posym 1.2.2

posym module

PoSym

A point symmetry analysis tool written in python designed for theoretical chemistry. This tool makes use of continuous symmetry measures (CSM) to provide a robust implementation to compute the symmetry of chemistry objects such as normal modes, wave function and electronic density.

Features

• Use as simple calculator for irreducible representations supporting direct sum and product
• Continuous symmetry measures (CSM) expressed in the basis or irreducible representation
• Determine symmetry of:
  • normal modes
  • functions defined in gaussian basis (molecular orbitals, electronic densities, operators)
  • wave functions defined as a slater determinant
  • wave functions defined as linear combination of slater determinants (Multi-reference/CI)
• Autogenerated high precision symmetry tables
• Compatibility with PySCF (https://pyscf.org) and PyQchem (http://www.github.com/abelcarreras/pyqchem)
• Designed to be easily extendable to other objects by subclassing the SymmetryObject main class

Requisites

• numpy
• scipy
• pandas
• yaml

Use as a simple symmetry calculation

Posym allows to create basic continuous symmetry python objects that can be operated using direct sum (+) and direct product (*).

from posym import PointGroup, SymmetryObject

pg = PointGroup(group='Td')
print(pg)

a1 = SymmetryObject(group='Td', rep='A1')
a2 = SymmetryObject(group='Td', rep='A2')
e = SymmetryObject(group='Td', rep='E')
t1 = SymmetryObject(group='Td', rep='T1')

print('t1 * t1:', t1 * t1)
print('t1 * e:', t1 * e)
print('e * (e + a1):', e * (e + a1))

Determine the symmetry of normal modes

Symmetry objects can be obtained from normal modes using SymmetryModes.

from posym import SymmetryNormalModes

coordinates = [[0.00000, 0.0000000, -0.0808819],
               [-1.43262, 0.0000000, -1.2823700],
               [1.43262, 0.0000000, -1.2823700]]

symbols = ['O', 'H', 'H']

normal_modes = [[[0., 0., -0.075],
                 [-0.381, -0., 0.593],
                 [0.381, -0., 0.593]],  # mode 1

                [[-0., -0., 0.044],
                 [-0.613, -0., -0.35],
                 [0.613, 0., -0.35]],  # mode 2

                [[-0.073, -0., -0.],
                 [0.583, 0., 0.397],
                 [0.583, 0., -0.397]]]  # mode 3

frequencies = [1737.01, 3988.5, 4145.43]

sym_modes_gs = SymmetryNormalModes(group='c2v', coordinates=coordinates, modes=normal_modes, symbols=symbols)
for i in range(len(normal_modes)):
  print('Mode {:2}: {:8.3f} :'.format(i + 1, frequencies[i]), sym_modes_gs.get_state_mode(i))

print('Total symmetry: ', sym_modes_gs)

Determine the symmetry of a molecular geometry

Continuous symmetry measure (CSM) is obtained using measure method.

from posym import SymmetryMolecule

coordinates = [[0.0000000000, 0.0000000000, 0.0000000000],
               [0.5541000000, 0.7996000000, 0.4965000000],
               [0.6833000000, -0.8134000000, -0.2536000000],
               [-0.7782000000, -0.3735000000, 0.6692000000],
               [-0.4593000000, 0.3874000000, -0.9121000000]]

symbols = ['C', 'H', 'H', 'H', 'H']

sym_geom = SymmetryMolecule(group='Td', coordinates=coordinates, symbols=symbols)
print('Symmetry measure Td : ', sym_geom.measure)

sym_geom = SymmetryMolecule(group='C3v', coordinates=coordinates, symbols=symbols)
print('Symmetry measure C3v : ', sym_geom.measure)

sym_geom = SymmetryMolecule(group='C4v', coordinates=coordinates, symbols=symbols)
print('Symmetry measure C4v : ', sym_geom.measure)

