Optimizers

pypomp provides a variety of class-based optimization algorithms that can be passed to model training methods to customize parameter estimation.

Classes

SGD([clip_norm, scale, ls, c, max_ls_itn])	Stochastic Gradient Descent optimizer.
Adam([clip_norm, scale, ls, c, max_ls_itn, ...])	Adam optimizer.
FullMatrixAdam([clip_norm, scale, ls, c, ...])	Full-Matrix Adam optimizer.
BFGS([clip_norm, scale, ls, c, max_ls_itn])	Quasi-Newton BFGS optimizer.
Newton([clip_norm, scale, ls, c, max_ls_itn])	Classic Second-Order Newton-Raphson optimizer.
WeightedNewton([clip_norm, scale, ls, c, ...])	Weighted Newton optimizer with decaying history.