v0.18.0

Breaking changes

• Posteriors saved under sbi v0.17.2 or older can not be loaded under sbi
v0.18.0 or newer.
• sample_with can no longer be passed to .sample(). Instead, the user has to rerun
  .build_posterior(sample_with=...). (#573)
• the posterior no longer has the the method .sample_conditional(). Using this
  feature now requires using the sampler interface (see tutorial
  here) (#573)
• retrain_from_scratch_each_round is now called retrain_from_scratch (#598, thanks to @jnsbck)
• API changes that had been introduced in sbi v0.14.0 and v0.15.0 are not enforced. Using the interface prior to
  those changes leads to an error (#645)
• prior passed to SNPE / SNLE / SNRE must be a PyTorch distribution (#655), see FAQ-7 for how to pass use custom prior.

Major changes and bug fixes

• new sampler interface (#573)
• posterior quality assurance with simulation-based calibration (SBC) (#501)
• added Sequential Neural Variational Inference (SNVI) (Glöckler et al. 2022) (#609, thanks to @manuelgloeckler)
• bugfix for SNPE-C with mixture density networks (#573)
• bugfix for sampling-importance resampling (SIR) as init_strategy for MCMC (#646)
• new density estimator for neural likelihood estimation with mixed data types (MNLE, #638)
• MCMC can now be parallelized across CPUs (#648)
• improved device check to remove several GPU issues (#610, thanks to @LouisRouillard)

Enhancements

• pairplot takes ax and fig (#557)
• bugfix for rejection sampling (#561)
• remove warninig when using multiple transforms with NSF in single dimension (#537)
• Sampling-importance-resampling (SIR) is now the default init_strategy for MCMC (#605)
• change mp_context to allow for multi-chain pyro samplers (#608, thanks to @sethaxen)
• tutorial on posterior predictive checks (#592, thanks to @LouisRouillard)
• add FAQ entry for using a custom prior (#595, thanks to @jnsbck)
• add methods to plot tensorboard data (#593, thanks to @lappalainenj)
• add option to pass the support for custom priors (#602)
• plotting method for 1D marginals (#600, thanks to @GuyMoss)
• fix GPU issues for conditional_pairplot and ActiveSubspace (#613)
• MCMC can be performed in unconstrained space also when using a MultipleIndependent distribution as prior (#619)
• added z-scoring option for structured data (#597, thanks to @rdgao)
• refactor c2st; change its default classifier to random forest (#503, thanks to @psteinb)
• MCMC init_strategy is now called proposal instead of prior (#602)
• inference objects can be serialized with pickle (#617)
• preconfigured fully connected embedding net (#644, thanks to @JuliaLinhart #624)
• posterior ensembles (#612, thanks to @jnsbck)
• remove gradients before returning the posterior (#631, thanks to @tomMoral)
• reduce batchsize of rejection sampling if few samples are left (#631, thanks to @tomMoral)
• tutorial for how to use SBC (#629, thanks to @psteinb)
• tutorial for how to use SBI with trial-based data and mixed data types (#638)
• allow to use a RestrictedPrior as prior for SNPE (#642)

v0.17.2

Minor changes

• bug fix for transforms in KDE (#552)

v0.17.1

Minor changes

• improve kwarg handling for rejection abc and smcabc
• typo and link fixes (#549, thanks to @pitmonticone)
• tutorial notebook on crafting summary statistics with sbi (#511, thanks to @ybernaerts)
• small fixes and improved documenentation for device handling (#544, thanks to @milagorecki)

v0.17.0

Major changes

• New API for specifying sampling methods (#487). Old syntax:

posterior = inference.build_posterior(sample_with_mcmc=True)

New syntax:

posterior = inference.build_posterior(sample_with="mcmc")  # or "rejection"

• Rejection sampling for likelihood(-ratio)-based posteriors (#487)
• MCMC in unconstrained and z-scored space (#510)
• Prior is now allowed to lie on GPU. The prior has to be on the same device as the one
  passed for training (#519).
• Rejection-ABC and SMC-ABC now return the accepted particles / parameters by default,
  or a KDE fit on those particles (kde=True) (#525).
• Fast analytical sampling, evaluation and conditioning for DirectPosterior trained
  with MDNs (thanks @jnsbck #458).

