gaussian.evaluation

Reusable observation ensembles, candidate measures and model inference datasets.

Functions

unit_sobol(count, dimension, seed)	Use a base-two draw then truncate; non-power-of-two counts are explicitly allowed.
cartesian(axes)	Caller checks the product size before allocating.
truth_settings(context)
candidate_settings(context)
inference_settings(context)
sobol_truth_count(problem, settings)	sobol_truths at one inferred parameter, doubling with each additional parameter.
truth_layout(problem, settings)	Return the actual truth design and count; size_adaptive switches to Sobol above max_grid_truths.
inference_observations(context, dependencies, writer)
candidate_layout(problem, prior, settings)
candidates(context, dependencies, writer)
inference(context, dependencies, writer)

Module Contents

gaussian.evaluation.unit_sobol(count, dimension, seed)

Use a base-two draw then truncate; non-power-of-two counts are explicitly allowed.

gaussian.evaluation.cartesian(axes)

Caller checks the product size before allocating.

gaussian.evaluation.truth_settings(context)

gaussian.evaluation.candidate_settings(context)

gaussian.evaluation.inference_settings(context)

gaussian.evaluation.sobol_truth_count(problem, settings)

sobol_truths at one inferred parameter, doubling with each additional parameter.

gaussian.evaluation.truth_layout(problem, settings)

Return the actual truth design and count; size_adaptive switches to Sobol above max_grid_truths.

gaussian.evaluation.inference_observations(context, dependencies, writer)

gaussian.evaluation.candidate_layout(problem, prior, settings)

gaussian.evaluation.candidates(context, dependencies, writer)

gaussian.evaluation.inference(context, dependencies, writer)