetpressure values
            n_print (int): Print period

        Returns:

        """
        if e_tol is not None:
            warnings.warn(
                "e_tol is deprecated as of vers. 0.3.0. It is not guaranteed to be in service in vers. 0.4.0"
            )
        if f_tol is not None:
            warnings.warn(
                "f_tol is deprecated as of vers. 0.3.0. It is not guaranteed to be in service in vers. 0.4.0"
            )
        self._generic_input["calc_mode"] = "minimize"
        self._generic_input["max_iter"] = max_iter
        self._generic_input["pressure"] = pressure
        self._generic_input.remove_keys(
            ["temperature", "n_ionic_steps", "n_print", "velocity"]
        )

    def calc_static(self):
        """

        Returns:

        """
        self._generic_input["calc_mode"] = "static"
        self._generic_input.remove_keys(
            [
                "max_iter",
                "pressure",
                "temperature",
                "n_ionic_steps",
                "n_print",
                "velocity",
            ]
        )

    def calc_md(
        self,
        temperature=None,
        pressure=None,
        n_ionic_steps=1000,
        time_step=None,
        n_print=100,
        temperature_damping_timescale=100.0,
        pressure_damping_timescale=None,
        seed=None,
        tloop=None,
        initial_temperature=True,
        langevin=False,
    ):
        self._generic_input["calc_mode"] = "md"
        self._generic_input["temperature"] = temperature
        self._generic_input["n_ionic_steps"] = n_ionic_steps
        self._generic_input["n_print"] = n_print
        self._generic_input.remove_keys(["max_iter", "pressure"])

    def from_hdf(self, hdf=None, group_name=None):
        """
        Recreates instance from the hdf5 file
        Args:
            hdf (str): Path to the hdf5 file
            group_name (str): Name of the group which contains the object
        """
        super(AtomisticGenericJob, self).from_hdf(hdf=hdf, group_name=group_name)
        with self._hdf5.open("input") as hdf5_input:
            try:
                self._generic_input.from_hdf(hdf5_input)
            except ValueError:
                pass

    def to_hdf(self, hdf=None, group_name=None):
        """
        Store the GenericJob in an HDF5 file

        Args:
            hdf (ProjectHDFio): HDF5 group object - optional
            group_name (str): HDF5 subgroup name - optional
        """
        super(AtomisticGenericJob, self).to_hdf(hdf=hdf, group_name=group_name)
        with self._hdf5.open("input") as hdf5_input:
            self._generic_input.to_hdf(hdf5_input)

    def store_structure(self):
        """
        Create :class:`~.StructureContainer` job with the initial structure of
        the job and sets that jobs :attr:`~.parent_id` from this job.

        Returns:
            :class:`~.StructureContainer`: job containing initial structure of
            this job
        """
        if self.structure is not None:
            structure_container = self.create_job(
                job_type=self.project.job_type.StructureContainer,
                job_name=self.job_name + "_structure",
            )
            structure_container.structure = self.structure
            self.parent_id = structure_container.job_id
            return structure_container
        else:
            ValueError("There is no structure attached to the current Job.")

    def animate_structure(
        self,
        spacefill=True,
        show_cell=True,
        stride=1,
        center_of_mass=False,
        particle_size=0.5,
    ):
        """
        Animates the job if a trajectory is present

        Args:
            spacefill (bool):
            show_cell (bool):
            stride (int): show animation every stride [::stride]
                          use value >1 to make animation faster
                           default=1
            center_of_mass (bool):

        Returns:
            animation: nglview IPython widget

        """
        try:
            import nglview
        except ImportError:
            raise ImportError(
                "The animate() function requires the package nglview to be installed"
            )

        animation = nglview.show_asetraj(
            self.trajectory(stride=stride, center_of_mass=center_of_mass)
        )
        if spacefill:
            animation.add_spacefill(radius_type="vdw", scale=0.5, radius=particle_size)
            animation.remove_ball_and_stick()
        else:
            animation.add_ball_and_stick()
        if show_cell:
            if self.structure.cell is not None:
                animation.add_unitcell()
        return animation

    def view_structure(self, snapshot=-1, spacefill=True, show_cell=True):
        """

        Args:
            snapshot (int): Snapshot of the trajectory one wants
            spacefill (bool):
            show_cell (bool):

