dig.ggraph3D.utils

Utilities under dig.ggraph3D.utils.

collect_bond_dists	Collect the lengths for each type of chemical bond in given valid molecular geometries.
compute_mmd	Calculate the maximum mean discrepancy distance between two sample set.
compute_prop	Calculate the quantum property score of the given molecular geometry with PySCF.
xyz2mol	Convert the given molecular geometry to a 2D molecular graph and check if it is chemically valid through valency check with the algorithm in this paper.

collect_bond_dists(mols_dict, valid_list, con_mat_list)

Collect the lengths for each type of chemical bond in given valid molecular geometries.

Parameters

• mol_dicts (dict) – A python dict where the key is the number of atoms, and the value indexed by that key is another python dict storing the atomic number matrix (indexed by the key ‘_atomic_numbers’) and the coordinate tensor (indexed by the key ‘_positions’) of all generated molecular geometries with that atom number.

• valid_list (list) – the list of bool values indicating whether each molecular geometry is chemically valid. Note that only the bond lengths of valid molecular geometries will be collected.

• con_mat_list (list) – the list of bond order matrices.

Return type

dict a python dict where the key is the bond type, and the value indexed by that key is the list of all bond lengths of that bond.

compute_mmd(source, target, batch_size=1000, kernel_mul=2.0, kernel_num=5, fix_sigma=None)

Calculate the maximum mean discrepancy distance between two sample set. This implementation is based on this open source code.

Parameters

• source (pytorch tensor) – the pytorch tensor containing data samples of the source distribution.

• target (pytorch tensor) – the pytorch tensor containing data samples of the target distribution.

Return type

float

compute_prop(atomic_number, position, prop_name)

Calculate the quantum property score of the given molecular geometry with PySCF.

Parameters

• atomic_number (numpy array) – the numpy array indicating the atomic number of atoms in the molecular geometry.

• position (numpy array) – the numpy array indicating the coordinates of atoms in the molecular geometry.

• prop_name (string) – the name of quantum property, ‘gap’ for HOMO-LUMO gap, ‘alpha’ for isotropic polarizability.

Return type

float

xyz2mol(atoms, coordinates, use_graph=True)

Convert the given molecular geometry to a 2D molecular graph and check if it is chemically valid through valency check with the algorithm in this paper.

Parameters

• atoms (numpy array) – the numpy array indicating the atomic number of atoms in the molecular geometry.

• coordinates (numpy array) – the numpy array indicating the coordinates of atoms in the molecular geometry.

• use_graph (bool) – whether use networkx to compute the maximum-weighted matching of the graph or not. (default: True)

Return type

BO: bond order matrix. valid: a bool value denoting whether the molecular geometry is valid or not.