API reference

Prior and simulator

sbi.utils.BoxUniform	Uniform distribution in multiple dimensions.
sbi.utils.MultipleIndependent	Wrap a sequence of PyTorch distributions into a joint PyTorch distribution.
sbi.utils.process_prior	Return PyTorch distribution-like prior from user-provided prior.
sbi.utils.process_simulator	Returns a simulator that meets the requirements for usage in sbi.
sbi.utils.RestrictedPrior	Distribution that restricts the prior distribution to a smaller region.
sbi.utils.RestrictionEstimator	Classifier to estimate regions of the prior that give good simulation results.
sbi.inference.simulate_for_sbi	Returns pairs \((\theta, x)\) by sampling proposal and running simulations.

Neural nets

sbi.neural_nets.classifier_nn	Returns a function that builds a classifier for learning density ratios.
sbi.neural_nets.likelihood_nn	Returns a function that builds a density estimator for learning the likelihood.
sbi.neural_nets.marginal_nn	Returns a function that builds a density estimator for learning the marginal.
sbi.neural_nets.posterior_flow_nn	Build util function that builds a FlowMatchingEstimator object for flow-based posteriors.
sbi.neural_nets.posterior_nn	Returns a function that builds a density estimator for learning the posterior.
sbi.neural_nets.posterior_score_nn	Build util function that builds a ScoreEstimator object for score-based posteriors.

Embedding nets

sbi.neural_nets.embedding_nets.CausalCNNEmbedding	Embedding network that uses 1D causal convolutions.
sbi.neural_nets.embedding_nets.CNNEmbedding	Convolutional embedding network (1D or 2D convolutions).
sbi.neural_nets.embedding_nets.FCEmbedding	Fully-connected multi-layer neural network to be used as embedding network.
sbi.neural_nets.embedding_nets.LRUEmbedding	Embedding network backed by a stack of Linear Recurrent Unit (LRU) blocks.
sbi.neural_nets.embedding_nets.PermutationInvariantEmbedding	Permutation invariant embedding network.
sbi.neural_nets.embedding_nets.ResNetEmbedding1D	Residual neural network mapping vector-like data to a fixed-size vector.
sbi.neural_nets.embedding_nets.ResNetEmbedding2D	Residual neural network mapping image-like data to a fixed sized vector.
sbi.neural_nets.embedding_nets.SpectralConvEmbedding
sbi.neural_nets.embedding_nets.TransformerEmbedding	Transformer-based embedding network for time series and image data.

Training

sbi.inference.BNRE	Balanced Neural Ratio Estimation (BNRE) as in Delaunoy et al. (2022) [1].
sbi.inference.FMPE	Flow Matching Posterior Estimation (FMPE) [1].
sbi.inference.MarginalTrainer	Utility class for training a marginal density estimator $p(x)$.
sbi.inference.MCABC	Monte-Carlo Approximate Bayesian Computation (Rejection ABC).
sbi.inference.MNLE	Mixed Neural Likelihood Estimation for discrete and continuous data [1].
sbi.inference.MNPE	Method that can infer discrete and continuous parameters (Mixed NPE).
sbi.inference.NLE_A	Neural Likelihood Estimation (NLE) as in Papamakarios et al. (2019) [1].
sbi.inference.NPE_A	Neural Posterior Estimation algorithm as in Papamakarios et al. (2016) [1].
sbi.inference.NPE_B	Neural Posterior Estimation algorithm (NPE-B) as in Lueckmann et al. (2017) [1].
sbi.inference.NPE_C	Neural Posterior Estimation algorithm (NPE-C) as in Greenberg et al. (2019) [1].
sbi.inference.NPSE	Neural Posterior Score Estimation (NPSE) [1, 2].
sbi.inference.NRE_A	Neural Ratio Estimation (NRE-A / AALR) as in Hermans et al. (2020) [1].
sbi.inference.NRE_B	Neural Ratio Estimation (NRE-B / SRE) as in Durkan et al. (2020) [1].
sbi.inference.NRE_C	Neural Ratio Estimation (NRE-C / CNRE) as in Miller et al. (2022) [1].
sbi.inference.SMCABC	Sequential Monte Carlo Approximate Bayesian Computation.

