cohp.py

# coding: utf-8
# Copyright (c) Pymatgen Development Team.
# Distributed under the terms of the MIT License.

"""
This module defines classes to represent crystal orbital Hamilton
populations (COHP) and integrated COHP (ICOHP), but can also be used
for crystal orbital overlap populations (COOP).
"""

import re
import sys
import warnings

import numpy as np
from monty.json import MSONable
from scipy.interpolate import InterpolatedUnivariateSpline
from pymatgen.core.sites import PeriodicSite
from pymatgen.core.structure import Structure
from pymatgen.electronic_structure.core import Orbital, Spin
from pymatgen.io.lmto import LMTOCopl
from pymatgen.io.lobster import Cohpcar
from pymatgen.util.coord import get_linear_interpolated_value
from pymatgen.util.num import round_to_sigfigs

__author__ = "Marco Esters, Janine George"
__copyright__ = "Copyright 2017, The Materials Project"
__version__ = "0.2"
__maintainer__ = "Marco Esters, Janine George"
__email__ = "esters@uoregon.edu, janine.george@uclouvain.be"
__date__ = "Dec 13, 2017"


class Cohp(MSONable):
    """
    Basic COHP object.
    """

    def __init__(self, efermi, energies, cohp, are_coops=False, icohp=None):
        """
        Args:
            are_coops: Indicates whether this object describes COHPs or COOPs.
            efermi: Fermi energy.
            energies: A sequence of energies.
            cohp ({Spin: np.array}): representing the COHP for each spin.
            icohp ({Spin: np.array}): representing the ICOHP for each spin.
        """
        self.are_coops = are_coops
        self.efermi = efermi
        self.energies = np.array(energies)
        self.cohp = cohp
        self.icohp = icohp

    def __repr__(self):
        return self.__str__()

    def __str__(self):
        """
        Returns a string that can be easily plotted (e.g. using gnuplot).
        """
        cohpstring = "COOP" if self.are_coops else "COHP"
        header = ["Energy", cohpstring + "Up"]
        data = [self.energies, self.cohp[Spin.up]]
        if Spin.down in self.cohp:
            header.append(cohpstring + "Down")
            data.append(self.cohp[Spin.down])
        if self.icohp:
            header.append("I" + cohpstring + "Up")
            data.append(self.icohp[Spin.up])
            if Spin.down in self.cohp:
                header.append("I" + cohpstring + "Down")
                data.append(self.icohp[Spin.down])
        formatheader = "#" + " ".join(["{:15s}" for __ in header])
        formatdata = " ".join(["{:.5f}" for __ in header])
        stringarray = [formatheader.format(*header)]
        for i, __ in enumerate(self.energies):
            stringarray.append(formatdata.format(*[d[i] for d in data]))
        return "\n".join(stringarray)

    def as_dict(self):
        """
        Json-serializable dict representation of COHP.
        """
        d = {
            "@module": self.__class__.__module__,
            "@class": self.__class__.__name__,
            "are_coops": self.are_coops,
            "efermi": self.efermi,
            "energies": self.energies.tolist(),
            "COHP": {str(spin): pops.tolist() for spin, pops in self.cohp.items()},
        }
        if self.icohp:
            d["ICOHP"] = {str(spin): pops.tolist() for spin, pops in self.icohp.items()}
        return d

    def get_cohp(self, spin=None, integrated=False):
        """
        Returns the COHP or ICOHP for a particular spin.

        Args:
            spin: Spin. Can be parsed as spin object, integer (-1/1)
                or str ("up"/"down")
            integrated: Return COHP (False) or ICOHP (True)

        Returns:
            Returns the CHOP or ICOHP for the input spin. If Spin is
            None and both spins are present, both spins will be returned
            as a dictionary.
        """
        if not integrated:
            populations = self.cohp
        else:
            populations = self.icohp

        if populations is None:
            return None
        if spin is None:
            return populations
        if isinstance(spin, int):
            spin = Spin(spin)
        elif isinstance(spin, str):
            s = {"up": 1, "down": -1}[spin.lower()]
            spin = Spin(s)
        return {spin: populations[spin]}

    def get_icohp(self, spin=None):
        """
        Convenient alternative to get the ICOHP for a particular spin.
        """
        return self.get_cohp(spin=spin, integrated=True)

    def get_interpolated_value(self, energy, integrated=False):
        """
        Returns the COHP for a particular energy.

        Args:
            energy: Energy to return the COHP value for.
        """
        inter = {}
        for spin in self.cohp:
            if not integrated:
                inter[spin] = get_linear_interpolated_value(self.energies, self.cohp[spin], energy)
            elif self.icohp is not None:
                inter[spin] = get_linear_interpolated_value(self.energies, self.icohp[spin], energy)
            else:
                raise ValueError("ICOHP is empty.")
        return inter

    def has_antibnd_states_below_efermi(self, spin=None, limit=0.01):
        """
        Returns dict indicating if there are antibonding states below the Fermi level depending on the spin
            spin: Spin
            limit: -COHP smaller -limit will be considered.

        """
        warnings.warn("This method has not been tested on many examples. Check the parameter limit, pls!")

        populations = self.cohp
        number_energies_below_efermi = len([x for x in self.energies if x <= self.efermi])

        if populations is None:
            return None
        if spin is None:
            dict_to_return = {}
            for sp, cohpvalues in populations.items():
                if (max(cohpvalues[0:number_energies_below_efermi])) > limit:
                    dict_to_return[sp] = True
                else:
                    dict_to_return[sp] = False
        else:
            dict_to_return = {}
            if isinstance(spin, int):
                spin = Spin(spin)
            elif isinstance(spin, str):
                s = {"up": 1, "down": -1}[spin.lower()]
                spin = Spin(s)
            if (max(populations[spin][0:number_energies_below_efermi])) > limit:
                dict_to_return[spin] = True
            else:
                dict_to_return[spin] = False

        return dict_to_return

    @classmethod
    def from_dict(cls, d):
        """
        Returns a COHP object from a dict representation of the COHP.

