class probnum.filtsmooth.optim.GaussNewton(kalman, stopping_criterion=None)

Bases: probnum.filtsmooth.optim.StateSpaceOptimizer

Gauss-Newton optimizer in state-space models.

Equivalent to the iterated Kalman smoother.


This class of algorithms solves regression problems. As such, it computes a posterior distribution over states. It has not much to do with hyperparameter optimization.

Methods Summary

solution_generator(regression_problem, ...)

Generate optimization steps.

solve(regression_problem, initial_guess)

Solve a state-space optimization problem.

Methods Documentation

solution_generator(regression_problem, initial_guess)[source]

Generate optimization steps.

solve(regression_problem, initial_guess)

Solve a state-space optimization problem.