pymatgen.optimization package

Optimization utilities.

Submodules

pymatgen.optimization.linear_assignment module

This module contains the LAPJV algorithm to solve the Linear Assignment Problem.

class LinearAssignment(costs: np.ndarray, epsilon: float = 1e-13)[source]

Bases: object

This class finds the solution to the Linear Assignment Problem. It finds a minimum cost matching between two sets, given a cost matrix.

This class is an implementation of the LAPJV algorithm described in: R. Jonker, A. Volgenant. A Shortest Augmenting Path Algorithm for Dense and Sparse Linear Assignment Problems. Computing 38, 325-340 (1987)

Parameters:

  • costs – The cost matrix of the problem. cost[i,j] should be the cost of matching x[i] to y[j]. The cost matrix may be rectangular

  • epsilon – Tolerance for determining if solution vector is < 0

min_cost[source]

The minimum cost of the matching.

solution[source]

The matching of the rows to columns. i.e solution = [1, 2, 0] would match row 0 to column 1, row 1 to column 2 and row 2 to column 0. Total cost would be c[0, 1] + c[1, 2] + c[2, 0].

pymatgen.optimization.neighbors module

find_points_in_spheres(all_coords, center_coords, r, pbc, lattice, tol=1e-08, min_r=1.0)[source]

For each point in center_coords, get all the neighboring points in all_coords that are within the cutoff radius r. All the coordinates should be Cartesian.

Parameters:

  • all_coords – (np.ndarray[double, dim=2]) all available points. When periodic boundary is considered, this is all the points in the lattice.

  • center_coords – (np.ndarray[double, dim=2]) all centering points

  • r – (float) cutoff radius

  • pbc – (np.ndarray[np.int64_t, dim=1]) whether to set periodic boundaries

  • lattice – (np.ndarray[double, dim=2]) 3x3 lattice matrix

  • tol – (float) numerical tolerance

  • min_r – (float) minimal cutoff to calculate the neighbor list directly. If the cutoff is less than this value, the algorithm will calculate neighbor list using min_r as cutoff and discard those that have larger distances.

Returns:

Indexes of center_coords. index2 (n, ): Indexes of all_coords that form the neighbor pair. offset_vectors (n, 3): The periodic image offsets for all_coords. distances (n, ).

Return type:

index1 (n, )