pypomp.PanelPomp.mif

PanelPomp.mif(J: int, M: int, rw_sd: RWSigma, key: Array | None = None, theta: PanelParameters | None = None, thresh: float = 0.0, n_monitors: int = 0, block: bool = True) → None

Estimate parameters using the (Marginalized) Panel Iterated Filtering (MPIF/PIF) algorithm.

Performs parameter estimation using the (Marginalized) Panel Iterated Filtering (MPIF/PIF) algorithm (Bretó et al. 2020 [1]; Wheeler et al. 2025 [2]).

Parameters:

• J (int) – Number of particles per unit.

• M (int) – Number of iterations (cooling cycles).

• rw_sd (RWSigma) – Random walk standard deviations and cooling schedule.

• key (jax.Array or None, optional) – JAX random key. If None, uses the model’s fresh_key.

• theta (PanelParameters or None, optional) – Initial parameter estimates. If None, defaults to self.theta.

• thresh (float, optional) – Resampling threshold for the particle filter. Defaults to 0.0.

• n_monitors (int, optional) – Number of unperturbed particle filter runs to estimate log-likelihood at each iteration. Defaults to 0 (use perturbed filter).

• block (bool, optional) – Whether to use block updates (MPIF). If False, uses standard PIF. Defaults to True.

Returns:

Updates self.theta with final estimates and appends a Result to the history.

Return type:

None

References

[1]

Bretó, Carles, Edward L. Ionides, and Aaron A. King. “Panel Data Analysis via Mechanistic Models.” Journal of the American Statistical Association 115, no. 531 (2020): 1178–1188. https://doi.org/10.1080/01621459.2019.1604367.

[2]

Wheeler, Jesse, Aaron J. Abkemeier, and Edward L. Ionides. “Iterating marginalized Bayes maps for likelihood maximization with application to nonlinear panel models.” arXiv preprint arXiv:2511.17438 (2025). https://arxiv.org/abs/2511.17438.