        """
        if "ICOHP" in d:
            icohp = {Spin(int(key)): np.array(val) for key, val in d["ICOHP"].items()}
        else:
            icohp = None
        return Cohp(
            d["efermi"],
            d["energies"],
            {Spin(int(key)): np.array(val) for key, val in d["COHP"].items()},
            icohp=icohp,
            are_coops=d["are_coops"],
        )


class CompleteCohp(Cohp):
    """
    A wrapper class that defines an average COHP, and individual COHPs.

    .. attribute: are_coops

         Indicates whether the object is of COOPs or COHPs.

    .. attribute: efermi

         Fermi energy

    .. attribute: energies

         Sequence of energies

    .. attribute: structure

         Structure associated with the COHPs.

    .. attribute: cohp, icohp

         The average COHP/ICOHP.

    .. attribute: all_cohps

         A dict of COHPs for individual bonds of the form {label: COHP}

    .. attribute: orb_res_cohp

        Orbital-resolved COHPs.
    """

    def __init__(
        self,
        structure,
        avg_cohp,
        cohp_dict,
        bonds=None,
        are_coops=False,
        orb_res_cohp=None,
    ):
        """
        Args:
            structure: Structure assosciated with this COHP.
            avg_cohp: The average cohp as a COHP object.
            cohps: A dict of COHP objects for individual bonds of the form
                {label: COHP}
            bonds: A dict containing information on the bonds of the form
                {label: {key: val}}. The key-val pair can be any information
                the user wants to put in, but typically contains the sites,
                the bond length, and the number of bonds. If nothing is
                supplied, it will default to an empty dict.
            are_coops: indicates whether the Cohp objects are COHPs or COOPs.
                Defauls to False for COHPs.
            orb_res_cohp: Orbital-resolved COHPs.
        """
        super().__init__(
            avg_cohp.efermi,
            avg_cohp.energies,
            avg_cohp.cohp,
            are_coops=are_coops,
            icohp=avg_cohp.icohp,
        )
        self.structure = structure
        self.are_coops = are_coops
        self.all_cohps = cohp_dict
        self.orb_res_cohp = orb_res_cohp
        if bonds is None:
            self.bonds = {label: {} for label in self.all_cohps.keys()}
        else:
            self.bonds = bonds

    def __str__(self):
        if self.are_coops:
            return "Complete COOPs for " + str(self.structure)
        return "Complete COHPs for " + str(self.structure)

    def as_dict(self):
        """
        Json-serializable dict representation of CompleteCohp.
        """
        d = {
            "@module": self.__class__.__module__,
            "@class": self.__class__.__name__,
            "are_coops": self.are_coops,
            "efermi": self.efermi,
            "structure": self.structure.as_dict(),
            "energies": self.energies.tolist(),
            "COHP": {"average": {str(spin): pops.tolist() for spin, pops in self.cohp.items()}},
        }

        if self.icohp is not None:
            d["ICOHP"] = {"average": {str(spin): pops.tolist() for spin, pops in self.icohp.items()}}

        for label in self.all_cohps.keys():
            d["COHP"].update({label: {str(spin): pops.tolist() for spin, pops in self.all_cohps[label].cohp.items()}})
            if self.all_cohps[label].icohp is not None:
                if "ICOHP" not in d.keys():
                    d["ICOHP"] = {
                        label: {str(spin): pops.tolist() for spin, pops in self.all_cohps[label].icohp.items()}
                    }
                else:
                    d["ICOHP"].update(
                        {label: {str(spin): pops.tolist() for spin, pops in self.all_cohps[label].icohp.items()}}
                    )
        if False in [bond_dict == {} for bond_dict in self.bonds.values()]:
            d["bonds"] = {
                bond: {
                    "length": self.bonds[bond]["length"],
                    "sites": [site.as_dict() for site in self.bonds[bond]["sites"]],
                }
                for bond in self.bonds
            }
        if self.orb_res_cohp:
            orb_dict = {}
            for label in self.orb_res_cohp:
                orb_dict[label] = {}
                for orbs in self.orb_res_cohp[label]:
                    cohp = {str(spin): pops.tolist() for spin, pops in self.orb_res_cohp[label][orbs]["COHP"].items()}
                    orb_dict[label][orbs] = {"COHP": cohp}
                    icohp = {str(spin): pops.tolist() for spin, pops in self.orb_res_cohp[label][orbs]["ICOHP"].items()}
                    orb_dict[label][orbs]["ICOHP"] = icohp
                    orbitals = [[orb[0], orb[1].name] for orb in self.orb_res_cohp[label][orbs]["orbitals"]]
                    orb_dict[label][orbs]["orbitals"] = orbitals
            d["orb_res_cohp"] = orb_dict

        return d

    def get_cohp_by_label(self, label, summed_spin_channels=False):
        """
        Get specific COHP object.

