pypomp.core.pomp.Pomp.dpop_train

Pomp.dpop_train(J: int, M: int, eta: LearningRate, optimizer: str = 'Adam', alpha: float = 0.8, decay: float = 0.0, process_weight_state: str | None = None, key: Array | None = None, theta: Mapping[str, int | float | number | Array] | Sequence[Mapping[str, int | float | number | Array]] | pypomp.core.parameters.PompParameters | None = None) → tuple[Array, Array][source]

Optimizes model parameters using the DPOP differentiable particle filter and gradient-based methods.

This method trains the model parameters to maximize the DPOP objective function using first-order optimizers like Adam or SGD, with optional learning rate decay. Gradients are computed efficiently via JAX reverse-mode automatic differentiation.

Parameters:

  • J (int) – Number of particles.

  • M (int) – Number of gradient steps.

  • eta (LearningRate) – Learning rates per parameter as a LearningRate object.

  • optimizer (str, default "Adam") – Optimizer to use: “Adam” or “SGD”.

  • alpha (float, default 0.8) – DPOP discount / cooling factor.

  • decay (float, default 0.0) – Learning-rate decay coefficient. At iteration m, the effective learning rate is eta / (1 + decay * m).

  • process_weight_state (str or None, default None) – Name of the state component that stores the accumulated process log-weight (e.g. "logw").

  • key (jax.Array or None, default None) – Random key. If None, uses self.fresh_key.

  • theta (ThetaInput, default None) – Optional initial parameter(s). Accepts dict[str, Numeric], list[dict[str, Numeric]], or PompParameters. Numeric values are coerced to floats. Defaults to self.theta.

Returns:

  • nll_history (jax.Array, shape (M+1,)) – Mean DPOP negative log-likelihood per observation at each step.

  • theta_history (jax.Array, shape (M+1, p)) – Parameter vector (estimation space) at each step.