Source code for hikari.dataframes.hkl

import copy
import random
from typing import Union, Iterable

import numpy as np
import numpy.linalg as lin
import pandas as pd

import hikari
from hikari.dataframes import BaseFrame
from hikari.symmetry import PG, SG, Group
from hikari.resources import hkl_formats, hkl_aliases, hkl_mercury_style, \
    characteristic_radiation
from hikari.utility import cubespace, chemical_elements, make_abspath, \
    rescale_list_to_range, rescale_list_to_other

pd.options.mode.chained_assignment = 'raise'


[docs] class HklKeyRegistrar(type): """Metaclass for `HklKey`s which registers them if they define `name`.""" REGISTRY = {} default: None dtype: np.dtype def __new__(mcs, name, bases, attrs): new_cls = type.__new__(mcs, name, bases, attrs) if hasattr(new_cls, 'name') and new_cls.name: mcs.REGISTRY[new_cls.name] = new_cls return new_cls @property def IMPERATIVES(self): # noqa - capital letters to avoid attribute clash return [k for k, v in self.REGISTRY.items() if v.imperative]
[docs] class HklKey(metaclass=HklKeyRegistrar): """Base Class for every subsequent HklKey.""" name: str = '' # if not empty, how key will be registered default = None # default to set if value was not provided dtype: str = '' # dtype to use when defining numpy array imperative: bool = False # is key imperative for every table? reduce_behaviour: str = 'keep' # 'add', 'average', 'discard' or 'keep' 1st type: type = type(None) # python type used when defining instances
[docs] class HklKeyImperativeInt8(HklKey): default = 0 dtype = np.int8 imperative = True
[docs] class HklKeyAveragedFloat64(HklKey): default = 0.0 dtype = np.float64 reduce_behaviour = 'average'
[docs] class HklKeyKeptFloat64(HklKey): default = 0.0 dtype = np.float64 reduce_behaviour = 'keep'
[docs] class HklKeyIndexH(HklKeyImperativeInt8): """Reciprocal lattice Miller index h""" name = 'h'
[docs] class HklKeyIndexK(HklKeyImperativeInt8): """Reciprocal lattice Miller index k""" name = 'k'
[docs] class HklKeyIndexL(HklKeyImperativeInt8): """Reciprocal lattice Miller index l""" name = 'l'
[docs] class HklKeyStructureFactor(HklKeyAveragedFloat64): """Crystallographic structure factor F""" name = 'F' default = 1.0
[docs] class HklKeyIntensity(HklKeyAveragedFloat64): """Crystallographic intensity I_obs""" name = 'I' default = 1.0
[docs] class HklKeyIntensityCalculated(HklKeyAveragedFloat64): """Crystallographic calculated intensity I_calc""" name = 'Ic' default = 1.0
[docs] class HklKeyIntensityUncertainty(HklKeyAveragedFloat64): """Uncertainty of intensity I determination""" name = 'si'
[docs] class HklKeyStructureFactorUncertainty(HklKeyAveragedFloat64): """Uncertainty of structure factor F determination""" name = 'sf'
[docs] class HklKeyStructureFactorToUncertaintyRatio(HklKeyAveragedFloat64): """Structure factor to its uncertainty ratio""" name = 'u'
[docs] class HklKeyBatchNumber(HklKey): """Batch or run number""" name = 'b' default = 0 dtype = np.int16 reduce_behaviour = 'discard'
[docs] class HklKeyCrystalNumber(HklKey): """Crystal domain or twin number""" name = 'c' default = 0 dtype = np.int16 reduce_behaviour = 'discard'
[docs] class HklKeyMultiplicity(HklKey): """Multiplicity i.e. how many observation contributed to this reflection""" name = 'm' default = 1 dtype = np.int16 imperative = True reduce_behaviour = 'add'
[docs] class HklKeyWavelength(HklKey): """Wavelength expressed in Angstrom""" name = 'la' default = 0.0 dtype = np.float64 reduce_behaviour = 'discard'
[docs] class HklKeyPhase(HklKeyAveragedFloat64): """Reflection phase expressed in radians""" name = 'ph' default = 0.0
[docs] class HklKeyRadius(HklKey): """Distance from 000 node, i.e. double the sin(th)/la resolution in A^-1""" name = 'r' dtype = np.float64
[docs] class HklKeyTransmissionPathLength(HklKeyKeptFloat64): """Absorption-weighted transmission path length in centimeters""" name = 't'
[docs] class HklKeyCosineU1(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'u1'
[docs] class HklKeyCosineU2(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'u2'
[docs] class HklKeyCosineU3(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'u3'
[docs] class HklKeyCosineV1(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'v1'
[docs] class HklKeyCosineV2(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'v2'
[docs] class HklKeyCosineV3(HklKeyKeptFloat64): """Direction cosines of a vector (unused, see XD manual for reference)""" name = 'v3'
[docs] class HklKeyCoordinateX(HklKeyKeptFloat64): """Reciprocal space coordinate x*""" name = 'x' dtype = np.float32
[docs] class HklKeyCoordinateY(HklKeyKeptFloat64): """Reciprocal space coordinate x*""" name = 'y' dtype = np.float32
[docs] class HklKeyCoordinateZ(HklKeyKeptFloat64): """Reciprocal space coordinate x*""" name = 'z' dtype = np.float32
[docs] class HklKeyZeta(HklKeyAveragedFloat64): """weighted diff. in observed I - calculated Ic intensity""" name = 'ze'