        Args:
            label: string (for newer Lobster versions: a number)
            summed_spin_channels: bool, will sum the spin channels and return the sum in Spin.up if true

        Returns:
            Returns the COHP object to simplify plotting
        """
        # TODO: possibly add option to only get one spin channel as Spin.down in the future!
        if label.lower() == "average":
            divided_cohp = self.cohp
            divided_icohp = self.icohp

        else:
            divided_cohp = self.all_cohps[label].get_cohp(spin=None, integrated=False)
            divided_icohp = self.all_cohps[label].get_icohp(spin=None)

        if summed_spin_channels and Spin.down in self.cohp:
            final_cohp = {}
            final_icohp = {}
            final_cohp[Spin.up] = np.sum([divided_cohp[Spin.up], divided_cohp[Spin.down]], axis=0)
            final_icohp[Spin.up] = np.sum([divided_icohp[Spin.up], divided_icohp[Spin.down]], axis=0)
        else:
            final_cohp = divided_cohp
            final_icohp = divided_icohp

        return Cohp(
            efermi=self.efermi,
            energies=self.energies,
            cohp=final_cohp,
            are_coops=self.are_coops,
            icohp=final_icohp,
        )

    def get_summed_cohp_by_label_list(self, label_list, divisor=1, summed_spin_channels=False):
        """
        Returns a COHP object that includes a summed COHP divided by divisor

        Args:
            label_list: list of labels for the COHP that should be included in the summed cohp
            divisor: float/int, the summed cohp will be divided by this divisor
            summed_spin_channels: bool, will sum the spin channels and return the sum in Spin.up if true
        Returns:
            Returns a COHP object including a summed COHP
        """
        # check if cohps are spinpolarized or not
        first_cohpobject = self.get_cohp_by_label(label_list[0])
        summed_cohp = first_cohpobject.cohp.copy()
        summed_icohp = first_cohpobject.icohp.copy()
        for label in label_list[1:]:
            cohp_here = self.get_cohp_by_label(label)
            summed_cohp[Spin.up] = np.sum([summed_cohp[Spin.up], cohp_here.cohp[Spin.up]], axis=0)

            if Spin.down in summed_cohp:
                summed_cohp[Spin.down] = np.sum([summed_cohp[Spin.down], cohp_here.cohp[Spin.down]], axis=0)

            summed_icohp[Spin.up] = np.sum([summed_icohp[Spin.up], cohp_here.icohp[Spin.up]], axis=0)

            if Spin.down in summed_icohp:
                summed_icohp[Spin.down] = np.sum([summed_icohp[Spin.down], cohp_here.icohp[Spin.down]], axis=0)

        divided_cohp = {}
        divided_icohp = {}
        divided_cohp[Spin.up] = np.divide(summed_cohp[Spin.up], divisor)
        divided_icohp[Spin.up] = np.divide(summed_icohp[Spin.up], divisor)
        if Spin.down in summed_cohp:
            divided_cohp[Spin.down] = np.divide(summed_cohp[Spin.down], divisor)
            divided_icohp[Spin.down] = np.divide(summed_icohp[Spin.down], divisor)

        if summed_spin_channels and Spin.down in summed_cohp:
            final_cohp = {}
            final_icohp = {}
            final_cohp[Spin.up] = np.sum([divided_cohp[Spin.up], divided_cohp[Spin.down]], axis=0)
            final_icohp[Spin.up] = np.sum([divided_icohp[Spin.up], divided_icohp[Spin.down]], axis=0)
        else:
            final_cohp = divided_cohp
            final_icohp = divided_icohp

        return Cohp(
            efermi=first_cohpobject.efermi,
            energies=first_cohpobject.energies,
            cohp=final_cohp,
            are_coops=first_cohpobject.are_coops,
            icohp=final_icohp,
        )

    def get_summed_cohp_by_label_and_orbital_list(
        self, label_list, orbital_list, divisor=1, summed_spin_channels=False
    ):
        """
        Returns a COHP object that includes a summed COHP divided by divisor

        Args:
            label_list: list of labels for the COHP that should be included in the summed cohp
            orbital_list: list of orbitals for the COHPs that should be included in the summed cohp (same order as
                label_list)
            divisor: float/int, the summed cohp will be divided by this divisor
            summed_spin_channels: bool, will sum the spin channels and return the sum in Spin.up if true

        Returns:
            Returns a COHP object including a summed COHP
        """
        # check length of label_list and orbital_list:
        if not len(label_list) == len(orbital_list):
            raise ValueError("label_list and orbital_list don't have the same length!")
        # check if cohps are spinpolarized or not
        first_cohpobject = self.get_orbital_resolved_cohp(label_list[0], orbital_list[0])
        summed_cohp = first_cohpobject.cohp.copy()
        summed_icohp = first_cohpobject.icohp.copy()
        for ilabel, label in enumerate(label_list[1:], 1):
            cohp_here = self.get_orbital_resolved_cohp(label, orbital_list[ilabel])
            summed_cohp[Spin.up] = np.sum([summed_cohp[Spin.up], cohp_here.cohp.copy()[Spin.up]], axis=0)
            if Spin.down in summed_cohp:
                summed_cohp[Spin.down] = np.sum([summed_cohp[Spin.down], cohp_here.cohp.copy()[Spin.down]], axis=0)
            summed_icohp[Spin.up] = np.sum([summed_icohp[Spin.up], cohp_here.icohp.copy()[Spin.up]], axis=0)
            if Spin.down in summed_icohp:
                summed_icohp[Spin.down] = np.sum([summed_icohp[Spin.down], cohp_here.icohp.copy()[Spin.down]], axis=0)

        divided_cohp = {}
        divided_icohp = {}
        divided_cohp[Spin.up] = np.divide(summed_cohp[Spin.up], divisor)
        divided_icohp[Spin.up] = np.divide(summed_icohp[Spin.up], divisor)
        if Spin.down in summed_cohp:
            divided_cohp[Spin.down] = np.divide(summed_cohp[Spin.down], divisor)
            divided_icohp[Spin.down] = np.divide(summed_icohp[Spin.down], divisor)

        if summed_spin_channels and Spin.down in divided_cohp:
            final_cohp = {}
            final_icohp = {}

            final_cohp[Spin.up] = np.sum([divided_cohp[Spin.up], divided_cohp[Spin.down]], axis=0)
            final_icohp[Spin.up] = np.sum([divided_icohp[Spin.up], divided_icohp[Spin.down]], axis=0)
        else:
            final_cohp = divided_cohp
            final_icohp = divided_icohp

        return Cohp(
            efermi=first_cohpobject.efermi,
            energies=first_cohpobject.energies,
            cohp=final_cohp,
            are_coops=first_cohpobject.are_coops,
            icohp=final_icohp,
        )

    def get_orbital_resolved_cohp(self, label, orbitals, summed_spin_channels=False):
        """
        Get orbital-resolved COHP.