Define basis set functions in gaussian basis

Define basis function as linear combination of gaussian that act as normal python functions

from posym.basis import PrimitiveGaussian, BasisFunction

# Oxigen atom
sa = PrimitiveGaussian(alpha=130.70932)
sb = PrimitiveGaussian(alpha=23.808861)
sc = PrimitiveGaussian(alpha=6.4436083)
s_O = BasisFunction([sa, sb, sc],
                    [0.154328969, 0.535328136, 0.444634536],
                    center=[0.0000000000, 0.000000000, -0.0808819]) # Bohr

sa = PrimitiveGaussian(alpha=5.03315132)
sb = PrimitiveGaussian(alpha=1.1695961)
sc = PrimitiveGaussian(alpha=0.3803890)
s2_O = BasisFunction([sa, sb, sc],
                     [-0.099967228, 0.399512825, 0.700115461],
                     center=[0.0000000000, 0.000000000, -0.0808819])

pxa = PrimitiveGaussian(alpha=5.0331513, l=[1, 0, 0])
pxb = PrimitiveGaussian(alpha=1.1695961, l=[1, 0, 0])
pxc = PrimitiveGaussian(alpha=0.3803890, l=[1, 0, 0])

pya = PrimitiveGaussian(alpha=5.0331513, l=[0, 1, 0])
pyb = PrimitiveGaussian(alpha=1.1695961, l=[0, 1, 0])
pyc = PrimitiveGaussian(alpha=0.3803890, l=[0, 1, 0])

pza = PrimitiveGaussian(alpha=5.0331513, l=[0, 0, 1])
pzb = PrimitiveGaussian(alpha=1.1695961, l=[0, 0, 1])
pzc = PrimitiveGaussian(alpha=0.3803890, l=[0, 0, 1])

px_O = BasisFunction([pxa, pxb, pxc],
                     [0.155916268, 0.6076837186, 0.3919573931],
                     center=[0.0000000000, 0.000000000, -0.0808819])
py_O = BasisFunction([pya, pyb, pyc],
                     [0.155916268, 0.6076837186, 0.3919573931],
                     center=[0.0000000000, 0.000000000, -0.0808819])
pz_O = BasisFunction([pza, pzb, pzc],
                     [0.155916268, 0.6076837186, 0.3919573931],
                     center=[0.0000000000, 0.000000000, -0.0808819])

# Hydrogen atoms
sa = PrimitiveGaussian(alpha=3.42525091)
sb = PrimitiveGaussian(alpha=0.62391373)
sc = PrimitiveGaussian(alpha=0.1688554)
s_H = BasisFunction([sa, sb, sc],
                    [0.154328971, 0.535328142, 0.444634542],
                    center=[-1.43262, 0.000000000, -1.28237])

s2_H = BasisFunction([sa, sb, sc],
                     [0.154328971, 0.535328142, 0.444634542],
                     center=[1.43262, 0.000000000, -1.28237])

basis_set = [s_O, s2_O, px_O, py_O, pz_O, s_H, s2_H]

# Operate with basis functions in analytic form

px_O2 = px_O * px_O
print('integral from -inf to inf:', px_O2.integrate)

# plot functions
from matplotlib import pyplot as plt
import numpy as np

xrange = np.linspace(-5, 5, 100)
plt.plot(xrange, [s_O(x, 0, 0) for x in xrange] , label='s_O')
plt.plot(xrange, [px_O(x, 0, 0) for x in xrange] , label='px_O')
plt.legend()

Create molecular orbitals from basis set

Define molecular orbitals straightforwardly from molecular orbitals coefficients using usual operators

# Orbital 1
o1 = s_O * 0.994216442 + s2_O * 0.025846814 + px_O * 0.0 + py_O * 0.0 + pz_O * -0.004164076 + s_H * -0.005583712 + s2_H * -0.005583712

# Orbital 2
o2 = s_O * 0.23376666 + s2_O * -0.844456594 + px_O * 0.0 + py_O * 0.0 + pz_O * 0.122829781 + s_H * -0.155593214 + s2_H * -0.155593214

# Orbital 3
o3 = s_O * 0.0 + s2_O * 0.0 + px_O * 0.612692349 + py_O * 0.0 + pz_O * 0.0 + s_H * -0.44922168 + s2_H * 0.449221684

# Orbital 4
o4 = s_O * -0.104033343 + s2_O * 0.538153649 + px_O * 0.0 + py_O * 0.0 + pz_O * 0.755880259 + s_H * -0.295107107 + s2_H * -0.2951071074

# Orbital 5
o5 = s_O * 0.0 + s2_O * 0.0 + px_O * 0.0 + py_O * -1.0 + pz_O * 0.0 + s_H * 0.0 + s2_H * 0.0

# Orbital 6
o6 = s_O * -0.125818566 + s2_O * 0.820120983 + px_O * 0.0 + py_O * 0.0 + pz_O * -0.763538862 + s_H * -0.769155124 + s2_H * -0.769155124


# Check orthogonality
print('<o1|o1>: ', (o1*o1).integrate)
print('<o2|o2>: ', (o2*o2).integrate)
print('<o1|o2>: ', (o1*o2).integrate)