[docs] class HklKeyZetaSquared(HklKeyAveragedFloat64): """weighted diff. in observed I - calculated Ic intensity, squared""" name = 'ze2'
[docs] class HklKeyCalculatedIntensityToSigma(HklKeyAveragedFloat64): """Calculated intensity of reflection divided by experimental sigma""" name = 'Icsi' default = 1.0
[docs] class HklKeyObservedIntensityToSigma(HklKeyAveragedFloat64): """Observed intensity of reflection divided by experimental sigma""" name = 'Iosi' default = 1.0
[docs] class HklKeyEquivalenceCode(HklKey): """Integer unique for each set of symmetrically equivalent reflections""" name = 'equiv' default = 0 dtype = np.int64
[docs] class HklKeyDummy(HklKey): """Dummy column for loading or holding irrelevant string data""" name = 'None' default = '' dtype = np.str_
[docs] class HklFrame(BaseFrame): """ A master object which manages single-crystal diffraction files. It utilises other `Hkl*` classes to import, store, manipulate and output information about single-crystal diffraction patterns. HklFrame acts as a container which stores the diffraction data (Pandas dataframe, :attr:`table`) and elementary crystal cell data (:class:`hikari.dataframes.Base`). Demanding methods belonging to this class are vectorized, providing relatively satisfactory performance and high memory capacity. HklFrame methods are designed to work in-place, so the work strategy is to create a new instance of HklFrame for each reflection dataset, manipulate it using methods, eg. :func:`merge` or :func:`trim`, and :func:`copy` to other object or output using :func:`write` if needed. The HklFrame always initiates empty and does not accept any arguments. Some magic methods, such as :func:`__len__` and :func:`__add__` are defined and describe/operate on the :attr:`frame`. """ HKL_LIMIT = 127 """Highest absolute value of h, k or l index, which can be interpreted correctly by current version of the software.""" def __init__(self): """HklFrame constructor""" super().__init__() self.__la = 0.71069 """Wavelength of radiation used in experiment.""" self.table = pd.DataFrame() """ Pandas dataframe containing diffraction data information. Each row represents one reflection observation, while each column has one piece of information about the reflections. For a list of available keys, see :class:`HklKeys`, whose instance is used to menage the keys of this table. """
[docs] def __add__(self, other): """ :param other: HklFrame to be added to data :type other: HklFrame :return: concatenated :attr:`table` dataframes with metadata from first :rtype: HklFrame """ _copied = self.copy() _copied.table = pd.concat([self.table, other.table], ignore_index=True) return _copied
[docs] def __len__(self): """ :return: Number of rows (individual reflections) in `self.data` :rtype: int """ return self.table.shape[0]
[docs] def __str__(self): """ :return: Human-readable representation of `self.data` :rtype: str """ return self.table.__str__()
@property def la(self): """ Wavelength of radiation used in the diffraction experiment. Can be set using popular abbreviations such as "MoKa" or "CuKb", where *a* and *b* stand for *alpha* and *beta*. Implemented cathode materials include: "Ag", "Co", "Cr", "Cu", "Fe", "Mn", "Mo", "Ni", "Pd", "Rh", "Ti", "Zn" and have been imported from International Tables of Crystallography, Volume C, Table 4.2.4.1, 3rd Edition. :return: wavelength of radiation used in experiment :rtype: float """ return self.__la @la.setter def la(self, wavelength): try: self.__la = characteristic_radiation[wavelength[:4].lower()] except TypeError: self.__la = float(wavelength) @property def r_lim(self): """ :return: Radius of limiting sphere in A^-1 calculated as 2/:attr:`la` :rtype: float """ return 2.0 / self.la
[docs] def _in_dacs(self, opening_angle, vectors): oa = np.deg2rad(opening_angle) xyz = self.table.loc[:, ('x', 'y', 'z')].to_numpy() r = self.table.loc[:, 'r'].to_numpy() v = np.array(vectors) v = (v.T / lin.norm(v, axis=1)).T # normalise vectors v m1 = np.matmul(v, xyz.T) # dist from dac plane p phi = np.abs(np.arcsin((m1 / r).clip(-1, 1))) # angle <(plane p, v) lim = self.r_lim * np.sin(oa - phi) # True if in dac return r[None, :] < lim
[docs] def dac_trim(self, opening_angle: float = 35.0, vector=None): r""" Remove reflections outside the opening_angle DAC-accessible volume. Sample/DAC orientation can be supplied either via specifying crystal orientation in :class:`hikari.dataframes.BaseFrame`, in :attr:`orientation` or providing a xyz\* *vector* perpendicular to the dac-accessible disc. For further details, see `*Tchoń & Makal, IUCrJ 8, 1006-1017 (2021)* <https://doi.org/10.1107/s2052252521009532>`_. :param opening_angle: DAC single opening angle in degrees, default 35. :type opening_angle: float :param vector: Provides information about orientation of crystal relative to DAC. If None, :attr:`orientation` is used instead. :type vector: tuple[float] :return: HklFrame containing only reflections in dac-accessible region. :rtype: HklFrame """ if vector is None: l_v = np.array((1.0, 0.0, 0.0)) # vector parallel to beam hkl = lin.inv(self.orientation) @ l_v # calculate hkl vector v = hkl @ self.A_r # calculate xyz* vector else: v = vector in_dac = self._in_dacs(opening_angle, vectors=np.array([v, ]))[0] self.table = self.table[in_dac] self.table.reset_index(drop=True, inplace=True)
