# API reference

Every public symbol lives in the flat `setu` namespace: import it as
`from setu import NLE` regardless of which submodule defines it. The tables
below group the surface by purpose for readability; each symbol is documented
once, and the linked page carries the full signature, parameters, and return
type.

```{eval-rst}
.. currentmodule:: setu
```

## Neural Likelihood Estimation (NLE)

Learn the conditional density $p(x \mid \theta)$ with a normalizing flow.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   NLE
   TrainingResult
   create_nle
   fit_nle
```

## Mixed Neural Likelihood Estimation (MixedNLE)

NLE for mixed discrete + continuous data (a flow for the continuous part, a
CategoricalMADE for the discrete part).

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   MixedNLE
   MixedNLETrainingResult
   create_mixed_nle
   fit_mixed_nle
```

## Discrete Neural Likelihood Estimation (DiscreteNLE)

NLE for pure-discrete observations: a categorical head for bounded counts, or a
CountMADE (ZINB / NegBinom / Poisson) for count data.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   DiscreteNLE
   DiscreteNLETrainingResult
   create_discrete_nle
   fit_discrete_nle
```

## Neural Ratio Estimation (NRE)

Learn a likelihood-to-evidence ratio with a classifier instead of a normalized
density.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   NRE
   NRETrainingResult
   create_nre
   fit_nre
```

## Data

Simulation datasets and shape helpers. See {doc}`shapes` for the
`(G, S, T, *E)` conventions these follow.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   SimulationDataset
   ValidationDataset
   simulate_dataset
   one_hot
   broadcast_conditions
```

## Serialization

Save and load trained estimators (Equinox leaves plus a JSON metadata sidecar).

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   save_estimator
   load_estimator
```

## Transforms

Z-scoring transforms applied to parameters and observations.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   Standardizer
```

## Metrics

Sample-based distances for comparing estimator draws against ground truth.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   c2st
   mmd
   sliced_wasserstein
   total_variation
```

## Simulators

Reference simulators and their analytical likelihoods, used in tests and
tutorials.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   linear_gaussian
   sample_prior
   analytical_log_likelihood
   analytical_log_ratio
   analytical_marginal_log_prob
```

## PyMC integration

Export a trained estimator as a PyMC likelihood term. See {doc}`shapes` for the
`(theta, x_observed)` argument order.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   to_pymc
   to_pymc_hierarchical
   mixed_nle_to_pymc
   discrete_nle_to_pymc
   nre_to_pymc
   nre_to_pymc_hierarchical
```

## NumPyro integration

Export a trained estimator as a NumPyro factor.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   to_numpyro
   to_numpyro_hierarchical
   discrete_nle_to_numpyro
   nre_to_numpyro
   nre_to_numpyro_hierarchical
```

## Leave-one-out cross-validation

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   compute_log_likelihood
```

## Tuning

Optuna-based hyperparameter search. Requires the `tuning` extra
(`pip install setu-sbi[tuning]`).

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   tune_nle
   tune_nre
   tune_mixed_nle
   TuningResult
   NLESearchSpace
   NRESearchSpace
   MixedNLESearchSpace
```

## Predictive checks

Prior and posterior predictive sampling.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   prior_predictive
   posterior_predictive
   PriorPredictiveResult
```

## Validation

Diagnostics that confirm a learned likelihood is trustworthy before export.

```{eval-rst}
.. autosummary::
   :toctree: _autosummary
   :nosignatures:

   validate_nle
   validate_estimator
   ValidationConfig
   ValidationResult
   ValidationError
```