        Args:
            label: bond label (Lobster: labels as in ICOHPLIST/ICOOPLIST.lobster).

            orbitals: The orbitals as a label, or list or tuple of the form
                [(n1, orbital1), (n2, orbital2)]. Orbitals can either be str,
                int, or Orbital.

            summed_spin_channels: bool, will sum the spin channels and return the sum in Spin.up if true

        Returns:
            A Cohp object if CompleteCohp contains orbital-resolved cohp,
            or None if it doesn't.

        Note: It currently assumes that orbitals are str if they aren't the
            other valid types. This is not ideal, but the easiest way to
            avoid unicode issues between python 2 and python 3.
        """
        if self.orb_res_cohp is None:
            return None
        if isinstance(orbitals, (list, tuple)):
            cohp_orbs = [d["orbitals"] for d in self.orb_res_cohp[label].values()]
            orbs = []
            for orbital in orbitals:
                if isinstance(orbital[1], int):
                    orbs.append(tuple((orbital[0], Orbital(orbital[1]))))
                elif isinstance(orbital[1], Orbital):
                    orbs.append(tuple((orbital[0], orbital[1])))
                elif isinstance(orbital[1], str):
                    orbs.append(tuple((orbital[0], Orbital[orbital[1]])))
                else:
                    raise TypeError("Orbital must be str, int, or Orbital.")
            orb_index = cohp_orbs.index(orbs)
            orb_label = list(self.orb_res_cohp[label].keys())[orb_index]
        elif isinstance(orbitals, str):
            orb_label = orbitals
        else:
            raise TypeError("Orbitals must be str, list, or tuple.")
        try:
            icohp = self.orb_res_cohp[label][orb_label]["ICOHP"]
        except KeyError:
            icohp = None

        start_cohp = self.orb_res_cohp[label][orb_label]["COHP"]
        start_icohp = icohp

        if summed_spin_channels and Spin.down in start_cohp:
            final_cohp = {}
            final_icohp = {}
            final_cohp[Spin.up] = np.sum([start_cohp[Spin.up], start_cohp[Spin.down]], axis=0)
            if start_icohp is not None:
                final_icohp[Spin.up] = np.sum([start_icohp[Spin.up], start_icohp[Spin.down]], axis=0)
        else:
            final_cohp = start_cohp
            final_icohp = start_icohp

        return Cohp(
            self.efermi,
            self.energies,
            final_cohp,
            icohp=final_icohp,
            are_coops=self.are_coops,
        )

    @classmethod
    def from_dict(cls, d):
        """
        Returns CompleteCohp object from dict representation.
        """
        cohp_dict = {}
        efermi = d["efermi"]
        energies = d["energies"]
        structure = Structure.from_dict(d["structure"])
        if "bonds" in d.keys():
            bonds = {
                bond: {
                    "length": d["bonds"][bond]["length"],
                    "sites": tuple(PeriodicSite.from_dict(site) for site in d["bonds"][bond]["sites"]),
                }
                for bond in d["bonds"]
            }
        else:
            bonds = None
        for label in d["COHP"]:
            cohp = {Spin(int(spin)): np.array(d["COHP"][label][spin]) for spin in d["COHP"][label]}
            try:
                icohp = {Spin(int(spin)): np.array(d["ICOHP"][label][spin]) for spin in d["ICOHP"][label]}
            except KeyError:
                icohp = None
            if label == "average":
                avg_cohp = Cohp(efermi, energies, cohp, icohp=icohp)
            else:
                cohp_dict[label] = Cohp(efermi, energies, cohp, icohp=icohp)

        if "orb_res_cohp" in d.keys():
            orb_cohp = {}
            for label in d["orb_res_cohp"]:
                orb_cohp[label] = {}
                for orb in d["orb_res_cohp"][label]:
                    cohp = {
                        Spin(int(s)): np.array(d["orb_res_cohp"][label][orb]["COHP"][s], dtype=float)
                        for s in d["orb_res_cohp"][label][orb]["COHP"]
                    }
                    try:
                        icohp = {
                            Spin(int(s)): np.array(d["orb_res_cohp"][label][orb]["ICOHP"][s], dtype=float)
                            for s in d["orb_res_cohp"][label][orb]["ICOHP"]
                        }
                    except KeyError:
                        icohp = None
                    orbitals = [tuple((int(o[0]), Orbital[o[1]])) for o in d["orb_res_cohp"][label][orb]["orbitals"]]
                    orb_cohp[label][orb] = {
                        "COHP": cohp,
                        "ICOHP": icohp,
                        "orbitals": orbitals,
                    }
                # If no total COHPs are present, calculate the total
                # COHPs from the single-orbital populations. Total COHPs
                # may not be present when the cohpgenerator keyword is used
                # in LOBSTER versions 2.2.0 and earlier.
                if label not in d["COHP"] or d["COHP"][label] is None:
                    cohp = {
                        Spin.up: np.sum(
                            np.array([orb_cohp[label][orb]["COHP"][Spin.up] for orb in orb_cohp[label]]),
                            axis=0,
                        )
                    }
                    try:
                        cohp[Spin.down] = np.sum(
                            np.array([orb_cohp[label][orb]["COHP"][Spin.down] for orb in orb_cohp[label]]),
                            axis=0,
                        )
                    except KeyError:
                        pass

