estimate_cutoff#

datafold.pcfold.estimate_cutoff(pcm, n_subsample=1000, k=10, random_state=None, distance_matrix=None)[source]#

Estimates a good choice of cut-off for a Gaussian radial basis kernel, given a certain tolerance below which the kernel values are considered zero.

Parameters:
  • pcm – point cloud to compute pair-wise kernel matrix with

  • n_subsample (int) – Maximum subsample used for the estimation. Ignored if distance_matrix is not None.

  • k (int) – Compute the k-th nearest neighbor distance to estimate the cut-off distance.

  • random_state (Optional[int]) – sets np.random.default_rng(random_state)

  • distance_matrix – pre-computed distance matrix instead of using the internal cdist method

Return type:

float