[docs] def dacs_count(self, opening_angle: float = 35.0, vectors: np.ndarray = np.array((1, 0, 0))): """ Count unique dac-accessible reflections for n crystals placed such that vector n is perpendicular to diamond. For details see :meth:`dac_trim`. :param opening_angle: DAC single opening angle in degrees, default 35. :type opening_angle: float :param vectors: Array with rotational axes of available DAC-discs. :type vectors: np.array :return: Array with numbers of unique reflns in DAC-accessible region. :rtype: np.array """ memory_estimate = 26 * len(self) * len(vectors) # estimate memory use cycles_needed = -(memory_estimate // -hikari.MEMORY_SIZE) # and split if cycles_needed > 1: vectors_split = np.array_split(vectors, cycles_needed) return np.hstack([self.dacs_count(opening_angle, vectors=v) for v in vectors_split]) else: in_dac = self._in_dacs(opening_angle, vectors) return np.array([self.table.loc[in_dac[n, :], 'equiv'].nunique() for n in range(vectors.shape[0])])
[docs] def copy(self): """ :return: An exact deep copy of this HklFrame. :rtype: HklFrame """ return copy.deepcopy(self)
[docs] def extinct(self, space_group: Group = SG['P1']): """ Removes from dataframe reflections which should be extinct based on space :class:`hikari.symmetry.group.Group`. For ref. see ITC-A12.3.5. :param space_group: Space group used to extinct the reflections. :type space_group: hikari.symmetry.group.Group """ hkls = self.table.loc[:, ['h', 'k', 'l']].to_numpy() extinct_flag_list = [o.extincts(hkls) for o in space_group.operations] extinct_flag_list_union = np.logical_or.reduce(extinct_flag_list) self.table = self.table[~extinct_flag_list_union] self.table.reset_index(drop=True, inplace=True)
[docs] def find_equivalents(self, point_group: Group = PG['1']): """ Assign each reflection its symmetry equivalence identifier and store it in the `hikari.dataframes.HklFrame.data['equiv']` column. The ID is an integer unique for each set of equivalent reflections. In order to provide an information about equivalence, a *point_group* of reciprocal space must be provided (default PG['1']). Point groups and their notation can be found in :mod:`hikari.symmetry` sub-package. :param point_group: Point group used to determine symmetry equivalence :type point_group: hikari.symmetry.Group """ inc = 10 ** (int(np.log10(self.HKL_LIMIT)) + 2) equiv_dtype = HklKey.REGISTRY['equiv'].dtype self.table.reset_index(drop=True, inplace=True) self.table['equiv'] = -inc**3 _hkl_matrix = self.table.loc[:, ('h', 'k', 'l')].to_numpy() for op in point_group.operations: _new_hkl_matrix = op.transform(_hkl_matrix).astype(equiv_dtype) new_equiv = _new_hkl_matrix @ np.array([inc ** 2, inc, 1]) _to_update = self.table['equiv'] < new_equiv self.table.loc[_to_update, 'equiv'] = new_equiv[_to_update]
[docs] def from_dict(self, dictionary: dict): """ Construct the `self.data` using information stored in dictionary. The dictionary keys must be valid strings, see :class:`HklKeys` for a list of valid keys. The dictionary values must be iterable of equal size, preferably `numpy.ndarray`. :param dictionary: Dictionary with "key - iterable of values" pairs. :type dictionary: Dict[str, numpy.ndarray] """ df = pd.DataFrame() for key, value in dictionary.items(): typ = HklKey.REGISTRY[key].dtype df[key] = pd.Series(value, dtype=typ, name=key) self.table = df[(df['h'] != 0) | (df['k'] != 0) | (df['l'] != 0)].copy() if not('x' in self.table.columns): self.place() self._recalculate_structure_factors_and_intensities()
[docs] def fill(self, radius: float = 2.0) -> None: """ Fill dataframe with all reflections within *radius* from space origin. :param radius: Maximum distance from the reciprocal space origin to placed reflection (in reciprocal Angstrom). """ hkl_ratios = radius / np.array([self.a_r, self.b_r, self.c_r]) hkl_limits = np.ceil(hkl_ratios).astype(np.int16) def hkl_walls(h, k, l_): hw = np.mgrid[h:h:1j, -k:k:2j*k+1j, -l_:l_:2j*l_+1j].reshape(3, -1) kw = np.mgrid[-h:h:2j*h+1j, k:k:1j, -l_:l_:2j*l_+1j].reshape(3, -1) lw = np.mgrid[-h:h:2j*h+1j, -k:k:2j*k+1j, l_:l_:1j].reshape(3, -1) return hw.T, kw.T, lw.T for _ in range(self.HKL_LIMIT): limits_too_small = [any(lin.norm(hkls @ self.A_r, axis=1) <= radius) for hkls in hkl_walls(*hkl_limits)] if any(limits_too_small): hkl_limits += limits_too_small else: hkl_limits -= 1 break if any(hkl_limits > self.HKL_LIMIT): msg = 'Attempting to use hkl indices {} above HKL_LIMIT of {}' raise ValueError(msg.format(hkl_limits, self.HKL_LIMIT)) hkls = np.indices(2 * hkl_limits + 1, np.int16).reshape(3, -1).T \ - hkl_limits _h, _k, _l = hkls[lin.norm(hkls @ self.A_r, axis=1) <= radius].T ones = np.ones_like(np.array(_h)[0]) self.from_dict({'h': np.squeeze(_h), 'k': np.squeeze(_k), 'l': np.squeeze(_l), 'I': ones, 'si': ones, 'm': ones})
