scarlet2.infer

Inference methods

Classes

FitValidator(scene, observation)	A class containing all of the validation checks for a Scene objects after calling .fit().
Measurement(name, mean, std)	The mean and standard deviation of a measured quantity for one source.
PairSimilarity(weight[, eps])	Configuration for the pair-similarity regularizer used by fit().

Functions

fit(scene, observations, *args[, schedule, ...])	Fit model parameters of every source in scene to match observations.
get_scene_samples(scene, uncertainties[, ...])	Draw posterior sample scenes from a Laplace approximation.
hvp(f, primals, tangents)	Calculate the Hessian-vector product of a function f
hvp_grad(grad_f, primals, tangents)	Calculate the Hessian-vector product of a gradient function grad_f
hvp_rad(hvp, shape[, key, max_iter])	Approximate the diagonal of the Hessian
sample(scene, observations, *args[, seed, ...])	Sample parameters of every source in scene to get posteriors given observations.
source_uncertainty(scene, uncertainties, ...)	Propagate parameter uncertainties to a per-source measurement.
uncertainty(scene, observations[, ...])	Estimate the per-parameter curvature (diagonal Hessian) of a fitted model.