ott.solvers.linear.semidiscrete.SemidiscreteState

ott.solvers.linear.semidiscrete.SemidiscreteState#

class ott.solvers.linear.semidiscrete.SemidiscreteState(it, epsilon, g, g_ema, opt_state, losses, grad_norms, errors)[source]#

State of the SemidiscreteSolver.

Parameters:
  • it (Array) – Iteration number.

  • epsilon (Array) – Epsilon value at the current iteration.

  • g (Array) – Dual potential.

  • g_ema (Array) – Exponential moving average of the dual potential.

  • opt_state (Any) – State of the optimizer.

  • losses (Array) – Dual losses.

  • grad_norms (Array) – Norms of the gradients.

  • errors (Array) – Marginal deviation errors.

Methods

Attributes