                orb_res_icohp = None in [orb_cohp[label][orb]["ICOHP"] for orb in orb_cohp[label]]
                if (label not in d["ICOHP"] or d["ICOHP"][label] is None) and orb_res_icohp:
                    icohp = {
                        Spin.up: np.sum(
                            np.array([orb_cohp[label][orb]["ICOHP"][Spin.up] for orb in orb_cohp[label]]),
                            axis=0,
                        )
                    }
                    try:
                        icohp[Spin.down] = np.sum(
                            np.array([orb_cohp[label][orb]["ICOHP"][Spin.down] for orb in orb_cohp[label]]),
                            axis=0,
                        )
                    except KeyError:
                        pass
        else:
            orb_cohp = None

        if "average" not in d["COHP"].keys():
            # calculate average
            cohp = np.array([np.array(c) for c in d["COHP"].values()]).mean(axis=0)
            try:
                icohp = np.array([np.array(c) for c in d["ICOHP"].values()]).mean(axis=0)
            except KeyError:
                icohp = None
            avg_cohp = Cohp(efermi, energies, cohp, icohp=icohp)

        return CompleteCohp(
            structure,
            avg_cohp,
            cohp_dict,
            bonds=bonds,
            are_coops=d["are_coops"],
            orb_res_cohp=orb_cohp,
        )

    @classmethod
    def from_file(cls, fmt, filename=None, structure_file=None, are_coops=False):
        """
        Creates a CompleteCohp object from an output file of a COHP
        calculation. Valid formats are either LMTO (for the Stuttgart
        LMTO-ASA code) or LOBSTER (for the LOBSTER code).

        Args:
            cohp_file: Name of the COHP output file. Defaults to COPL
                for LMTO and COHPCAR.lobster/COOPCAR.lobster for LOBSTER.

            are_coops: Indicates whether the populations are COOPs or
                COHPs. Defaults to False for COHPs.

            fmt: A string for the code that was used to calculate
                the COHPs so that the output file can be handled
                correctly. Can take the values "LMTO" or "LOBSTER".

            structure_file: Name of the file containing the structure.
                If no file name is given, use CTRL for LMTO and POSCAR
                for LOBSTER.

        Returns:
            A CompleteCohp object.
        """
        fmt = fmt.upper()
        if fmt == "LMTO":
            # LMTO COOPs and orbital-resolved COHP cannot be handled yet.
            are_coops = False
            orb_res_cohp = None
            if structure_file is None:
                structure_file = "CTRL"
            if filename is None:
                filename = "COPL"
            cohp_file = LMTOCopl(filename=filename, to_eV=True)
        elif fmt == "LOBSTER":
            if structure_file is None:
                structure_file = "POSCAR"
            if filename is None:
                filename = "COOPCAR.lobster" if are_coops else "COHPCAR.lobster"
            warnings.warn(
                "The bond labels are currently consistent with ICOHPLIST.lobster/ICOOPLIST.lobster, not with "
                "COHPCAR.lobster/COOPCAR.lobster. Please be aware!"
            )
            cohp_file = Cohpcar(filename=filename, are_coops=are_coops)
            orb_res_cohp = cohp_file.orb_res_cohp
        else:
            raise ValueError("Unknown format %s. Valid formats are LMTO " "and LOBSTER." % fmt)

        structure = Structure.from_file(structure_file)
        efermi = cohp_file.efermi
        cohp_data = cohp_file.cohp_data
        energies = cohp_file.energies

        # Lobster shifts the energies so that the Fermi energy is at zero.
        # Shifting should be done by the plotter object though.

        spins = [Spin.up, Spin.down] if cohp_file.is_spin_polarized else [Spin.up]
        if fmt == "LOBSTER":
            energies += efermi

        if orb_res_cohp is not None:
            # If no total COHPs are present, calculate the total
            # COHPs from the single-orbital populations. Total COHPs
            # may not be present when the cohpgenerator keyword is used
            # in LOBSTER versions 2.2.0 and earlier.
            # TODO: Test this more extensively
            # pylint: disable=E1133,E1136
            for label in orb_res_cohp:
                if cohp_file.cohp_data[label]["COHP"] is None:
                    # print(label)
                    cohp_data[label]["COHP"] = {
                        sp: np.sum(
                            [orb_res_cohp[label][orbs]["COHP"][sp] for orbs in orb_res_cohp[label]],
                            axis=0,
                        )
                        for sp in spins
                    }
                if cohp_file.cohp_data[label]["ICOHP"] is None:
                    cohp_data[label]["ICOHP"] = {
                        sp: np.sum(
                            [orb_res_cohp[label][orbs]["ICOHP"][sp] for orbs in orb_res_cohp[label]],
                            axis=0,
                        )
                        for sp in spins
                    }

        if fmt == "LMTO":
            # Calculate the average COHP for the LMTO file to be
            # consistent with LOBSTER output.
            avg_data = {"COHP": {}, "ICOHP": {}}
            for i in avg_data:
                for spin in spins:
                    rows = np.array([cohp_data[label][i][spin] for label in cohp_data])
                    avg = np.average(rows, axis=0)
                    # LMTO COHPs have 5 significant figures
                    avg_data[i].update({spin: np.array([round_to_sigfigs(a, 5) for a in avg], dtype=float)})
            avg_cohp = Cohp(efermi, energies, avg_data["COHP"], icohp=avg_data["ICOHP"])
        else:
            avg_cohp = Cohp(
                efermi,
                energies,
                cohp_data["average"]["COHP"],
                icohp=cohp_data["average"]["COHP"],
                are_coops=are_coops,
            )
            del cohp_data["average"]

        cohp_dict = {
            label: Cohp(
                efermi,
                energies,
                cohp_data[label]["COHP"],
                icohp=cohp_data[label]["ICOHP"],
                are_coops=are_coops,
            )
            for label in cohp_data
        }

        bond_dict = {
            label: {
                "length": cohp_data[label]["length"],
                "sites": [structure.sites[site] for site in cohp_data[label]["sites"]],
            }
            for label in cohp_data
        }

        return CompleteCohp(
            structure,
            avg_cohp,
            cohp_dict,
            bonds=bond_dict,
            are_coops=are_coops,
            orb_res_cohp=orb_res_cohp,
        )


class IcohpValue(MSONable):
    """
    Class to store information on an ICOHP or ICOOP value

