User Guide

These notebooks provide in-depth coverage of specific ProbPipe features. For a guided introduction, start with the Getting Started tutorial.

The chapters below match the 10-notebook target tracked in issue #127.

#	Notebook	Description
1	Distribution basics	The seven distribution ops (sample, log_prob, mean, variance, cov, expectation, condition_on); univariate, multivariate, and non-parametric (Empirical / Bootstrap / BootstrapReplicate) families; the SupportsX protocol family; from_distribution for converting between representations.
2	Records and Record Distributions	The 2×2 of structured containers: Record / NumericRecord (non-random values) paired with RecordDistribution / NumericRecordDistribution (random named-component distributions), plus the RecordArray and DistributionArray "array of" forms.
3	Broadcasting and workflow functions	Automatic uncertainty propagation, empirical enumeration, cartesian products, vectorization backends, and seeded reproducibility.
4	Joint distributions	ProductDistribution, SequentialJointDistribution, JointGaussian, JointEmpirical / NumericJointEmpirical; component views; condition_on; flat-vector interop.
5	External backends	How condition_on dispatches to TFP NUTS, Stan, PyMC, nutpie, and sbijax; pinning a specific method; the inference method registry.
6	Converting between representations	Bijectors + TransformedDistribution, from_distribution moment matching, and the converter registry for satisfying protocols like SupportsLogProb.
7	Sequential updating	Batch-wise Bayesian updating with IncrementalConditioner, auto KDE conversion, and provenance chain.
8	JAX interop	End-to-end JAX gradients through distributions: score functions, sensitivity analysis, maximum likelihood estimation, and variational inference.
9	Bagged posteriors	BootstrapReplicateDistribution, broadcasting condition_on over resampled datasets, and the between- vs. within-replicate spread as a stability / misspecification diagnostic.
10	Random functions and Gaussian emulators	RandomFunction / GaussianRandomFunction / LinearBasisFunction; joint-input / joint-output modes; algebraic operations; fitting to data; GP emulators and synthetic-likelihood surrogates for simulation-based inference.