[docs] def stats(self, bins: int = 10, space_group: Group = SG['P1']): """ Returns completeness, redundancy, number of all, unique & theoretically possible reflections within equal-volume `bins` in given `space group`. :param bins: Number of equal-volume bins to divide the data into. :type bins: int :param space_group: Group used to calculate equivalence and extinctions :type space_group: hikari.symmetry.Group :return: String containing table with stats as a function of resolution :rtype: str """ point_group = space_group.reciprocate() hkl_base = self.copy() hkl_base.extinct(space_group) hkl_base.find_equivalents(point_group) hkl_base.table['_i_to_si'] = hkl_base.table['I'] / hkl_base.table['si'] hkl_full = self.copy() hkl_full.fill(radius=max(self.table['r'])) hkl_full.merge(point_group) hkl_full.extinct(space_group) hkl_full.find_equivalents(point_group) def group_by_resolution(hkl_): bin_limits = cubespace(0.0, max(self.table['r']), num=bins + 1) return hkl_.table.groupby(by=pd.cut(hkl_.table['r'], bin_limits), observed=False) grouped_base = group_by_resolution(hkl_base) grouped_full = group_by_resolution(hkl_full) observed = grouped_base.size() independent = grouped_base['equiv'].nunique() theory = grouped_full['equiv'].nunique() cpl = independent.div(theory) red = observed.div(independent) i2si = grouped_base['_i_to_si'].mean() out = pd.concat([observed, independent, theory, i2si, cpl, red], axis=1) out.columns = ['Obser', 'Indep', 'Theory', 'I/si(I)', 'Cplt', 'Red.'] return out # use .reset_index().to_string(index=False) to flatten
[docs] def merge(self, point_group=PG['1']): """ Average down each set of redundant reflections present in the table, to one reflection. The redundancy is determined using the :meth:`find_equivalents` method with appropriate point group. Thus, the merging can be used in different ways depending on point group: - For PG['1'], only reflections with exactly the same h, k, l indices will be merged. Resulting dataframe will not contain any duplicates. - For PG['-1'] reflections with the same h, k and l as well as their Friedel pairs will be merged together to one reflection. - For PG['mmm'] all equivalent reflections of "mmm" point group will be merged. Since "mmm" is centrosymmetric, Friedel pairs will be merged. - For PG['mm2'] symmetry-equivalent reflections within the "mmm" point group will be merged, but the Friedel pairs will be preserved. The procedure will have a different effect on different dataframe keys, depending on their "reduce_behaviour" specified in :class:`HklKeys`. Fixed parameters *h, k, l, x, y, z, r* and *equiv* will be preserved; Floating points such as intensity *I*, structure factor *F* and their uncertainties *si* and *sf* will be averaged using arithmetic mean; Multiplicity *m* will be summed; Other parameters which would lose their meaning such as batch number *b* will be discarded. The merging inevitably removes some information from the dataframe, but it can be necessary for some operations. For example, the drawing procedures work faster and provide clearer image if multiple points occupying the same position in space are reduced to one instance. :param point_group: Point Group used to determine symmetry equivalence :type point_group: hikari.symmetry.Group """ self.find_equivalents(point_group=point_group) # group the dataframe and obtain all existing keys grouped = self.table.groupby('equiv') grouped_first = grouped.first().reset_index() grouped_mean = grouped.mean().reset_index() grouped_sum = grouped.sum().reset_index() # for each key apply a necessary reduce operation and add it to data data = dict() for key in self.table.keys(): if HklKey.REGISTRY[key].reduce_behaviour == 'keep': data[key] = grouped_first[key] elif HklKey.REGISTRY[key].reduce_behaviour == 'add': data[key] = grouped_sum[key] elif HklKey.REGISTRY[key].reduce_behaviour == 'average': data[key] = grouped_mean[key] self.from_dict(data)
[docs] def place(self): """ Assign reflections their positions in reciprocal space ("x", "y", "z") and calculate their distance from origin ("r") in reciprocal Angstrom. Save four new keys and their values into the dataframe. """ hkl = self.table.loc[:, ('h', 'k', 'l')].to_numpy() xyz = hkl @ self.A_r self.table.loc[:, 'x'] = xyz[:, 0] self.table.loc[:, 'y'] = xyz[:, 1] self.table.loc[:, 'z'] = xyz[:, 2] self.table.loc[:, 'r'] = lin.norm(xyz, axis=1)