    .. attribute:: num_bonds
            number of bonds used for the average cohp (relevant for Lobster versions <3.0) (int)

    .. attribute:: are_coops
            Boolean to indicate whether ICOOP or not

    .. attribute:: icohp
            dict={Spin.up: icohpvalue for spin.up, Spin.down: icohpvalue for spin.down}

    .. attribute:: summed_icohp:
            sum of icohp/icoop of both spin channels

    """

    def __init__(self, label, atom1, atom2, length, translation, num, icohp, are_coops=False):
        """
        Args:
            label: label for the icohp
            atom1: str of atom that is contributing to the bond
            atom2: str of second atom that is contributing to the bond
            length: float of bond lengths
            translation: translation list, e.g. [0,0,0]
            num: integer describing how often the bond exists
            icohp: dict={Spin.up: icohpvalue for spin.up, Spin.down: icohpvalue for spin.down}
        """
        self._are_coops = are_coops
        self._label = label
        self._atom1 = atom1
        self._atom2 = atom2
        self._length = length
        self._translation = translation
        self._num = num
        self._icohp = icohp
        if Spin.down in self._icohp:
            self._is_spin_polarized = True
        else:
            self._is_spin_polarized = False

    def __str__(self):

        if not self._are_coops:
            if self._is_spin_polarized:
                return (
                    "ICOHP "
                    + str(self._label)
                    + " between "
                    + str(self._atom1)
                    + " and "
                    + str(self._atom2)
                    + " ("
                    + str(self._translation)
                    + "): "
                    + str(self._icohp[Spin.up])
                    + " eV (Spin up) and "
                    + str(self._icohp[Spin.down])
                    + " eV (Spin down)"
                )
            return (
                "ICOHP "
                + str(self._label)
                + " between "
                + str(self._atom1)
                + " and "
                + str(self._atom2)
                + " ("
                + str(self._translation)
                + "): "
                + str(self._icohp[Spin.up])
                + " eV (Spin up)"
            )
        if self._is_spin_polarized:
            return (
                "ICOOP "
                + str(self._label)
                + " between "
                + str(self._atom1)
                + " and "
                + str(self._atom2)
                + " ("
                + str(self._translation)
                + "): "
                + str(self._icohp[Spin.up])
                + " (Spin up) and "
                + str(self._icohp[Spin.down])
                + " (Spin down)"
            )
        return (
            "ICOOP "
            + str(self._label)
            + " between "
            + str(self._atom1)
            + " and "
            + str(self._atom2)
            + " ("
            + str(self._translation)
            + "): "
            + str(self._icohp[Spin.up])
            + " (Spin up)"
        )

    @property
    def num_bonds(self):
        """
        tells the number of bonds for which the ICOHP value is an average
        Returns:
            Int
        """
        return self._num

    @property
    def are_coops(self):
        """
        tells if ICOOPs or not
        Returns:
            Boolean
        """
        return self._are_coops

    @property
    def is_spin_polarized(self):
        """
        tells if spin polarized calculation or not
        Returns:
            Boolean

        """
        return self._is_spin_polarized

    def icohpvalue(self, spin=Spin.up):
        """
        Args:
            spin: Spin.up or Spin.down
        Returns:
            icohpvalue (float) corresponding to chosen spin
        """
        if not self.is_spin_polarized and spin == Spin.down:
            raise ValueError("The calculation was not performed with spin polarization")

        return self._icohp[spin]

    @property
    def icohp(self):
        """
        dict with icohps for spinup and spindown
        Return:
            dict={Spin.up: icohpvalue for spin.up, Spin.down: icohpvalue for spin.down}
        """
        return self._icohp

    @property
    def summed_icohp(self):
        """
        Adds ICOHPs of both spin channels for spin polarized compounds
        Returns:
             icohp value in eV
        """
        if self._is_spin_polarized:
            sum_icohp = self._icohp[Spin.down] + self._icohp[Spin.up]
        else:
            sum_icohp = self._icohp[Spin.up]
        return sum_icohp


class IcohpCollection(MSONable):
    """
    Class to store IcohpValues

    .. attribute:: are_coops
        Boolean to indicate whether ICOHPs or ICOOPs are stored

    .. attribute:: is_spin_polarized
        Boolean to indicate if the Lobster calculation was done spin polarized or not

    """

    def __init__(
        self,
        list_labels,
        list_atom1,
        list_atom2,
        list_length,
        list_translation,
        list_num,
        list_icohp,
        is_spin_polarized,
        are_coops=False,
    ):
        """
        Args:
            is_spin_polarized: Boolean to indicate if the Lobster calculation was done spin polarized or not Boolean to
                indicate if the Lobster calculation was done spin polarized or not
            are_coops: Boolean to indicate whether ICOHPs or ICOOPs are stored
            list_labels: list of labels for ICOHP/ICOOP values
            list_atom1: list of str of atomnames e.g. "O1"
            list_atom2: list of str of atomnames e.g. "O1"
            list_length: list of lengths of corresponding bonds in Angstrom
            list_translation: list of translation list, e.g. [0,0,0]
            list_num: list of equivalent bonds, usually 1 starting from Lobster 3.0.0
            list_icohp: list of dict={Spin.up: icohpvalue for spin.up, Spin.down: icohpvalue for spin.down}
        """
        self._are_coops = are_coops
        self._icohplist = {}
        self._is_spin_polarized = is_spin_polarized
        self._list_labels = list_labels
        self._list_atom1 = list_atom1
        self._list_atom2 = list_atom2
        self._list_length = list_length
        self._list_translation = list_translation
        self._list_num = list_num
        self._list_icohp = list_icohp