Minor changes

• scatter allowed for diagonal entries in pairplot (#510)
• Changes to default hyperparameters for SNPE_A (thanks @famura, #496, #497)
• bugfix for within_prior checks (#506)

v0.16.0

Major changes

• Implementation of SNPE-A (thanks @famura and @theogruner, #474, #478, #480, #482)
• Option to do inference over iid observations with SNLE and SNRE (#484, #488)

Minor changes

• Fixed unused argument num_bins when using nsf as density estimator (#465)
• Fixes to adapt to the new support handling in torch v1.8.0 (#469)
• More scalars for monitoring training progress (thanks @psteinb #471)
• Fixed bug in minimal.py (thanks @psteinb, #485)
• Depend on pyknos v0.14.2

v0.15.1

• add option to pass torch.data.DataLoader kwargs to all inference methods (thanks @narendramukherjee, #445)
• fix bug due to release of torch v1.8.0 (#451)
• expose leakage_correction parameters for log_prob correction in unnormalized
  posteriors (thanks @famura, #454)

v0.15.0

Major changes

• Active subspaces for sensitivity analysis (#394, tutorial)
• Method to compute the maximum-a-posteriori estimate from the posterior (#412)

API changes

• pairplot(), conditional_pairplot(), and conditional_corrcoeff() should now be imported from sbi.analysis instead of sbi.utils (#394).
• Changed fig_size to figsize in pairplot (#394).
• moved user_input_checks to sbi.utils (#430).

Minor changes

• Depend on new joblib=1.0.0 and fix progress bar updates for multiprocessing (#421).
• Fix for embedding nets with SNRE (thanks @adittmann, #425).
• Is it now optional to pass a prior distribution when using SNPE (#426).
• Support loading of posteriors saved after sbi v0.15.0 (#427, thanks @psteinb).
• Neural network training can be resumed (#431).
• Allow using NSF to estimate 1D distributions (#438).
• Fix type checks in input checks (thanks @psteinb, #439).
• Bugfix for GPU training with SNRE_A (thanks @glouppe, #442).
• Fixup for conditional correlation matrix (thanks @JBeckUniTb, #404)
• z-score data using only the training data (#411)

v0.14.2

• Small fix for SMC-ABC with semi-automatic summary statistics (#402)

v0.14.1

• Support for training and sampling on GPU including fixes from nflows (#331)
• Bug fix for SNPE with neural spline flow and MCMC (#398)
• Small fix for SMCABC particles covariance
• Small fix for rejection-classifier (#396)

v0.14.0

• New flexible interface API (#378). This is going to be a breaking change for users of
  the flexible interface and you will have to change your code. Old syntax:

from sbi.inference import SNPE, prepare_for_sbi

simulator, prior = prepare_for_sbi(simulator, prior)
inference = SNPE(simulator, prior)

# Simulate, train, and build posterior.
posterior = inference(num_simulation=1000)

New syntax:

from sbi.inference import SNPE, prepare_for_sbi, simulate_for_sbi

simulator, prior = prepare_for_sbi(simulator, prior)
inference = SNPE(prior)

theta, x = simulate_for_sbi(simulator, proposal=prior, num_simulations=1000)
density_estimator = inference.append_simulations(theta, x).train()
posterior = inference.build_posterior(density_estimator)  # MCMC kwargs go here.

More information can be found here here.

• Fixed typo in docs for infer (thanks @glouppe, #370)
• New RestrictionEstimator to learn regions of bad simulation outputs (#390)
• Improvements for and new ABC methods (#395)
  • Linear regression adjustment as in Beaumont et al. 2002 for both MCABC and SMCABC
  • Semi-automatic summary statistics as in Fearnhead & Prangle 2012 for both MCABC and SMCABC
  • Small fixes to perturbation kernel covariance estimation in SMCABC.