        Returns:
            view: nglview IPython widget

        """
        import nglview

        atoms = self.get_structure(snapshot)
        picture = nglview.show_ase(atoms)
        if spacefill:
            picture.add_spacefill(radius_type="vdw", scale=0.5)
            picture.remove_ball_and_stick()
        else:
            picture.add_ball_and_stick()
        if show_cell:
            if atoms.cell is not None:
                picture.add_unitcell()
        return picture

    def validate_ready_to_run(self):
        """

        Returns:

        """
        if not self.structure and self._generic_input["structure"] == "atoms":
            raise ValueError(
                "This job does not contain a valid structure: {}".format(self.job_name)
            )

    def db_entry(self):
        """
        Generate the initial database entry

        Returns:
            (dict): db_dict
        """
        db_dict = super(AtomisticGenericJob, self).db_entry()
        if self.structure:
            if isinstance(self.structure, Atoms):
                parent_structure = self.structure.get_parent_basis()
            else:
                parent_structure = self.structure.copy()
            db_dict["ChemicalFormula"] = parent_structure.get_chemical_formula()
        return db_dict

    def restart(self, job_name=None, job_type=None):
        """
        Restart a new job created from an existing calculation.
        Args:
            project (pyiron.project.Project instance): Project instance at which the new job should be created
            job_name (str): Job name
            job_type (str): Job type

        Returns:
            new_ham: New job
        """
        new_ham = super(AtomisticGenericJob, self).restart(
            job_name=job_name, job_type=job_type
        )
        if isinstance(new_ham, GenericMaster) and not isinstance(self, GenericMaster):
            new_child = self.restart(job_name=None, job_type=None)
            new_ham.append(new_child)
        new_ham.structure = self.get_structure(iteration_step=-1)
        if new_ham.structure is None:
            new_ham.structure = self.structure.copy()
        new_ham._generic_input['structure'] = 'atoms'
        return new_ham

    # Required functions
    def continue_with_restart_files(self, job_type=None, job_name=None):
        """

        Args:
            job_type:
            job_name:

        Returns:

        """
        if job_name is None:
            job_name = "{}_continue".format(self.job_name)
        new_ham = self.restart(job_type=job_type, job_name=job_name)
        if self.status.initialized:
            self._job_id = self.save()
        new_ham.parent_id = self.job_id
        new_ham._generic_input["structure"] = "continue_final"
        return new_ham

    def continue_with_final_structure(self, job_type=None, job_name=None):
        """

        Args:
            job_type:
            job_name:

        Returns:

        """
        if job_name is None:
            job_name = "{}_continue".format(self.job_name)
        if job_type is None:
            job_type = self.__name__
        new_ham = self.create_job(job_type, job_name)
        if self.status.initialized:
            self._job_id = self.save()
        new_ham.parent_id = self.job_id
        if self.status.finished:
            new_ham.structure = self.get_structure(iteration_step=-1)
            new_ham._generic_input["structure"] = "atoms"
        else:
            new_ham._generic_input["structure"] = "continue_final"
        return new_ham

    def trajectory(
        self, stride=1, center_of_mass=False, atom_indices=None,
            snapshot_indices=None, overwrite_positions=None, overwrite_cells=None
    ):
        """

        Args:
            stride (int): The trajectories are generated with every 'stride' steps
            center_of_mass (list/numpy.ndarray): The center of mass
            atom_indices (list/numpy.ndarray): The atom indices for which the trajectory should be generated
            snapshot_indices (list/numpy.ndarray): The snapshots for which the trajectory should be generated
            overwrite_positions (list/numpy.ndarray): List of positions that are meant to overwrite the existing
                                                      trajectory. Useful to wrap coordinates for example
            overwrite_cells(list/numpy.ndarray): List of cells that are meant to overwrite the existing
                                                 trajectory. Only used when `overwrite_positions` is defined. This must
                                                 have the same length of `overwrite_positions`

        Returns:
            pyiron.atomistics.job.atomistic.Trajectory: Trajectory instance

