setu.prior_predictive#

setu.prior_predictive(simulator_fn, prior_fn, conditions, n_sims_per_condition, key)[source]#

Run prior predictive simulations organized by condition group.

Draws n_sims_per_condition theta samples from the prior for each condition, runs the simulator, and returns a SimulationDataset per group.

Parameters:

  • simulator_fn (Callable) – (theta, key, conditions) -> x. Must accept three positional arguments (use simulate_dataset for simulators without conditions).

  • prior_fn (Callable) – (key) -> theta.

  • conditions (Array) – Unique condition vectors, shape (n_conditions, condition_dim).

  • n_sims_per_condition (int) – Simulations per condition group.

  • key (Array) – JAX random key.

Return type:

PriorPredictiveResult

Returns:

PriorPredictiveResult with one SimulationDataset per condition.