Source code for pymatgen.io.zeopp

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

"""
Module implementing classes and functions to use Zeo++.

Zeo++ Installation Steps:
========================
1) Zeo++ requires Voro++. Download Voro++ from code.lbl.gov using
   subversion:
   "svn checkout --username anonsvn https://code.lbl.gov/svn/voro/trunk
   Password is anonsvn.
2) Stable version of Zeo++ can be obtained from
   http://www.maciejharanczyk.info/Zeopp/
   Alternatively it can be obtained from code.lbl.gov. Replace voro
   with zeo.
3) (Optional) Install cython from pip
Mac OS X:
4) (a) Edit the Voro++/voro/trunk/config.mk file to suit your environment
   (compiler, linker).
   (b) Run make command
5) (a) Edit the Zeo++/trunk/cython_wrapper/setup.py to correctly point to
   Voro++ directory.
   (b) Run "python setup.py develop" to install Zeo++ python bindings.
   Be patient, it will take a while.
Linux:
4) (a) Edit the Voro++/voro/trunk/config.mk file to suit your environment.
   (b) Also add -fPIC option to CFLAGS variable in config.mk file.
   (c) Run make command
5) (a) Go to Zeo++/zeo/trunk folder and compile zeo++ library using the
   command "make dylib".
   (b) Edit the Zeo++/trunk/cython_wrapper/setup_alt.py to correctly
   point to Voro++ directory.
   (c) Run "python setup_alt.py develop" to install Zeo++ python bindings.

Zeo++ Post-Installation Checking:
==============================
1) Go to pymatgen/io/tests and run "python test_zeoio.py"
   If Zeo++ python bindings are properly installed, the tests should
   pass. One or two tests will be skipped.
b) Go to pymatgen/analysis/defects/tests and run
   "python test_point_defects.py". Lots of tests will be skipped if GULP
   is not installed. But there should be no errors.
"""

import os
import re

from monty.io import zopen
from monty.dev import requires
from monty.tempfile import ScratchDir

from pymatgen.core.structure import Structure, Molecule
from pymatgen.core.lattice import Lattice
from pymatgen.io.cssr import Cssr
from pymatgen.io.xyz import XYZ

try:
    from zeo.netstorage import AtomNetwork
    from zeo.area_volume import volume, surface_area
    from zeo.cluster import prune_voronoi_network_close_node

    zeo_found = True
except ImportError:
    zeo_found = False

__author__ = "Bharat Medasani"
__copyright = "Copyright 2013, The Materials Project"
__version__ = "0.1"
__maintainer__ = "Bharat Medasani"
__email__ = "mbkumar@gmail.com"
__data__ = "Aug 2, 2013"


