mbi package

Top-Level Classes

`Domain`	Represents a discrete domain defined by attributes and their sizes.
`Dataset`	A discrete tabular dataset backed by a dictionary of 1D numpy arrays.
`Factor`	Represents a factor defined over a discrete domain.
`CliqueVector`	Manages a collection of factors, each associated with a clique.
`Projectable`	An object whose marginals can be computed over subsets of attributes.
`MarginalLossFn`	A loss function over the concatenated vector of marginals.
`LinearMeasurement`	A class for representing a private linear measurement of a marginal.
`MarginalOracle`	Callable signature for stateless marginal oracle functions.
`MarkovRandomField`	Represents a learned graphical model.
`Estimator`	An object that estimates a Model from a marginal-based loss function.
`Constraint`	A structural constraint on allowed value combinations.

`save`	Save a JAX pytree to `.npz` format.
`load`	Load a JAX pytree from `.npz` format.
`summarize`	Return a structured summary of the model.
`set_log_fn`	Override the library-wide log function (default: print).

`marginal_oracles`	Marginal oracles for computing marginals from graphical model potentials.
`marginal_loss`	Defines loss functions based on linear measurements of marginals.
`estimation`	Algorithms for estimating graphical models from marginal-based loss functions.
`approximate_oracles`	Approximate marginal oracles with convex counting numbers.
`callbacks`	Defines callback mechanisms for monitoring optimization processes.
`junction_tree`	Utilities for constructing and working with junction trees.
`extensions`	Extensions for mbi providing alternative estimation approaches.