ott.tools package
Contents
ott.tools package#
The tools package contains high level functions that build on outputs produced by core functions. They can be used to compute Sinkhorn divergences [Séjourné et al., 2019], instantiate transport matrices, provide differentiable approximations to ranks and quantile functions [Cuturi et al., 2019], etc.
Segmented Sinkhorn#
|
Compute reg_ot_cost between subsets of vectors described in x & y. |
Sinkhorn Divergence#
Compute Sinkhorn divergence defined by a geometry, weights, parameters. |
|
Compute sinkhorn divergence between subsets of vectors given in x & y. |
Soft Sorting Algorithms#
|
Apply the soft quantile operator on the input tensor. |
|
Renormalize inputs so that its quantiles match those of targets/weights. |
|
Soft quantizes an input according using num_levels values along axis. |
|
Apply the soft trank operator on input tensor. |
|
Apply the soft sort operator on a given axis of the input. |
|
Sort a multidimensional array according to a real valued criterion. |
Clustering#
|
K-means clustering using Lloyd's algorithm [Lloyd, 1982]. |
|
Output of the |
ott.tools.gaussian_mixture package#
Gaussian Mixtures#
|
PyTree for a normal distribution. |
|
Pytree for a Gaussian Mixture model. |
Pytree for a coupled pair of Gaussian mixture models. |
|
|
Initialize a GMM via K-means++ with retries on failure. |
|
Fit a GMM using the EM algorithm. |
|
Get a function that performs penalized EM. |