        """
        cells = self.output.cells
        if len(self.output.indices) != 0:
            indices = self.output.indices
        else:
            indices = [self.structure.indices] * len(cells)  # Use the same indices throughout
        if overwrite_positions is not None:
            positions = np.array(overwrite_positions).copy()
            if overwrite_cells is not None:
                if overwrite_cells.shape == (len(positions), 3, 3):
                    cells = np.array(overwrite_cells).copy()
                else:
                    raise ValueError("overwrite_cells must be compatible with the positions!")
        else:
            positions = self.output.positions.copy()
        conditions = list()
        if isinstance(cells, (list, np.ndarray)):
            if len(cells) == 0:
                conditions.append(True)
            else:
                conditions.append(cells[0] is None)
        conditions.append(cells is None)
        if any(conditions):
            max_pos = np.max(np.max(positions, axis=0), axis=0)
            max_pos[np.abs(max_pos) < 1e-2] = 10
            cell = np.eye(3) * max_pos
            cells = np.array([cell] * len(positions))

        if len(positions) != len(cells):
            raise ValueError("The positions must have the same length as the cells!")
        if snapshot_indices is not None:
            positions = positions[snapshot_indices]
            cells = cells[snapshot_indices]
            indices = indices[snapshot_indices]
        if atom_indices is None:
            return Trajectory(
                positions[::stride],
                self.structure.get_parent_basis(),
                center_of_mass=center_of_mass,
                cells=cells[::stride],
                indices=indices[::stride]
            )
        else:
            sub_struct = self.structure.get_parent_basis()[atom_indices]
            if len(sub_struct.species) < len(self.structure.species):
                # Then `sub_struct` has had its indices remapped so they run from 0 to the number of species - 1
                # But the `indices` array is unaware of this and needs to be remapped to this new space
                original_symbols = np.array([el.Abbreviation for el in self.structure.species])
                sub_symbols = np.array([el.Abbreviation for el in sub_struct.species])

                map_ = np.array([np.argwhere(original_symbols == symbol)[0, 0] for symbol in sub_symbols], dtype=int)

                remapped_indices = np.array(indices)
                for i_sub, i_original in enumerate(map_):
                    np.place(remapped_indices, indices == i_original, i_sub)
            else:
                remapped_indices = indices

            return Trajectory(
                positions[::stride, atom_indices, :],
                sub_struct,
                center_of_mass=center_of_mass,
                cells=cells[::stride],
                indices=remapped_indices[::stride, atom_indices]
            )

    def write_traj(
        self,
        filename,
        file_format=None,
        parallel=True,
        append=False,
        stride=1,
        center_of_mass=False,
        atom_indices=None,
        snapshot_indices=None,
        overwrite_positions=None,
        overwrite_cells=None,
        **kwargs
    ):
        """
        Writes the trajectory in a given file file_format based on the `ase.io.write`_ function.

        Args:
            filename (str): Filename of the output
            file_format (str): The specific file_format of the output
            parallel (bool): ase parameter
            append (bool): ase parameter
            stride (int): Writes trajectory every `stride` steps
            center_of_mass (bool): True if the positions are centered on the COM
            atom_indices (list/numpy.ndarray): The atom indices for which the trajectory should be generated
            snapshot_indices (list/numpy.ndarray): The snapshots for which the trajectory should be generated
            overwrite_positions (list/numpy.ndarray): List of positions that are meant to overwrite the existing
                                                      trajectory. Useful to wrap coordinates for example
            overwrite_cells(list/numpy.ndarray): List of cells that are meant to overwrite the existing
                                                 trajectory. Only used when `overwrite_positions` is defined. This must
                                                 have the same length of `overwrite_positions`
            **kwargs: Additional ase arguments

        .. _ase.io.write: https://wiki.fysik.dtu.dk/ase/_modules/ase/io/formats.html#write
        """
        traj = self.trajectory(
            stride=stride,
            center_of_mass=center_of_mass,
            atom_indices=atom_indices,
            snapshot_indices=snapshot_indices,
            overwrite_positions=overwrite_positions,
            overwrite_cells=overwrite_cells
        )
        # Using thr ASE output writer
        ase_write(
            filename=filename,
            images=traj,
            format=file_format,
            parallel=parallel,
            append=append,
            **kwargs
        )

    # Compatibility functions
    def get_final_structure(self):
        """

        Returns:

        """
        warnings.warn(
            "get_final_structure() is deprecated - please use get_structure() instead.",
            DeprecationWarning,
        )
        return self.get_structure(iteration_step=-1)

    def get_structure(self, iteration_step=-1, wrap_atoms=True):
        """
        Gets the structure from a given iteration step of the simulation (MD/ionic relaxation). For static calculations
        there is only one ionic iteration step
        Args:
            iteration_step (int): Step for which the structure is requested
            wrap_atoms (bool): True if the atoms are to be wrapped back into the unit cell

