populations

anri.index.populations(f, cand, U_list, ops, radius_deg, p=4, m=16, eps=0.02, vb=16384)

Group each voxel’s occupied candidate orientations into up to p populations.

A voxel’s m most occupied candidates above eps x its total are grouped greedily: the most occupied free candidate seeds a population, and every free candidate within radius_deg of it (over the Laue group) joins. Each population gets its share of the voxel’s occupancy, the occupancy-weighted mean of its members (aligned to the seed over the symmetry, then projected back to a rotation) and their weighted rms misorientation from that mean. Populations closer than the radius come out as one, with a larger spread: what the data cannot separate is reported as a mean and a spread. The spread includes the orientation grid’s own spacing (one orientation between grid points shares itself among the nearby points), so it is an upper bound.

Parameters:

• f (Array | ndarray | bool | number | bool | int | float | complex) – [Nv, K] occupancies and orientation indices, from anri.index.fit_occupancy()

• cand (Array | ndarray | bool | number | bool | int | float | complex) – [Nv, K] occupancies and orientation indices, from anri.index.fit_occupancy()

• U_list (Array | ndarray | bool | number | bool | int | float | complex) – [Nq, 3, 3] the orientations that cand indexes

• ops (Array | ndarray | bool | number | bool | int | float | complex) – [n, 3, 3] Laue-group rotations, from anri.crystal.laue_rotations()

• radius_deg (float) – Largest misorientation from a population’s seed, e.g. 1.8 grid steps

• p (int, default: 4) – Populations per voxel at most

• m (int, default: 16) – Candidates considered per voxel

• eps (float, default: 0.02) – Smallest occupancy considered, as a fraction of the voxel’s total

• vb (int, default: 16384) – Voxels per call

Returns:

• fraction (np.ndarray) – [Nv, p] share of each voxel’s occupancy (0 where absent), largest first

• U (np.ndarray) – [Nv, p, 3, 3] mean orientations

• spread (np.ndarray) – [Nv, p] rms misorientation from the mean, degrees

• n (np.ndarray) – [Nv, p] candidates in each population