[docs] def calculate_fcf_statistics(self): """ Calculate values of zeta (I - Ic) / si on other stats based on contents of fcf files. Save new key and its values into the dataframe. """ ze = (self.table['I'] - self.table['Ic']) / self.table['si'] self.table['ze'] = ze self.table['ze2'] = ze ** 2 self.table['Iosi'] = self.table['I'] / self.table['si'] self.table['Icsi'] = self.table['Ic'] / self.table['si']
[docs] def read(self, hkl_path, hkl_format='shelx_4'): """ Read the contents of .hkl file as specified by path and format, and store them in the pandas dataframe in `self.data`. For a list of all available .hkl formats, please refer to :attr:`hikari.dataframes.HklIo.format`. :param hkl_path: Absolute or relative path to the .hkl file. :type hkl_path: str :param hkl_format: Format of provided .hkl file. :type hkl_format: union[int, str, dict] """ reader = HklReader(hkl_file_path=hkl_path, hkl_file_format=hkl_format) dict_of_data = reader.read() forgotten_keys = [k for k in HklKey.IMPERATIVES if k not in dict_of_data.keys()] for key in forgotten_keys: default = HklKey.REGISTRY[key].default length_of_data = max([len(v) for v in dict_of_data.values()]) dict_of_data[key] = [default] * length_of_data self.from_dict(dict_of_data)
[docs] def _recalculate_structure_factors_and_intensities(self): """ Calculate 'I' and 'si' or 'F' and 'sf', depending on which are missing. """ if 'I' in self.table.keys() and 'si' not in self.table.keys(): raise KeyError('Intensities "I" are defined, but "si" not.') if 'F' in self.table.keys() and 'sf' not in self.table.keys(): raise KeyError('Structure factors "F" are defined, but "sf" not.') if all(('I' in self.table.keys(), 'si' in self.table.keys(), 'F' not in self.table.keys(), 'sf' not in self.table.keys())): self._recalculate_structure_factors_from_intensities() if all(('F' in self.table.keys(), 'sf' in self.table.keys(), 'I' not in self.table.keys(), 'si' not in self.table.keys())): self._recalculate_intensities_from_structure_factors()
[docs] def _recalculate_structure_factors_from_intensities(self): r""" Recalculate the structure factor F and its uncertainty sf. Structure factor is calculated as follows: *F = signum(I) \* sqrt(abs(I))*. Structure factor's uncertainty is calculated as follows: *sf = si / (2 \* sqrt(abs(I)))*. """ new_data = copy.deepcopy(self.table) signum_of_i = new_data['I'].copy() signum_of_i[signum_of_i > 0] = 1 signum_of_i[signum_of_i < 0] = -1 absolute_sqrt_of_i = abs(new_data["I"]) ** 0.5 new_data['F'] = signum_of_i * absolute_sqrt_of_i new_data['sf'] = new_data["si"] / (2 * absolute_sqrt_of_i) self.table = new_data
[docs] def _recalculate_intensities_from_structure_factors(self): r""" Recalculate the intensity I and its uncertainty si. Intensity is calculated as follows: *I = signum(F) \* F \*\* 2*. Intensity's uncertainty is calculated as follows: *si = 2 \* sf \* abs(F)*. """ new_data = copy.deepcopy(self.table) signum_of_f = new_data['F'].copy() signum_of_f[signum_of_f > 0] = 1 signum_of_f[signum_of_f < 0] = -1 new_data['I'] = signum_of_f * (abs(new_data['F']) ** 2) new_data['si'] = 2 * new_data["sf"] * abs(new_data["F"]) self.table = new_data
[docs] def transform(self, operations: Union[Iterable[np.ndarray], np.ndarray]): """ Apply a symmetry operation or list of symmetry operations to transform the diffraction pattern. If one symmetry operation (3x3 or 4x4 numpy array) is provided, it effectively multiplies the hkl matrix by the operation matrix and accordingly alters the `self.data` dataframe. As a result, the length of `self.data` before and after transformation is the same. However, the function behaves slightly counter-intuitively if two or more operation matrices are provided. In such case the method applies the transformation procedure independently for each operation, and then *concatenates* resulting matrices. Resulting self.data is len(operations) times longer than the initial. The function can use 3x3 or larger (e.g. 4x4) matrices, as it selects only the upper-left 3x3 segment for the sake of calculations. Also, while reconstructing the symmetry of merged reflection file it is important to use all symmetry operations, not only generators. Single symmetry operations or their lists belonging to certain point groups can be imported from :py:mod:`hikari.symmetry` module. :param operations: Iterable of operation matrices to be applied :type operations: Union[Iterable[np.ndarray], np.ndarray] """ def _make_a_list_of_operations(): _ops = np.array(operations) if len(_ops.shape) == 1: raise ValueError('Operations should be a list of matrices') elif len(_ops.shape) == 2 and _ops.shape[0] == _ops.shape[1]: return [_ops] elif len(_ops.shape) == 3 and _ops.shape[1] == _ops.shape[2]: return operations else: raise ValueError('Operations should be a list of matrices') ops = _make_a_list_of_operations() def _build_transformed_dataframe(): dataframes = list() for op in ops: df = copy.deepcopy(self.table) mat = op[0:3, 0:3] hkl = (mat @ self.table.loc[:, ['h', 'k', 'l']].to_numpy().T).T df['h'] = hkl[:, 0] df['k'] = hkl[:, 1] df['l'] = hkl[:, 2] dataframes.append(df) return pd.concat(dataframes, axis=0, ignore_index=True) self.table = _build_transformed_dataframe() self.place()
