cellflow.utils.match_linear

cellflow.utils.match_linear(source_batch, target_batch, cost_fn=<ott.geometry.costs.SqEuclidean object>, epsilon=1.0, scale_cost='mean', tau_a=1.0, tau_b=1.0, threshold=None, **kwargs)[source]

Compute solution to a linear OT problem.

Parameters:

  • source_batch (Array) – Source point cloud of shape [n, d].

  • target_batch (Array) – Target point cloud of shape [m, d].

  • cost_fn (CostFn | None) – Cost function to use for the linear OT problem.

  • epsilon (float | None) – Regularization parameter.

  • scale_cost (Union[float, Literal['mean', 'max_cost', 'median']]) – Scaling of the cost matrix.

  • tau_a (float) – Parameter in \((0, 1]\) that defines how unbalanced the problem is in the source distribution. If \(1\), the problem is balanced in the source distribution.

  • tau_b (float) – Parameter in \((0, 1]\) that defines how unbalanced the problem is in the target distribution. If \(1\), the problem is balanced in the target distribution.

  • threshold (float | None) – Convergence criterion for the Sinkhorn algorithm.

  • kwargs (Any) – Additional arguments for ott.solvers.linear.sinkhorn.Sinkhorn.

Return type:

Array

Returns:

: Optimal transport matrix between 'source_batch' and 'target_batch'.