Source code for pymatgen.analysis.xas.spectrum

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

"""
This module defines classes to represent all xas and stitching methods
"""
import math
from typing import List
from scipy.interpolate import interp1d
import numpy as np
from pymatgen.analysis.structure_matcher import StructureMatcher
from pymatgen.core.spectrum import Spectrum
from pymatgen.symmetry.analyzer import SpacegroupAnalyzer

__author__ = "Chen Zheng, Yiming Chen"
__copyright__ = "Copyright 2012, The Materials Project"
__version__ = "3.0"
__maintainer__ = "Yiming Chen"
__email__ = "chz022@ucsd.edu, yic111@ucsd.edu"
__date__ = "April 25, 2020"


[docs]class XAS(Spectrum): """ Basic XAS object. Args: x: A sequence of x-ray energies in eV y: A sequence of mu(E) structure (Structure): Structure associated with the spectrum absorbing_element (Element): Element associated with the spectrum edge (str): Absorption edge associated with the spectrum spectrum_type (str): 'XANES' or 'EXAFS' absorbing_index (None or int): If None, the spectrum is assumed to be a site-weighted spectrum, which is comparable to experimental one. Otherwise, it indicates the absorbing_index for a site-wise spectrum. .. attribute: x The sequence of energies .. attribute: y The sequence of mu(E) .. attribute: absorbing_element The absorbing_element of the spectrum .. attribute: edge The edge of the spectrum .. attribute: spectrum_type XANES or EXAFS spectrum .. attribute: absorbing_index The absorbing_index of the spectrum """ XLABEL = 'Energy' YLABEL = 'Intensity' def __init__(self, x, y, structure, absorbing_element, edge="K", spectrum_type="XANES", absorbing_index=None): """ Initializes a spectrum object. """ super(XAS, self).__init__(x, y, structure, absorbing_element, edge) self.structure = structure self.absorbing_element = absorbing_element self.edge = edge self.spectrum_type = spectrum_type self.e0 = self.x[np.argmax(np.gradient(self.y) / np.gradient(self.x))] # Wavenumber, k is calculated based on equation # k^2=2*m/(hbar)^2*(E-E0) self.k = [np.sqrt((i-self.e0) / 3.8537) if i > self.e0 else -np.sqrt((self.e0-i) / 3.8537) for i in self.x] self.absorbing_index = absorbing_index def __str__(self): return "%s %s Edge %s for %s: %s" % ( self.absorbing_element, self.edge, self.spectrum_type, self.structure.composition.reduced_formula, super(XAS, self).__str__() )
[docs] def stitch(self, other: 'XAS', num_samples: int = 500, mode: str = "XAFS") \ -> 'XAS': """ Stitch XAS objects to get the full XAFS spectrum or L23 edge XANES spectrum depending on the mode. 1. Use XAFS mode for stitching XANES and EXAFS with same absorption edge. The stitching will be performed based on wavenumber, k. for k <= 3, XAS(k) = XAS[XANES(k)] for 3 < k < max(xanes_k), will interpolate according to XAS(k)=f(k)*mu[XANES(k)]+(1-f(k))*mu[EXAFS(k)] where f(k)=cos^2((pi/2) (k-3)/(max(xanes_k)-3) for k > max(xanes_k), XAS(k) = XAS[EXAFS(k)] 2. Use L23 mode for stitching L2 and L3 edge XANES for elements with atomic number <=30. Args: other: Another XAS object. num_samples(int): Number of samples for interpolation. mode(str): Either XAFS mode for stitching XANES and EXAFS or L23 mode for stitching L2 and L3. Returns: XAS object: The stitched spectrum. """ m = StructureMatcher() if not m.fit(self.structure, other.structure): raise ValueError( "The input structures for spectra mismatch") if not self.absorbing_element == other.absorbing_element: raise ValueError("The absorbing elements for spectra are different") if not self.absorbing_index == other.absorbing_index: raise ValueError("The absorbing indexes for spectra are different") if mode == "XAFS": if not self.edge == other.edge: raise ValueError("Only spectrum with the same absorption " "edge can be stitched in XAFS mode.") if self.spectrum_type == other.spectrum_type: raise ValueError("Need one XANES and one EXAFS spectrum to " "stitch in XAFS mode") xanes = self if self.spectrum_type == "XANES" else other exafs = self if self.spectrum_type == "EXAFS" else other if max(xanes.x) < min(exafs.x): raise ValueError( "Energy overlap between XANES and EXAFS is needed for stitching") # for k <= 3 wavenumber, mu = [], [] # type: List[float], List[float] idx = xanes.k.index(min(self.k, key=lambda x: (abs(x - 3), x))) mu.extend(xanes.y[:idx]) wavenumber.extend(xanes.k[:idx]) # for 3 < k < max(xanes.k) fs = [] # type: List[float] ks = np.linspace(3, max(xanes.k), 50) for k in ks: f = np.cos((math.pi / 2) * (k - 3) / (max(xanes.k) - 3)) ** 2 fs.append(f) f_xanes = interp1d( np.asarray(xanes.k), np.asarray(xanes.y), bounds_error=False, fill_value=0) f_exafs = interp1d( np.asarray(exafs.k), np.asarray(exafs.y), bounds_error=False, fill_value=0) mu_xanes = f_xanes(ks) mu_exafs = f_exafs(ks) mus = [fs[i] * mu_xanes[i] + (1 - fs[i]) * mu_exafs[i] for i in np.arange(len(ks))] mu.extend(mus) wavenumber.extend(ks) # for k > max(xanes.k) idx = exafs.k.index(min(exafs.k, key=lambda x: (abs(x - max(xanes.k))))) mu.extend(exafs.y[idx:]) wavenumber.extend(exafs.k[idx:]) # interpolation f_final = interp1d( np.asarray(wavenumber), np.asarray(mu), bounds_error=False, fill_value=0) wavenumber_final = np.linspace(min(wavenumber), max(wavenumber), num=num_samples) mu_final = f_final(wavenumber_final) energy_final = [3.8537 * i ** 2 + xanes.e0 if i > 0 else -3.8537 * i ** 2 + xanes.e0 for i in wavenumber_final] return XAS(energy_final, mu_final, self.structure, self.absorbing_element, xanes.edge, "XAFS") if mode == "L23": if self.spectrum_type != "XANES" or \ other.spectrum_type != "XANES": raise ValueError("Only XANES spectrum can be stitched in " "L23 mode.") if self.edge not in ["L2", "L3"] or other.edge not in ["L2", "L3"] \ or self.edge == other.edge: raise ValueError("Need one L2 and one L3 edge spectrum to stitch" "in L23 mode.") l2_xanes = self if self.edge == "L2" else other l3_xanes = self if self.edge == "L3" else other if l2_xanes.absorbing_element.number > 30: raise ValueError( "Does not support L2,3-edge XANES for {} element" .format(l2_xanes.absorbing_element)) l2_f = interp1d( l2_xanes.x, l2_xanes.y, bounds_error=False, fill_value=0) # will add value of l3.y[-1] to avoid sudden change in absorption coeff. l3_f = interp1d( l3_xanes.x, l3_xanes.y, bounds_error=False, fill_value=l3_xanes.y[-1]) energy = np.linspace(min(l3_xanes.x), max(l2_xanes.x), num=num_samples) mu = [i + j for i, j in zip(l2_f(energy), l3_f(energy))] return XAS(energy, mu, self.structure, self.absorbing_element, "L23", "XANES") raise ValueError("Invalid mode. Only XAFS and L23 are supported.")
[docs]def site_weighted_spectrum(xas_list: List['XAS'], num_samples: int = 500) -> 'XAS': """ Obtain site-weighted XAS object based on site multiplicity for each absorbing index and its corresponding site-wise spectrum. Args: xas_list([XAS]): List of XAS object to be weighted num_samples(int): Number of samples for interpolation Returns: XAS object: The site-weighted spectrum """ m = StructureMatcher() groups = m.group_structures([i.structure for i in xas_list]) if len(groups) > 1: raise ValueError("The input structures mismatch") if not len({i.absorbing_element for i in xas_list}) == \ len({i.edge for i in xas_list}) == 1: raise ValueError("Can only perform site-weighting for spectra with " "same absorbing element and same absorbing edge.") if len({i.absorbing_index for i in xas_list}) == 1 or \ None in {i.absorbing_index for i in xas_list}: raise ValueError("Need at least two site-wise spectra to perform " "site-weighting") sa = SpacegroupAnalyzer(groups[0][0]) ss = sa.get_symmetrized_structure() maxes, mines = [], [] fs = [] multiplicities = [] for xas in xas_list: multiplicity = len( ss.find_equivalent_sites(ss[xas.absorbing_index])) multiplicities.append(multiplicity) maxes.append(max(xas.x)) mines.append(min(xas.x)) # use 3rd-order spline interpolation for mu (idx 3) vs energy (idx 0). f = interp1d( np.asarray(xas.x), np.asarray(xas.y), bounds_error=False, fill_value=0, kind='cubic') fs.append(f) # Interpolation within the intersection of x-axis ranges. x_axis = np.linspace(max(mines), min(maxes), num=num_samples) weighted_spectrum = np.zeros(num_samples) sum_multiplicities = sum(multiplicities) for i in range(len(multiplicities)): weighted_spectrum += (multiplicities[i] * fs[i](x_axis)) \ / sum_multiplicities return XAS(x_axis, weighted_spectrum, ss, xas.absorbing_element, xas.edge, xas.spectrum_type)