[docs] def thin_out(self, target_cplt=1.0): """ Randomly remove reflections from dataframe in order to decrease the completeness to *target_cplt* (relatively to initial completeness). :param target_cplt: Percentage of data not removed from dataframe :type target_cplt: float """ assert 0.0 <= target_cplt <= 1.0, 'target_cplt outside the 0--1 range' number_to_delete = int((1 - target_cplt) * len(self)) indices_to_delete = random.sample(range(0, len(self)), number_to_delete) self.table.drop(indices_to_delete, inplace=True) self.table.reset_index(drop=True, inplace=True)
[docs] def to_res(self, path='hkl.res', colored='m'): """ Export the reflection information from table to .res file, so that a software used to visualize .res files can be used to visualize a diffraction data in three dimensions. :param colored: Which key of dataframe should be visualized using color :type colored: str :param path: Absolute or relative path where the file should be saved :type path: str """ HklToResConverter(self).convert(path)
[docs] def trim(self, limit: float): """ Remove reflections further than *limit* from reciprocal space origin. :param limit: Radius of the trimming sphere in reciprocal Angstrom :type limit: float """ self.table = self.table.loc[self.table['r'] <= limit]
[docs] def write(self, hkl_path, hkl_format='shelx_4'): """ Write the contents of dataframe to a .hkl file using specified *path* and *format*. For a list of all available .hkl formats, please refer to :attr:`hikari.dataframes.HklIo.format`. :param hkl_path: Absolute or relative path to the .hkl file. :type hkl_path: str :param hkl_format: Desired format of .hkl file. :type hkl_format: union[int, str, dict] """ writer = HklWriter(hkl_file_path=hkl_path, hkl_file_format=hkl_format) writer.write(hkl_data=self.table)
[docs] class HklIo: """ A helper class supporting HklFrame. Manages reading and writing hkl files into and out of HklFrame's dataframe. """ def __init__(self, hkl_file_path, hkl_file_format): self.use_separator = True self.file_path = make_abspath(hkl_file_path) self.formats_defined = hkl_formats self.formats_aliases = hkl_aliases self.__format = 'shelx_4' self.format = hkl_file_format
[docs] def _build_line_formatter(self): """ Prepare :meth:`line_formatter` to format hkl data while writing. :return: String for str.format() to format hkl data while writing. :rtype: str """ formatter = str() for label, width in zip(self._format_dict['labels'], self._format_dict['widths']): width = abs(width) text_width = abs(width) - int(self.use_separator) formatter += '{{{0}:>{1}.{2}}}'.format(label, width, text_width) formatter += '\n' self.__line_formatter = formatter
@property def format(self): """ Return a name of currently used hkl file format. Available file formats and their aliases are defined internally in .json files and have been presented in a table below: +----------+----------+------------------------+------+------+--------+ | Name | Aliases | Contents |Prefix|Suffix| Free | | | | (format string) | | | format | +==========+==========+========================+======+======+========+ | free_2 | | h -4 k -4 l -4 I -8 | NO | YES | YES | | | | si -8 b -4 la -8 | | (a) | | +----------+----------+------------------------+------+------+--------+ | free_3 | | h -4 k -4 l -4 | NO | YES | YES | | | | F -8 sf -8 b -4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | free_4 | | h -4 k -4 l -4 | NO | YES | YES | | | | I -8 si -8 b -4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | free_40 | free | h -4 k -4 l -4 | NO | YES | YES | | | | I -8 si -8 | | (a) | | +----------+----------+------------------------+------+------+--------+ | free_5 | | h -4 k -4 l -4 | NO | YES | YES | | | | I -8 si -8 c -4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | free_6 | | h -4 k -4 l -4 | NO | YES | YES | | | | I -8 si -8 m -4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | m80 | | h 4 k 4 l 4 b 4 F 12 | YES | YES | NO | | | | None 132 | (b) | (b) | | +----------+----------+------------------------+------+------+--------+ | shelx_2 | 2 | h 4 k 4 l 4 I 8 si 8 | NO | YES | NO | | | | b 4 la 8 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_3 | 3 | h 4 k 4 l 4 | NO | YES | NO | | | | F 8 sf 8 b 4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_4 | 4 | h 4 k 4 l 4 | NO | YES | NO | | | | I 8 si 8 b 4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_40 | 40 | h 4 k 4 l 4 | NO | YES | NO | | | | I 8 si 8 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_5 | 5 | h 4 k 4 l 4 | NO | YES | NO | | | | I 8 si 8 c 4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_6 | 6 | h 4 k 4 l 4 | NO | YES | NO | | | | I 8 si 8 m 4 | | (a) | | +----------+----------+------------------------+------+------+--------+ | shelx_fcf| fcf | h 4 k 4 l 4 | YES | NO | NO | | | | Ic 14 I 14 si 13 | (*) | | | +----------+----------+------------------------+------+------+--------+ | tonto_F | | h -4 k -4 l -4 | YES | YES | YES | | | | F -8 sf -8 | (c) | (c) | | +----------+----------+------------------------+------+------+--------+ | tonto_I | tonto | h -4 k -4 l -4 | YES | YES | YES | | | | I -8 si -8 | (c) | (c) | | +----------+----------+------------------------+------+------+--------+ | xd_F6 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | F -13 sf -13 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_F7 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | F -13 sf -13 t -10 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_F-7 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | F -13 sf -13 ph -10 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_F13 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | F -13 sf -13 t -10 | (d) | | | | | | u1 -10 u2 -10 u3 -10 | | | | | | | v1 -10 v2 -10 v3 -10 | | | | +----------+----------+------------------------+------+------+--------+ | xd_I6 | xd | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | I -13 si -13 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_I7 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | I -13 si -13 t -10 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_I-7 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | I -13 si -13 ph -10 | (d) | | | +----------+----------+------------------------+------+------+--------+ | xd_I13 | | h -4 k -4 l -4 b -3 | YES | NO | YES | | | | I -13 si -13 t -10 | (d) | | | | | | u1 -10 u2 -10 u3 -10 | | | | | | | v1 -10 v2 -10 v3 -10 | | | | +----------+----------+------------------------+------+------+--------+ | *custom* | | custom string as above | NO | NO | if all | | | | with keys and widths | | | widths | | | | | | | in | | | | | | | format | | | | | | | are <0 | +----------+----------+------------------------+------+------+--------+ Three different types of prefix / suffix are supported at the moment: - Suffix (a) is a zero-line: a shelx ending line with h = k = l = 0, - Prefix and suffix (b) express a superflip-style block/start end, - Prefix and suffix (c) are tonto-characteristic beginning/end of file, - Prefix (d) is an xd-characteristic line with info about file content. Pre/suffixes denoted with (*) are not supported in terms of writing. A custom hkl file format can be defined by providing a *format string* instead of 'Name'. The string should look like the ones in column "contents". For the meaning of keys ('I', 'b', 'c' etc.), please refer to :class:`HklKeys`. :return: Returns a name of currently used format. :rtype: str """ return self.__format @format.setter def format(self, new_format): if new_format in self.formats_defined.keys(): self.__format = new_format elif new_format in self.formats_aliases.keys(): self.__format = self.formats_aliases[new_format] else: raise KeyError('Unknown hkl format "{}" given'.format(new_format)) self._build_line_formatter() @property def _format_dict(self): """A dictionary with details concerning current format.""" return self.formats_defined[self.__format] @property def _line_formatter(self): """A string used by string.format() method to print hkl data.""" return self.__line_formatter
[docs] def _import_custom_format(self, custom_format_string): """ Import format string such as 'h 4 k 4 l 4 I 8 si 8' as format 'custom'. :param custom_format_string: string containing alternating data labels and column widths (all negative if free format) separated using ' ' :type custom_format_string: str :return: None """ custom_format_list = custom_format_string.strip(' ').split(' ') custom_labels = custom_format_list[0::2] custom_widths = [int(width) for width in custom_format_list[1::2]] assert len(custom_labels) == len(custom_widths), "Bad custom format" self.formats_defined['custom'] = {'labels': custom_labels, 'widths': custom_widths, 'prefix': '', 'suffix': ''}
@property def is_current_format_free(self): """ Return true if currently defined format is free, i.e. the columns are separated by whitespace. :return: True if all format widths are negative; False otherwise. :rtype: bool """ return all(width < 0 for width in self._format_dict['widths'])
[docs] class HklReader(HklIo): """ A helper class for HklFrame, Manages reading hkl files and importing data and keys from them """ def __init__(self, hkl_file_path, hkl_file_format): super().__init__(hkl_file_path, hkl_file_format)
[docs] def _parse_fixed_line(self, line): """ Parse data from a line, where data from each *label* has fixed *width*. :param line: string to be parsed based on format dictionary. :type line: str :return: list of strings extracted from parsed line :rtype: list """ slice_end = list(np.cumsum(self._format_dict['widths'])) slice_beg = [0] + slice_end[:-1] try: parsed = [line[beg:end] for beg, end in zip(slice_beg, slice_end)] parsed = np.array(parsed) assert min([len(p) for p in parsed]) > 0 assert len(parsed) == len(self._format_dict['widths']) except (ValueError, AssertionError): return None return parsed