[docs]class ZeoCssr(Cssr): """ ZeoCssr adds extra fields to CSSR sites to conform with Zeo++ input CSSR format. The coordinate system is rorated from xyz to zyx. This change aligns the pivot axis of pymatgen (z-axis) to pivot axis of Zeo++ (x-axis) for structurural modifications. """ def __init__(self, structure): """ Args: structure: A structure to create ZeoCssr object """ super().__init__(structure) def __str__(self): """ CSSR.__str__ method is modified to padd 0's to the CSSR site data. The padding is to conform with the CSSR format supported Zeo++. The oxidation state is stripped from site.specie Also coordinate system is rotated from xyz to zxy """ output = [ "{:.4f} {:.4f} {:.4f}".format(self.structure.lattice.c, self.structure.lattice.a, self.structure.lattice.b), "{:.2f} {:.2f} {:.2f} SPGR = 1 P 1 OPT = 1".format(self.structure.lattice.gamma, self.structure.lattice.alpha, self.structure.lattice.beta), "{} 0".format(len(self.structure)), "0 {}".format(self.structure.formula) ] for i, site in enumerate(self.structure.sites): # if not hasattr(site, 'charge'): # charge = 0 # else: # charge = site.charge charge = site.charge if hasattr(site, 'charge') else 0 # specie = site.specie.symbol specie = site.species_string output.append( "{} {} {:.4f} {:.4f} {:.4f} 0 0 0 0 0 0 0 0 {:.4f}".format( i + 1, specie, site.c, site.a, site.b, charge ) ) return "\n".join(output)
[docs] @staticmethod def from_string(string): """ Reads a string representation to a ZeoCssr object. Args: string: A string representation of a ZeoCSSR. Returns: ZeoCssr object. """ lines = string.split("\n") toks = lines[0].split() lengths = [float(i) for i in toks] toks = lines[1].split() angles = [float(i) for i in toks[0:3]] # Zeo++ takes x-axis along a and pymatgen takes z-axis along c a = lengths.pop(-1) lengths.insert(0, a) alpha = angles.pop(-1) angles.insert(0, alpha) latt = Lattice.from_parameters(*lengths, *angles) sp = [] coords = [] chrg = [] for l in lines[4:]: m = re.match(r'\d+\s+(\w+)\s+([0-9\-\.]+)\s+([0-9\-\.]+)\s+' + r'([0-9\-\.]+)\s+(?:0\s+){8}([0-9\-\.]+)', l.strip()) if m: sp.append(m.group(1)) # coords.append([float(m.group(i)) for i in xrange(2, 5)]) # Zeo++ takes x-axis along a and pymatgen takes z-axis along c coords.append([float(m.group(i)) for i in [3, 4, 2]]) chrg.append(m.group(5)) return ZeoCssr( Structure(latt, sp, coords, site_properties={'charge': chrg}) )
[docs] @staticmethod def from_file(filename): """ Reads a CSSR file to a ZeoCssr object. Args: filename: Filename to read from. Returns: ZeoCssr object. """ with zopen(filename, "r") as f: return ZeoCssr.from_string(f.read())
[docs]class ZeoVoronoiXYZ(XYZ): """ Class to read Voronoi Nodes from XYZ file written by Zeo++. The sites have an additional column representing the voronoi node radius. The voronoi node radius is represented by the site property voronoi_radius. """ def __init__(self, mol): """ Args: mol: Input molecule holding the voronoi node information """ super().__init__(mol)
[docs] @staticmethod def from_string(contents): """ Creates Zeo++ Voronoi XYZ object from a string. from_string method of XYZ class is being redefined. Args: contents: String representing Zeo++ Voronoi XYZ file. Returns: ZeoVoronoiXYZ object """ lines = contents.split("\n") num_sites = int(lines[0]) coords = [] sp = [] prop = [] coord_patt = re.compile( r"(\w+)\s+([0-9\-\.]+)\s+([0-9\-\.]+)\s+([0-9\-\.]+)\s+" + r"([0-9\-\.]+)" ) for i in range(2, 2 + num_sites): m = coord_patt.search(lines[i]) if m: sp.append(m.group(1)) # this is 1-indexed # coords.append(map(float, m.groups()[1:4])) # this is 0-indexed coords.append([float(j) for j in [m.group(i) for i in [3, 4, 2]]]) prop.append(float(m.group(5))) return ZeoVoronoiXYZ( Molecule(sp, coords, site_properties={'voronoi_radius': prop}) )
[docs] @staticmethod def from_file(filename): """ Creates XYZ object from a file. Args: filename: XYZ filename Returns: XYZ object """ with zopen(filename) as f: return ZeoVoronoiXYZ.from_string(f.read())
