geossm.stmodel.FitOptions¶

class geossm.stmodel.FitOptions(max_iter=20, tol_relat=0.001, verbose=True, force_iter=False, std=True)[source]¶

Bases: object

Stopping-rule and logging options for fit().

Parameters:

max_iter (int, default 20) – Maximum number of EM iterations to run.
tol_relat (float, default 1e-3) – Relative log-likelihood-improvement tolerance below which the EM iteration is considered converged and stops early (before max_iter is reached).
verbose (bool, default True) – If True, log per-iteration progress during fit.
force_iter (bool, default False) – If False, fit enforces a monotonically increasing log-likelihood: it stops at the first decrease and returns the last valid estimate. If True, a decrease only triggers a warning and the EM iterations continue (useful when poor initial values make the log-likelihood oscillate at the start). A NaN/Inf log-likelihood always stops the algorithm and returns the last valid estimate.
std (bool, default True) – If False, skip the in-sample predictive covariance Sigma_y_hat computed after the EM loop (left as None) – useful when the (p, p, T) array does not fit in device memory.

Attributes