Analyze symmetry of molecular orbitals

Get symmetry of molecular orbitals defined as BasisFunction type objects

from posym import SymmetryGaussianLinear

sym_o1 = SymmetryGaussianLinear('c2v', o1)
sym_o2 = SymmetryGaussianLinear('c2v', o2)
sym_o3 = SymmetryGaussianLinear('c2v', o3)
sym_o4 = SymmetryGaussianLinear('c2v', o4)
sym_o5 = SymmetryGaussianLinear('c2v', o5)
sym_o6 = SymmetryGaussianLinear('c2v', o6)

print('Symmetry O1: ', sym_o1)
print('Symmetry O2: ', sym_o2)
print('Symmetry O3: ', sym_o3)
print('Symmetry O4: ', sym_o4)
print('Symmetry O5: ', sym_o5)
print('Symmetry O6: ', sym_o6)

# Operate molecular orbitals symmetries to get the symmetry of non-degenerate wave functions

# restricted close shell
sym_wf_gs = sym_o1 * sym_o1 * sym_o2 * sym_o2 * sym_o3 * sym_o3 * sym_o4 * sym_o4 * sym_o5 * sym_o5
print('Symmetry WF (ground state): ', sym_wf_gs)

# restricted open shell
sym_wf_excited_1 = sym_o1 * sym_o1 * sym_o2 * sym_o2 * sym_o3 * sym_o3 * sym_o4 * sym_o4 * sym_o5 * sym_o6
print('Symmetry WF (excited state 1): ', sym_wf_excited_1)

# restricted close shell
sym_wf_excited_2 = sym_o1 * sym_o1 * sym_o2 * sym_o2 * sym_o3 * sym_o3 * sym_o4 * sym_o4 * sym_o6 * sym_o6
print('Symmetry WF (excited state 2): ', sym_wf_excited_2)

Compute the symmetry of wave functions defined as a Slater determinant

Use SymmetryWaveFunction class to determine the symmetry of a wave function from a set of occupied molecular orbitals defined as BasisFunction objects

from posym import SymmetrySingleDeterminant
from posym.tools import build_orbital

# get orbitals from basis set and MO coefficients
orbital1 = build_orbital(basis_set, coefficients['alpha'][0])  # A1
orbital2 = build_orbital(basis_set, coefficients['alpha'][1])  # A1
orbital3 = build_orbital(basis_set, coefficients['alpha'][2])  # T1
orbital4 = build_orbital(basis_set, coefficients['alpha'][3])  # T1
orbital5 = build_orbital(basis_set, coefficients['alpha'][4])  # T1

wf_sym = SymmetrySingleDeterminant('Td',
                                   alpha_orbitals=[orbital1, orbital2, orbital5],
                                   beta_orbitals=[orbital1, orbital2, orbital4],
                                   center=[0, 0, 0])

print('Configuration 1: ', wf_sym)  # T1 + T2

wf_sym = SymmetrySingleDeterminant('Td',
                                   alpha_orbitals=[orbital1, orbital2, orbital3],
                                   beta_orbitals=[orbital1, orbital2, orbital3],
                                   center=[0, 0, 0])

print('Configuration 2: ', wf_sym)  # A1 + E

Compute the symmetry of multi-reference wave functions

Use SymmetryWaveFunctionCI class to determine the symmetry of multi-reference wave function (defined as a liner combination of Slater determinants) from a set of occupied molecular orbitals defined as BasisFunction objects and a configurations dictionary.

from posym import SymmetryMultiDeterminant

configurations = [{'amplitude': -0.03216, 'occupations': {'alpha': [1, 1, 0, 0, 1], 'beta': [1, 1, 1, 0, 0]}},
                  {'amplitude': 0.70637, 'occupations': {'alpha': [1, 1, 0, 1, 0], 'beta': [1, 1, 1, 0, 0]}},
                  {'amplitude': 0.03216, 'occupations': {'alpha': [1, 1, 1, 0, 0], 'beta': [1, 1, 0, 0, 1]}},
                  {'amplitude': -0.70637, 'occupations': {'alpha': [1, 1, 1, 0, 0], 'beta': [1, 1, 0, 1, 0]}}]

wf_sym = SymmetryMultiDeterminant('Td',
                                  orbitals=[orbital1, orbital2, orbital3, orbital4, orbital5],
                                  configurations=configurations,
                                  center=[0, 0, 0])

print('State 1: ', wf_sym)  # T1

Compatible with pySCF

Usage of helper functions to interface with pySCF

from posym import SymmetryGaussianLinear
from posym.tools import get_basis_set_pyscf, build_orbital
from pyscf import gto, scf
import numpy as np


r = 1  # O-H distance
alpha = np.deg2rad(104.5)  # H-O-H angle

mol_pyscf = gto.M(atom=[['O', [0, 0, 0]],
                        ['H', [-r, 0, 0]],
                        ['H', [r*np.cos(np.pi - alpha), r*np.sin(np.pi - alpha), 0]]],
                  basis='3-21g',
                  charge=0,
                  spin=0)