        Returns:
            pyiron.atomistics.structure.atoms.Atoms: The required structure
        """
        if not (self.structure is not None):
            raise AssertionError()
        snapshot = self.structure.copy()
        conditions = list()
        if isinstance(self.output.cells, (list, np.ndarray)):
            if len(self.output.cells) == 0:
                conditions.append(True)
            else:
                conditions.append(self.output.cells[0] is None)
        if self.output.positions is not None and self.output.cells is None:
            conditions.append(self.output.cells is None)
        if any(conditions):
            snapshot.cell = None
        elif self.output.cells is not None:
            snapshot.cell = self.output.cells[iteration_step]
        if self.output.positions is not None:
            snapshot.positions = self.output.positions[iteration_step]
        indices = self.output.indices
        if indices is not None and len(indices) > max([iteration_step, 0]):
            snapshot.indices = indices[iteration_step]
        if wrap_atoms:
            return snapshot.center_coordinates_in_unit_cell()
        else:
            if len(self.output.unwrapped_positions) > max([iteration_step, 0]):
                snapshot.positions = self.output.unwrapped_positions[iteration_step]
            else:
                snapshot.positions += self.output.total_displacements[iteration_step]
            return snapshot

    def map(self, function, parameter_lst):
        master = self.create_job(
            job_type=self.project.job_type.MapMaster, job_name="map_" + self.job_name
        )
        master.modify_function = function
        master.parameter_list = parameter_lst
        return master

    def gui(self):
        """

        Returns:

        """
        ProjectGUI(self)

    def _structure_to_hdf(self):
        if self.structure is not None and self._generic_input["structure"] == "atoms":
            with self.project_hdf5.open("input") as hdf5_input:
                self.structure.to_hdf(hdf5_input)

    def _structure_from_hdf(self):
        if (
            "structure" in self.project_hdf5["input"].list_groups()
            and self._generic_input["structure"] == "atoms"
        ):
            with self.project_hdf5.open("input") as hdf5_input:
                self.structure = Atoms().from_hdf(hdf5_input)

    def _write_chemical_formular_to_database(self):
        if self.structure:
            parent_structure = self.structure.get_parent_basis()
            self.project.db.item_update(
                {"ChemicalFormula": parent_structure.get_chemical_formula()},
                self._job_id,
            )

    def _before_successor_calc(self, ham):
        if ham._generic_input["structure"] == "continue_final":
            ham.structure = self.get_structure(iteration_step=-1)
            ham.to_hdf()


def set_structure(job, parameter):
    job.structure = parameter
    return job


class MapFunctions(object):
    def __init__(self):
        self.set_structure = set_structure


class Trajectory(object):
    """
    A trajectory instance compatible with the ase.io class

    Args:
        positions (numpy.ndarray): The array of the trajectory in cartesian coordinates
        structure (pyiron.atomistics.structure.atoms.Atoms): The initial structure instance from which the species info
                                                             is derived
        center_of_mass (bool): False (default) if the specified positions are w.r.t. the origin
        cells (numpy.ndarray): Optional argument of the cell shape at every time step (Nx3x3 array) when the volume
                                varies
    """

    def __init__(self, positions, structure, center_of_mass=False, cells=None, indices=None):
        if center_of_mass:
            pos = np.copy(positions)
            pos[:, :, 0] = (pos[:, :, 0].T - np.mean(pos[:, :, 0], axis=1)).T
            pos[:, :, 1] = (pos[:, :, 1].T - np.mean(pos[:, :, 1], axis=1)).T
            pos[:, :, 2] = (pos[:, :, 2].T - np.mean(pos[:, :, 2], axis=1)).T
            self._positions = pos
        else:
            self._positions = positions
        self._structure = structure
        self._cells = cells
        self._indices = indices

    def __getitem__(self, item):
        new_structure = self._structure.copy()
        if self._cells is not None:
            new_structure.cell = self._cells[item]
        if self._indices is not None:
            new_structure.indices = self._indices[item]
        new_structure.positions = self._positions[item]
        # This step is necessary for using ase.io.write for trajectories
        new_structure.arrays["positions"] = new_structure.positions
        # new_structure.arrays['cells'] = new_structure.cell
        return new_structure

    def __len__(self):
        return len(self._positions)


class GenericInput(GenericParameters):
    def __init__(self, input_file_name=None, table_name="generic"):
        super(GenericInput, self).__init__(
            input_file_name=input_file_name,
            table_name=table_name,
            comment_char="#",
            separator_char="=",
        )

    def load_default(self):
        """
        Loads the default file content
        """
        file_content = """\
calc_mode=static # static, minimize, md
structure=atoms # atoms, continue_final
"""
        self.load_string(file_content)


class GenericOutput(object):
    def __init__(self, job):
        self._job = job