        for ilist, listel in enumerate(list_labels):
            self._icohplist[listel] = IcohpValue(
                listel,
                list_atom1[ilist],
                list_atom2[ilist],
                list_length[ilist],
                list_translation[ilist],
                list_num[ilist],
                list_icohp[ilist],
            )

    def __str__(self):
        joinstr = []
        for value in self._icohplist.values():
            joinstr.append(str(value))
        return "\n".join(joinstr)

    def get_icohp_by_label(self, label, summed_spin_channels=True, spin=Spin.up):
        """
        get an icohp value for a certain bond as indicated by the label (bond labels starting by "1" as in
        ICOHPLIST/ICOOPLIST)

        Args:
            label: label in str format (usually the bond number in Icohplist.lobster/Icooplist.lobster
            summed_spin_channels: Boolean to indicate whether the ICOHPs/ICOOPs of both spin channels should be summed
            spin: if summed_spin_channels is equal to False, this spin indicates which spin channel should be returned

        Returns:
            float describing ICOHP/ICOOP value
        """

        icohp_here = self._icohplist[label]
        if icohp_here._is_spin_polarized:
            if summed_spin_channels:
                return icohp_here.summed_icohp
            return icohp_here.icohpvalue(spin)
        return icohp_here.icohpvalue(spin)

    def get_summed_icohp_by_label_list(self, label_list, divisor=1.0, summed_spin_channels=True, spin=Spin.up):
        """
        get the sum of several ICOHP values that are indicated by a list of labels (labels of the bonds are the same as
        in ICOHPLIST/ICOOPLIST)

        Args:
            label_list: list of labels of the ICOHPs/ICOOPs that should be summed
            divisor: is used to divide the sum
            summed_spin_channels: Boolean to indicate whether the ICOHPs/ICOOPs of both spin channels should be summed
            spin: if summed_spin_channels is equal to False, this spin indicates which spin channel should be returned

        Returns:
             float that is a sum of all ICOHPs/ICOOPs as indicated with label_list
        """
        sum_icohp = 0
        for label in label_list:
            icohp_here = self._icohplist[label]
            if icohp_here.num_bonds != 1:
                warnings.warn("One of the ICOHP values is an average over bonds. This is currently not considered.")
            # prints warning if num_bonds is not equal to 1
            if icohp_here._is_spin_polarized:
                if summed_spin_channels:
                    sum_icohp = sum_icohp + icohp_here.summed_icohp
                else:
                    sum_icohp = sum_icohp + icohp_here.icohpvalue(spin)
            else:
                sum_icohp = sum_icohp + icohp_here.icohpvalue(spin)
        return sum_icohp / divisor

    def get_icohp_dict_by_bondlengths(self, minbondlength=0.0, maxbondlength=8.0):
        """
        get a dict of IcohpValues corresponding to certaind bond lengths
        Args:
            minbondlength: defines the minimum of the bond lengths of the bonds
            maxbondlength: defines the maximum of the bond lengths of the bonds
        Returns:
             dict of IcohpValues, the keys correspond to the values from the initial list_labels
        """
        newicohp_dict = {}
        for value in self._icohplist.values():
            if value._length >= minbondlength and value._length <= maxbondlength:
                newicohp_dict[value._label] = value
        return newicohp_dict

    def get_icohp_dict_of_site(
        self,
        site,
        minsummedicohp=None,
        maxsummedicohp=None,
        minbondlength=0.0,
        maxbondlength=8.0,
        only_bonds_to=None,
    ):
        """
        get a dict of IcohpValue for a certain site (indicated by integer)
        Args:
            site: integer describing the site of interest, order as in Icohplist.lobster/Icooplist.lobster, starts at 0
            minsummedicohp: float, minimal icohp/icoop of the bonds that are considered. It is the summed ICOHP value
                from both spin channels for spin polarized cases
            maxsummedicohp: float, maximal icohp/icoop of the bonds that are considered. It is the summed ICOHP value
                from both spin channels for spin polarized cases
            minbondlength: float, defines the minimum of the bond lengths of the bonds
            maxbondlength: float, defines the maximum of the bond lengths of the bonds
            only_bonds_to: list of strings describing the bonding partners that are allowed, e.g. ['O']
        Returns:
             dict of IcohpValues, the keys correspond to the values from the initial list_labels
        """

        newicohp_dict = {}
        for key, value in self._icohplist.items():
            atomnumber1 = int(re.split(r"(\d+)", value._atom1)[1]) - 1
            atomnumber2 = int(re.split(r"(\d+)", value._atom2)[1]) - 1
            if site in (atomnumber1, atomnumber2):
                # manipulate order of atoms so that searched one is always atom1
                if site == atomnumber2:
                    save = value._atom1
                    value._atom1 = value._atom2
                    value._atom2 = save

                if only_bonds_to is None:
                    second_test = True
                else:
                    second_test = re.split(r"(\d+)", value._atom2)[0] in only_bonds_to
                if value._length >= minbondlength and value._length <= maxbondlength and second_test:
                    if minsummedicohp is not None:
                        if value.summed_icohp >= minsummedicohp:
                            if maxsummedicohp is not None:
                                if value.summed_icohp <= maxsummedicohp:
                                    newicohp_dict[key] = value
                            else:
                                newicohp_dict[key] = value
                    else:
                        if maxsummedicohp is not None:
                            if value.summed_icohp <= maxsummedicohp:
                                newicohp_dict[key] = value
                        else:
                            newicohp_dict[key] = value