# run pySCF calculation
pyscf_scf = scf.RHF(mol_pyscf)
pyscf_scf = pyscf_scf.run()

# get electronic structure data
mo_coefficients = pyscf_scf.mo_coeff.T
overlap_matrix = pyscf_scf.get_ovlp(mol_pyscf)
basis_set = get_basis_set_pyscf(mol_pyscf)

# compute symmetry of Molecular orbitals
print('\nMO symmetry')
for i, orbital_vect in enumerate(mo_coefficients):
    orb = build_orbital(basis_set, orbital_vect)
    sym_orb = SymmetryGaussianLinear('c2v', orb)
    print('orbital {}: {}'.format(i, sym_orb))

Combine with PyQchem to create useful automations

PyQchem (https://github.com/abelcarreras/PyQchem) is a Python interface for Q-Chem (https://www.q-chem.com). PyQchem can be used to obtain wave functions and normal modes as Python objects that can be directly used in Posym.

from pyqchem import get_output_from_qchem, QchemInput, Structure
from pyqchem.parsers.basic import basic_parser_qchem
from posym import SymmetryGaussianLinear
# convenient functions to connect pyqchem - posym
from posym.tools import get_basis_set, build_orbital

# define molecules
butadiene = Structure(coordinates=[[-1.07076839, -2.13175980, 0.03234382],
                                   [-0.53741536, -3.05918866, 0.04995793],
                                   [-2.14073783, -2.12969357, 0.04016267],
                                   [-0.39112115, -0.95974916, 0.00012984],
                                   [0.67884827, -0.96181542, -0.00769025],
                                   [-1.15875076, 0.37505495, -0.02522296],
                                   [-0.62213437, 1.30041753, -0.05065831],
                                   [-2.51391203, 0.37767199, -0.01531698],
                                   [-3.04726506, 1.30510083, -0.03293196],
                                   [-3.05052841, -0.54769055, 0.01011971]],
                      symbols=['C', 'H', 'H', 'C', 'H', 'C', 'H', 'C', 'H', 'H'])

# create qchem input
qc_input = QchemInput(butadiene,
                      jobtype='sp',
                      exchange='hf',
                      basis='sto-3g',
                      )

# calculate and parse qchem output
data, ee = get_output_from_qchem(qc_input,
                                 read_fchk=True,
                                 processors=4,
                                 parser=basic_parser_qchem)

# extract required information from Q-Chem calculation
coordinates = ee['structure'].get_coordinates()
mo_coefficients = ee['coefficients']['alpha']
basis = ee['basis']

# print results
print('Molecular orbitals (alpha) symmetry')
basis_set = get_basis_set(coordinates, basis)
for i, orbital_coeff in enumerate(mo_coefficients):
  orbital = build_orbital(basis_set, orbital_coeff)
  sym_orbital = SymmetryGaussianLinear('c2v', orbital)
  print('Symmetry O{}: '.format(i + 1), sym_orbital)

Try an interactive example in Google Colab

Bibliography

This software is based on the theory described in the following works:

Pinsky M, Dryzun C, Casanova D, Alemany P, Avnir D, J Comput Chem. 29:2712-21 (2008) [link]
Pinsky M, Casanova D, Alemany P, Alvarez S, Avnir D, Dryzun C, Kizner Z, Sterkin A. J Comput Chem. 29:190-7 (2008) [link]
Casanova D, Alemany P. Phys Chem Chem Phys. 12(47):15523–9 (2010) [link]
Casanova D, Alemany P, Falceto A, Carreras A, Alvarez S. J Comput Chem 34(15):1321–31 (2013) [link]
A. Carreras, E. Bernuz, X. Marugan, M. Llunell, P. Alemany, Chem. Eur. J. 25, 673 – 691 (2019) [link]

Contact info

Abel Carreras
[e-mail]

Multiverse Computing SL
Donostia-San Sebastian (Spain)