NonlinearSSM

cultivars.state_space.nonlinear.NonlinearSSM

alias of _NonlinearStateSpace

Private methods

_draw_states(particles, rng, y_prev)	One transition draw per particle.
_gaussian_update(y_row, predicted_obs, ...)	One Gaussian measurement update from moment inputs.
_jacobian(func, point, out_dim)	Central-difference Jacobian of a batched map at one point.
_lag_stack_state_space(mean_map, *, order, ...)	Wrap a nonlinear autoregressive mean as a noisily observed state space.
_measure(y_row, particles)	Per-particle measurement log-density.
_predict(particles, y_prev)	The state mean map, conditioned on the previous row when the model says so.
_prepare(y)	Coerce data to (n, p), promoting a 1-D series when p is one.
_refuse_hooks(filter_name)	Refuse a linearizing filter on a model outside the additive form.
_rts_backward(outcome, cross_at, method)	Rauch's backward recursion from a forward pass and a gain rule.
_sigma_points(mean, cov, scale)	The 2m + 1 unscented points around one moment pair.
_smoother_gain(cross, predicted_cov)	The Rauch gain C P_{t+1|t}^{-1} from a cross-covariance.
_systematic_resample(weights, rng)	Systematic resampling indices for normalized weights.
_unscented_weights(alpha, beta, kappa)	Scaling constant and mean/covariance sigma-point weights.