        return newicohp_dict

    def extremum_icohpvalue(self, summed_spin_channels=True, spin=Spin.up):
        """
        get ICOHP/ICOOP of strongest bond
        Args:
            summed_spin_channels: Boolean to indicate whether the ICOHPs/ICOOPs of both spin channels should be summed

            spin: if summed_spin_channels is equal to False, this spin indicates which spin channel should be returned
        Returns:
            lowest ICOHP/largest ICOOP value (i.e. ICOHP/ICOOP value of strongest bond)
        """
        if not self._are_coops:
            extremum = sys.float_info.max
        else:
            extremum = -sys.float_info.max

        if not self._is_spin_polarized:
            if spin == Spin.down:
                warnings.warn("This spin channel does not exist. I am switching to Spin.up")
            spin = Spin.up

        for value in self._icohplist.values():
            if not value.is_spin_polarized or not summed_spin_channels:
                if not self._are_coops:
                    if value.icohpvalue(spin) < extremum:
                        extremum = value.icohpvalue(spin)
                        # print(extremum)
                else:
                    if value.icohpvalue(spin) > extremum:
                        extremum = value.icohpvalue(spin)
                        # print(extremum)
            else:
                if not self._are_coops:
                    if value.summed_icohp < extremum:
                        extremum = value.summed_icohp
                        # print(extremum)
                else:
                    if value.summed_icohp > extremum:
                        extremum = value.summed_icohp
                        # print(extremum)
        return extremum

    @property
    def is_spin_polarized(self):
        """
        :return: Whether it is spin polarized.
        """
        return self._is_spin_polarized

    @property
    def are_coops(self):
        """
        :return: Whether this is coops.
        """
        return self._are_coops


def get_integrated_cohp_in_energy_range(
    cohp, label, orbital=None, energy_range=None, relative_E_Fermi=True, summed_spin_channels=True
):
    """
    Method that can integrate completecohp objects which include data on integrated COHPs
    Args:
        cohp: CompleteCOHP object
        label: label of the COHP data
        orbital: If not None, a orbital resolved integrated COHP will be returned
        energy_range:   if None, returns icohp value at Fermi level;
                        if float, integrates from this float up to the Fermi level;
                        if [float,float], will integrate in between
        relative_E_Fermi: if True, energy scale with E_Fermi at 0 eV is chosen
        summed_spin_channels: if True, Spin channels will be summed

    Returns:
        float indicating the integrated COHP if summed_spin_channels==True, otherwise dict of the following form {
        Spin.up:float, Spin.down:float}
    """
    summedicohp = {}
    if orbital is None:
        icohps = cohp.all_cohps[label].get_icohp(spin=None)
        if summed_spin_channels and Spin.down in icohps:
            summedicohp[Spin.up] = icohps[Spin.up] + icohps[Spin.down]
        else:
            summedicohp = icohps
    else:
        icohps = cohp.get_orbital_resolved_cohp(label=label, orbitals=orbital).icohp
        if summed_spin_channels and Spin.down in icohps:
            summedicohp[Spin.up] = icohps[Spin.up] + icohps[Spin.down]
        else:
            summedicohp = icohps

    if energy_range is None:

        energies_corrected = cohp.energies - cohp.efermi
        spl_spinup = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.up], ext=0)

        if not summed_spin_channels and Spin.down in icohps:
            spl_spindown = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.down], ext=0)
            return {Spin.up: spl_spinup(0.0), Spin.down: spl_spindown(0.0)}
        if summed_spin_channels:
            return spl_spinup(0.0)

        return {Spin.up: spl_spinup(0.0)}

    # returns icohp value at the Fermi level!
    if isinstance(energy_range, float):
        if relative_E_Fermi:
            energies_corrected = cohp.energies - cohp.efermi
            spl_spinup = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.up], ext=0)

            if not summed_spin_channels and Spin.down in icohps:
                spl_spindown = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.down], ext=0)
                return {
                    Spin.up: spl_spinup(0) - spl_spinup(energy_range),
                    Spin.down: spl_spindown(0) - spl_spindown(energy_range),
                }
            if summed_spin_channels:
                return spl_spinup(0) - spl_spinup(energy_range)
            return {Spin.up: spl_spinup(0) - spl_spinup(energy_range)}

        energies_corrected = cohp.energies
        spl_spinup = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.up], ext=0)

        if not summed_spin_channels and Spin.down in icohps:
            spl_spindown = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.down], ext=0)
            return {
                Spin.up: spl_spinup(cohp.efermi) - spl_spinup(energy_range),
                Spin.down: spl_spindown(cohp.efermi) - spl_spindown(energy_range),
            }
        if summed_spin_channels:
            return spl_spinup(cohp.efermi) - spl_spinup(energy_range)
        return {Spin.up: spl_spinup(cohp.efermi) - spl_spinup(energy_range)}

    if relative_E_Fermi:
        energies_corrected = cohp.energies - cohp.efermi
    else:
        energies_corrected = cohp.energies

    spl_spinup = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.up], ext=0)

    if not summed_spin_channels and Spin.down in icohps:
        spl_spindown = InterpolatedUnivariateSpline(energies_corrected, summedicohp[Spin.down], ext=0)
        return {
            Spin.up: spl_spinup(energy_range[1]) - spl_spinup(energy_range[0]),
            Spin.down: spl_spindown(energy_range[1]) - spl_spindown(energy_range[0]),
        }
    if summed_spin_channels:
        return spl_spinup(energy_range[1]) - spl_spinup(energy_range[0])

    return {Spin.up: spl_spinup(energy_range[1]) - spl_spinup(energy_range[0])}