Ctrl+K

ott.neural.methods.flows.otfm.OTFlowMatching.transport

ott.neural.methods.flows.otfm.OTFlowMatching.transport#

OTFlowMatching.transport(x, condition=None, t0=0.0, t1=1.0, **kwargs)[source]#

Transport data with the learned map.

This method pushes-forward the data by solving the neural ODE parameterized by the velocity field.

Parameters:

  • x (Array) – Initial condition of the ODE of shape [batch_size, ...].

  • condition (Optional[Array]) – Condition of the input data of shape [batch_size, ...].

  • t0 (float) – Starting point of integration.

  • t1 (float) – End point of integration.

  • kwargs (Any) – Keyword arguments for the ODE solver.

Return type:

Array

Returns:

The push-forward or pull-back distribution defined by the learned transport plan.

Contents