def __str__(self): output = [str(len(self._mols[0])), self._mols[0].composition.formula] fmtstr = "{{}} {{:.{0}f}} {{:.{0}f}} {{:.{0}f}} {{:.{0}f}}".format( self.precision ) for site in self._mols[0]: output.append(fmtstr.format( site.specie.symbol, site.z, site.x, site.y, site.properties['voronoi_radius'] )) return "\n".join(output)
[docs]@requires(zeo_found, "get_voronoi_nodes requires Zeo++ cython extension to be " "installed. Please contact developers of Zeo++ to obtain it.") def get_voronoi_nodes(structure, rad_dict=None, probe_rad=0.1): """ Analyze the void space in the input structure using voronoi decomposition Calls Zeo++ for Voronoi decomposition. Args: structure: pymatgen.core.structure.Structure rad_dict (optional): Dictionary of radii of elements in structure. If not given, Zeo++ default values are used. Note: Zeo++ uses atomic radii of elements. For ionic structures, pass rad_dict with ionic radii probe_rad (optional): Sampling probe radius in Angstroms. Default is 0.1 A Returns: voronoi nodes as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure voronoi face centers as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure """ with ScratchDir('.'): name = "temp_zeo1" zeo_inp_filename = name + ".cssr" ZeoCssr(structure).write_file(zeo_inp_filename) rad_file = None rad_flag = False if rad_dict: rad_file = name + ".rad" rad_flag = True with open(rad_file, 'w+') as fp: for el in rad_dict.keys(): fp.write("{} {}\n".format(el, rad_dict[el].real)) atmnet = AtomNetwork.read_from_CSSR( zeo_inp_filename, rad_flag=rad_flag, rad_file=rad_file) vornet, vor_edge_centers, vor_face_centers = \ atmnet.perform_voronoi_decomposition() vornet.analyze_writeto_XYZ(name, probe_rad, atmnet) voro_out_filename = name + '_voro.xyz' voro_node_mol = ZeoVoronoiXYZ.from_file(voro_out_filename).molecule species = ["X"] * len(voro_node_mol.sites) coords = [] prop = [] for site in voro_node_mol.sites: coords.append(list(site.coords)) prop.append(site.properties['voronoi_radius']) lattice = Lattice.from_parameters(*structure.lattice.parameters) vor_node_struct = Structure( lattice, species, coords, coords_are_cartesian=True, to_unit_cell=True, site_properties={"voronoi_radius": prop}) # PMG-Zeo c<->a transformation for voronoi face centers rot_face_centers = [(center[1], center[2], center[0]) for center in vor_face_centers] rot_edge_centers = [(center[1], center[2], center[0]) for center in vor_edge_centers] species = ["X"] * len(rot_face_centers) prop = [0.0] * len(rot_face_centers) # Vor radius not evaluated for fc vor_facecenter_struct = Structure( lattice, species, rot_face_centers, coords_are_cartesian=True, to_unit_cell=True, site_properties={"voronoi_radius": prop}) species = ["X"] * len(rot_edge_centers) prop = [0.0] * len(rot_edge_centers) # Vor radius not evaluated for fc vor_edgecenter_struct = Structure( lattice, species, rot_edge_centers, coords_are_cartesian=True, to_unit_cell=True, site_properties={"voronoi_radius": prop}) return vor_node_struct, vor_edgecenter_struct, vor_facecenter_struct
[docs]def get_high_accuracy_voronoi_nodes(structure, rad_dict, probe_rad=0.1): """ Analyze the void space in the input structure using high accuracy voronoi decomposition. Calls Zeo++ for Voronoi decomposition. Args: structure: pymatgen.core.structure.Structure rad_dict (optional): Dictionary of radii of elements in structure. If not given, Zeo++ default values are used. Note: Zeo++ uses atomic radii of elements. For ionic structures, pass rad_dict with ionic radii probe_rad (optional): Sampling probe radius in Angstroms. Default is 0.1 A Returns: voronoi nodes as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure voronoi face centers as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure """ with ScratchDir('.'): name = "temp_zeo1" zeo_inp_filename = name + ".cssr" ZeoCssr(structure).write_file(zeo_inp_filename) rad_flag = True rad_file = name + ".rad" with open(rad_file, 'w+') as fp: for el in rad_dict.keys(): print("{} {}".format(el, rad_dict[el].real), file=fp) atmnet = AtomNetwork.read_from_CSSR( zeo_inp_filename, rad_flag=rad_flag, rad_file=rad_file) # vornet, vor_edge_centers, vor_face_centers = \ # atmnet.perform_voronoi_decomposition() red_ha_vornet = \ prune_voronoi_network_close_node(atmnet) # generate_simplified_highaccuracy_voronoi_network(atmnet) # get_nearest_largest_diameter_highaccuracy_vornode(atmnet) red_ha_vornet.analyze_writeto_XYZ(name, probe_rad, atmnet) voro_out_filename = name + '_voro.xyz' voro_node_mol = ZeoVoronoiXYZ.from_file(voro_out_filename).molecule species = ["X"] * len(voro_node_mol.sites) coords = [] prop = [] for site in voro_node_mol.sites: coords.append(list(site.coords)) prop.append(site.properties['voronoi_radius']) lattice = Lattice.from_parameters(*structure.lattice.parameters) vor_node_struct = Structure( lattice, species, coords, coords_are_cartesian=True, to_unit_cell=True, site_properties={"voronoi_radius": prop}) return vor_node_struct
[docs]@requires(zeo_found, "get_voronoi_nodes requires Zeo++ cython extension to be " "installed. Please contact developers of Zeo++ to obtain it.") def get_free_sphere_params(structure, rad_dict=None, probe_rad=0.1): """ Analyze the void space in the input structure using voronoi decomposition Calls Zeo++ for Voronoi decomposition. Args: structure: pymatgen.core.structure.Structure rad_dict (optional): Dictionary of radii of elements in structure. If not given, Zeo++ default values are used. Note: Zeo++ uses atomic radii of elements. For ionic structures, pass rad_dict with ionic radii probe_rad (optional): Sampling probe radius in Angstroms. Default is 0.1 A Returns: voronoi nodes as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure voronoi face centers as pymatgen.core.structure.Strucutre within the unit cell defined by the lattice of input structure """ with ScratchDir('.'): name = "temp_zeo1" zeo_inp_filename = name + ".cssr" ZeoCssr(structure).write_file(zeo_inp_filename) rad_file = None rad_flag = False if rad_dict: rad_file = name + ".rad" rad_flag = True with open(rad_file, 'w+') as fp: for el in rad_dict.keys(): fp.write("{} {}\n".format(el, rad_dict[el].real)) atmnet = AtomNetwork.read_from_CSSR( zeo_inp_filename, rad_flag=rad_flag, rad_file=rad_file) out_file = "temp.res" atmnet.calculate_free_sphere_parameters(out_file) if os.path.isfile(out_file) and os.path.getsize(out_file) > 0: with open(out_file, "rt") as fp: output = fp.readline() else: output = "" fields = [val.strip() for val in output.split()][1:4] if len(fields) == 3: fields = [float(field) for field in fields] free_sphere_params = {'inc_sph_max_dia': fields[0], 'free_sph_max_dia': fields[1], 'inc_sph_along_free_sph_path_max_dia': fields[2]} return free_sphere_params
# Deprecated. Not needed anymore
[docs]def get_void_volume_surfarea(structure, rad_dict=None, chan_rad=0.3, probe_rad=0.1): """ Computes the volume and surface area of isolated void using Zeo++. Useful to compute the volume and surface area of vacant site. Args: structure: pymatgen Structure containing vacancy rad_dict(optional): Dictionary with short name of elements and their radii. chan_rad(optional): Minimum channel Radius. probe_rad(optional): Probe radius for Monte Carlo sampling. Returns: volume: floating number representing the volume of void """ with ScratchDir('.'): name = "temp_zeo" zeo_inp_filename = name + ".cssr" ZeoCssr(structure).write_file(zeo_inp_filename) rad_file = None if rad_dict: rad_file = name + ".rad" with open(rad_file, 'w') as fp: for el in rad_dict.keys(): fp.write("{0} {1}".format(el, rad_dict[el])) atmnet = AtomNetwork.read_from_CSSR(zeo_inp_filename, True, rad_file) vol_str = volume(atmnet, 0.3, probe_rad, 10000) sa_str = surface_area(atmnet, 0.3, probe_rad, 10000) vol = None sa = None for line in vol_str.split("\n"): if "Number_of_pockets" in line: fields = line.split() if float(fields[1]) > 1: vol = -1.0 break if float(fields[1]) == 0: vol = -1.0 break vol = float(fields[3]) for line in sa_str.split("\n"): if "Number_of_pockets" in line: fields = line.split() if float(fields[1]) > 1: # raise ValueError("Too many voids") sa = -1.0 break if float(fields[1]) == 0: sa = -1.0 break sa = float(fields[3]) if not vol or not sa: raise ValueError("Error in zeo++ output stream") return vol, sa