Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the user guide for the big picture.

3.3.5. fmralign.methods.OptimalTransport

class fmralign.methods.OptimalTransport(solver='sinkhorn_epsilon_scaling', metric='euclidean', reg=0.01, max_iter=1000, tol=0.001)[source]

Compute the optimal coupling between X and Y with entropic regularization, using the pure Python POT (https://pythonot.github.io/) package.

Parameters:
solverstr (optional)

solver from POT called to find optimal coupling ‘sinkhorn’, ‘greenkhorn’, ‘sinkhorn_stabilized’,’sinkhorn_epsilon_scaling’, ‘exact’ see POT/ot/bregman on Github for source code of solvers

metricstr (optional)

metric used to create transport cost matrix, see full list in scipy.spatial.distance.cdist doc

regint (optional)

level of entropic regularization

Attributes:
Rscipy.sparse.csr_matrix

Mixing matrix containing the optimal permutation

__init__(solver='sinkhorn_epsilon_scaling', metric='euclidean', reg=0.01, max_iter=1000, tol=0.001)[source]

3.3.5.1. Examples using fmralign.methods.OptimalTransport

Alignment methods benchmark (template-based ROI case)

Alignment methods benchmark (template-based ROI case)

Alignment on simulated 2D data.

Alignment on simulated 2D data.