[docs] def _parse_free_line(self, line): """ Parse data from line, where data from *labels* is separated with space. :param line: string to be parsed based on format dictionary. :type line: str :return: list of strings extracted from parsed line :rtype: list """ parsed = np.array(line.strip().split()) try: parsed.astype('float64') assert len(parsed) == len(self._format_dict['widths']) except (ValueError, AssertionError): return None return parsed
[docs] def read(self): """ Read the contents of file currently pointed by :attr:`hkl_file_path` and format :attr:`hkl_file_format` and return them to a dictionary. :return: A dictionary containing information read from .hkl file. :rtype: dict """ if self.is_current_format_free: def parse_line(line): return self._parse_free_line(line) else: def parse_line(line): return self._parse_fixed_line(line) def read_file_to_list_of_data(): list_of_reflections = list() with open(self.file_path, 'r') as hkl_file: for line in hkl_file.read().splitlines(): parsed_line = parse_line(line) if parsed_line is None: continue list_of_reflections.append(parsed_line) return np.array(list_of_reflections).transpose() array_of_reflections = read_file_to_list_of_data() def build_dict_of_reflections(_hkl_array): dict_of_data = dict() for index, key in enumerate(self._format_dict['labels']): key_dtype = HklKey.REGISTRY[key].dtype dict_of_data[key] = _hkl_array[index].astype(key_dtype) return dict_of_data return build_dict_of_reflections(array_of_reflections)
[docs] class HklWriter(HklIo): """ A helper class for HklFrame, Manages writing hkl files and exporting data to them """ def __init__(self, hkl_file_path, hkl_file_format): super().__init__(hkl_file_path, hkl_file_format)
[docs] def write(self, hkl_data): """ Write data from pandas dataframe `hkl_data` to the file specified at :attr:`hkl_file_path` of format :attr:`hkl_file_format`. :param hkl_data: Dataframe containing reflection information. :type hkl_data: pandas.dataframe """ needed_data = hkl_data.loc[:, self._format_dict['labels']].astype(str) with open(self.file_path, 'w') as hkl_file: hkl_file.write(self._format_dict['prefix']) for row_tuple in needed_data.itertuples(index=False): row_dict = dict(zip(self._format_dict['labels'], row_tuple)) hkl_file.write(self._line_formatter.format(**row_dict)) hkl_file.write(self._format_dict['suffix'])
[docs] class HklToResConverter: """A class responsible for representing hkl data using .res format""" MIN_DISTANCE = 10.0 ELEMENTS = chemical_elements[:100] MIN_U = 0.00001 MAX_U = 4.99999 def __init__(self, hkl_dataframe): self.df = hkl_dataframe @property def abc_scale_factor(self): return self.largest_absolute_hkl * self.MIN_DISTANCE \ / min(self.df.a_r, self.df.b_r, self.df.c_r) @property def largest_absolute_hkl(self): return self.df.table[['h', 'k', 'l']].abs().max().max() @property def x(self): return self.df.table['h'] / self.largest_absolute_hkl @property def y(self): return self.df.table['k'] / self.largest_absolute_hkl @property def z(self): return self.df.table['l'] / self.largest_absolute_hkl @property def u(self): if 'F' in self.df.table.columns: return rescale_list_to_range(self.df.table['F'], (self.MIN_U, self.MAX_U)) else: return [1.0] * len(self.df.table) @property def c(self): if 'm' in self.df.table.columns: return rescale_list_to_other(self.df.table['m'], self.ELEMENTS) else: return [self.ELEMENTS[0]] * len(self.df.table) @property def res_header(self): return "TITL Reflection visualisation\n" \ "REM Special file to be used in mercury with hkl.msd style.\n" \ "REM Reciprocal unit cell inflated by {sc} to prevent bonds\n" \ "CELL {la:7f} {a:7f} {b:7f} {c:7f} {al:7f} {be:7f} {ga:7f}\n" \ "LATT -1\n\n".format(sc=self.abc_scale_factor, la=self.df.la, a=self.df.a_r * self.abc_scale_factor, b=self.df.b_r * self.abc_scale_factor, c=self.df.c_r * self.abc_scale_factor, al=np.rad2deg(self.df.al_r), be=np.rad2deg(self.df.be_r), ga=np.rad2deg(self.df.ga_r)) @property def atom_list(self): cols = [self.c, self.df.table['h'], self.df.table['k'], self.df.table['l'], self.x, self.y, self.z, self.u] return [list(row) for row in zip(*cols)]
[docs] @staticmethod def res_line(_c, _h, _k, _l, _x, _y, _z, _u): """ :return: res line filled based on input element, hkl, xyz and u. :rtype: str """ label = f'{_c}({_h},{_k},{_l})' pos = f' {_x: 7.5f} {_y: 7.5f} {_z: 7.5f}' return f'{label:16} 1{pos} 11.0 {_u: 7.5f}\n'
[docs] def convert(self, path='~'): with open(make_abspath(path), 'w') as res: res.write(self.res_header) for row in self.atom_list: res.write(self.res_line(*row)) with open(make_abspath(path, '../hkl.msd'), 'w') as style: style.write(hkl_mercury_style)
# TODO wrap table/data in getter/setter and make it automatically place, # TODO refresh, set keys etc.