    @property
    def cells(self):
        return self._job["output/generic/cells"]

    @property
    def energy_pot(self):
        return self._job["output/generic/energy_pot"]

    @property
    def energy_tot(self):
        return self._job["output/generic/energy_tot"]

    @property
    def forces(self):
        return self._job["output/generic/forces"]

    @property
    def force_max(self):
        """
            maximum force magnitude of each step which is used for
            convergence criterion of structure optimizations
        """
        return np.linalg.norm(self.forces, axis=-1).max(axis=-1)

    @property
    def positions(self):
        return self._job["output/generic/positions"]

    @property
    def pressures(self):
        return self._job["output/generic/pressures"]

    @property
    def steps(self):
        return self._job["output/generic/steps"]

    @property
    def temperature(self):
        return self._job["output/generic/temperature"]

    @property
    def computation_time(self):
        return self._job["output/generic/computation_time"]

    @property
    def unwrapped_positions(self):
        unwrapped_positions = self._job["output/generic/unwrapped_positions"]
        if unwrapped_positions is not None:
            return unwrapped_positions
        else:
            return self._job.structure.positions+self.total_displacements

    @property
    def volume(self):
        return self._job["output/generic/volume"]

    @property
    def indices(self):
        return self._job["output/generic/indices"]

    @property
    def displacements(self):
        """
        Output for 3-d displacements between successive snapshots, with minimum image convention.
        For the total displacements from the initial configuration, use total_displacements
        This algorithm collapses if:
        - the ID's are not consistent (i.e. you can also not change the number of atoms)
        - there are atoms which move by more than half a box length in any direction within two snapshots (due to
        periodic boundary conditions)
        """
        # Check if the volume changes in any snapshot
        vol = np.linalg.det(self.cells)
        varying_cell = np.sqrt(np.average((vol - vol[0])**2)) > 1e-5
        return self.get_displacements(self._job.structure, self.positions, self.cells, varying_cell=varying_cell)

    @staticmethod
    def get_displacements(structure, positions, cells, varying_cell=False):
        """
        Output for 3-d displacements between successive snapshots, with minimum image convention.
        For the total displacements from the initial configuration, use total_displacements
        This algorithm collapses if:
        - the ID's are not consistent (i.e. you can also not change the number of atoms)
        - there are atoms which move by more than half a box length in any direction within two snapshots (due to
        periodic boundary conditions)

        Args:
            structure (pyiron.atomistics.structure.atoms.Atoms): The initial structure
            positions (numpy.ndarray/list): List of positions in cartesian coordinates (N_steps x N_atoms x 3)
            cells (numpy.ndarray/list): List of cells (N_steps x 3 x 3)
            varying_cell (bool): True if the cell shape varies during the trajectory (raises a warning)

        Returns:
            numpy.ndarray: Displacements (N_steps x N_atoms x 3)

        """
        if not varying_cell:
            displacement = np.tensordot(positions, np.linalg.inv(cells[-1]), axes=([2, 0]))
            displacement -= np.append(structure.get_scaled_positions(),
                                      displacement).reshape(len(positions) + 1, len(structure), 3)[:-1]
            displacement -= np.rint(displacement)
            displacement = np.tensordot(displacement, cells[-1], axes=([2, 0]))
        else:
            warnings.warn("You are computing displacements in a simulation with periodic boundary conditions \n"
                          "and a varying cell shape.")
            displacement = np.array(
                [np.tensordot(pos, np.linalg.inv(cell), axes=([1, 1])) for pos, cell in zip(positions, cells)])
            displacement -= np.append(structure.get_scaled_positions(),
                                      displacement).reshape(len(positions) + 1, len(structure), 3)[:-1]
            displacement -= np.rint(displacement)
            displacement = np.einsum('nki,nji->nkj', displacement, cells)
        return displacement

    @property
    def total_displacements(self):
        """
        Output for 3-d total displacements from the initial configuration, with minimum image convention.
        For the diplacements for the successive snapshots, use displacements
        This algorithm collapses if:
        - the ID's are not consistent (i.e. you can also not change the number of atoms)
        - there are atoms which move by more than half a box length in any direction within two snapshots (due to periodic boundary conditions)
        """
        return np.cumsum(self.displacements, axis=0)

    def __dir__(self):
        hdf5_path = self._job["output/generic"]
        if hdf5_path is not None:
            return hdf5_path.list_nodes()
        else:
            return []