explore (pmrf.explore)

(experimental) model exploration via random sampling.

Parameter space exploration and active learning module.

This module provides engines for non-adaptive (One-Shot) and adaptive (Active Learning) sampling of the model’s parameter space.

Exploration is performed by instantiating an engine (e.g., LatinHypercube or EqxLearnUncertainty) and passing it to the unified sample function along with the model. The router automatically delegates to the correct execution loop, evaluating the physical parameters and extracted features.

Results are returned as an ExploreResult, which contains the batched models and raw arrays for downstream plotting or surrogate training.

Classes

ExploreResult(model, frequency, ...[, ...])	Container for the results of a parameter space exploration process.
AbstractSampler()	Abstract base class for all exploration engines, mirroring the Optimistix AbstractIterativeSolver paradigm.
AbstractOneShotSampler()	One-shot samplers generate all points immediately during init.
AbstractAdaptiveSampler([initial_models, ...])	Adaptive samplers iterate until a budget is reached or convergence is met.
LatinHypercube()	Sampler using Latin Hypercube Sampling (LHS).
Uniform()	Sampler using uniform random sampling.
FieldSampler([initial_models, batch_size, ...])	Samples new points at the maxima of a learned scalar field.
EqxLearnUncertainty([initial_models, ...])	Adaptive sampler targeting regions of high surrogate uncertainty.

Functions

sample(model, sampler, *[, max_samples, ...])

Explore the parameter space of a model using a specified sampling engine.