Potentials

sbi.inference.likelihood_estimator_based_potential	Returns potential \(\log(p(x_o|\theta)p(\theta))\) for likelihood estimator.
sbi.inference.mixed_likelihood_estimator_based_potential	Returns $log(p(x_o|theta)p(theta))$ for mixed likelihood-based methods.
sbi.inference.posterior_estimator_based_potential	Returns the potential for posterior-based methods.
sbi.inference.ratio_estimator_based_potential	Returns the potential for ratio-based methods.
sbi.inference.vector_field_estimator_based_potential	Returns the potential function gradient for vector field estimators.

Posteriors

sbi.inference.DirectPosterior	Posterior based on neural networks that directly estimate the posterior (NPE).
sbi.inference.EnsemblePosterior	Wrapper for bundling together different posterior instances into an ensemble.
sbi.inference.ImportanceSamplingPosterior	Provides importance sampling to sample from the posterior.
sbi.inference.MCMCPosterior	Provides MCMC to sample from the posterior.
sbi.inference.RejectionPosterior	Provides rejection sampling to sample from the posterior.
sbi.inference.VectorFieldPosterior	Posterior based on flow- or score-matching estimators.
sbi.inference.VIPosterior	Provides VI (Variational Inference) to sample from the posterior.

Posterior Parameters

sbi.inference.posteriors.DirectPosteriorParameters	Parameters for initializing DirectPosterior.
sbi.inference.posteriors.ImportanceSamplingPosteriorParameters	Parameters for initializing ImportanceSamplingPosterior.
sbi.inference.posteriors.MCMCPosteriorParameters	Parameters for initializing MCMCPosterior.
sbi.inference.posteriors.RejectionPosteriorParameters	Parameters for initializing RejectionPosterior.
sbi.inference.posteriors.VectorFieldPosteriorParameters	Parameters for initializing VectorFieldPosterior.
sbi.inference.posteriors.VIPosteriorParameters	Parameters for VIPosterior, supporting both single-x and amortized VI.

Diagnostics

sbi.diagnostics.calc_misspecification_logprob	Perform hypothesis test to check if estimator.log_prob(x_o) is unusually low.
sbi.diagnostics.calc_misspecification_mmd	Misspecification test based on MMD in data- or embedding space.
sbi.diagnostics.check_sbc	Return uniformity checks and data-averaged posterior checks for SBC.
sbi.diagnostics.check_tarp	check the obtained TARP credibitlity levels and expected coverage probabilities.
sbi.diagnostics.get_nltp	Return negative log prob of true parameters under the posterior.
sbi.diagnostics.LC2ST	L-C2ST: Local Classifier Two-Sample Test.
sbi.diagnostics.run_sbc	Run simulation-based calibration (SBC) or expected coverage.
sbi.diagnostics.run_tarp	Estimates coverage of samples given true values thetas with the TARP method.

Analysis

sbi.analysis.ActiveSubspace	Identify the active subspace of the posterior for sensitivity analyses.
sbi.analysis.conditional_corrcoeff	Returns the conditional correlation matrix of a distribution.
sbi.analysis.conditional_pairplot	Plot conditional distribution given all other parameters.
sbi.analysis.conditional_potential	Returns potential function that can be used to sample the conditional potential.
sbi.analysis.marginal_plot	Plot samples in a row showing 1D marginals of selected dimensions.
sbi.analysis.pairplot	Plot samples in a 2D grid showing marginals and pairwise marginals.
sbi.analysis.plot_tarp	Plots the expected coverage probability (ECP) against the credibility level,alpha, for a given alpha grid.
sbi.analysis.pp_plot	Probability - Probability (P-P) plot for hypothesis tests to assess the validity of one (or several) estimator(s).
sbi.analysis.pp_plot_lc2st	Probability - Probability (P-P) plot for LC2ST.
sbi.analysis.sbc_rank_plot	Plot simulation-based calibration ranks as empirical CDFs or histograms.

Utilities

sbi.utils.get_density_thresholder	Returns function that thresholds a density at a particular 1-quantile.
sbi.utils.mcmc_transform	Builds a transform that is applied to parameters during MCMC.
sbi.utils.transformed_potential	Return potential after a transformation by adding the log-abs-determinant.