What's Changed

• Tests/reduce disk space by @bertiqwerty in #672
• 631 doeutils model matrix by @dlinzner-bcs in #673
• Docs quarto by @bertiqwerty in #667
• Feature/isolate repair by @LukasHebing in #668
• Implement new cache_root behavior from botorch by @jduerholt in #682
• 652 decorrelated mol features by @KislayaRavi in #683
• Feature/aggregations2 by @jduerholt in #678
• 675 fim matrix rank etc (replacing other pull request) by @rosonaeldred in #684

New Contributors

• @KislayaRavi made their first contribution in #683
• @rosonaeldred made their first contribution in #684

Full Changelog: v0.3.0...v0.3.1

What's Changed

• fix typo MBO_LENGTHCALE_PRIOR -> MBO_LENGTHSCALE_PRIOR by @Jimbo994 in #610
• number_of_categorical_combinations by @jduerholt in #616
• Introduce LogTransformedInterval Prior Constraint by @jduerholt in #618
• 562 pydantic deprecation warnings by @Jimbo994 in #617
• Update for botorch 0.15.0 by @jduerholt in #619
• Make notebook render well by @jduerholt in #622
• Tidy-up Tutorial on Optimality Criteria by @jduerholt in #621
• Update link to Gauche by @jduerholt in #623
• improve link to chemistry by @jduerholt in #620
• Fix/624 by @LukasHebing in #625
• Fix/628 by @LukasHebing in #630
• Wrapper für Botorch Test Functions by @jduerholt in #632
• suggestions to NChooseK repair function by @Jimbo994 in #633
• Active Learning Strategy: fix missing input processing for acqf samples. by @jkeupp in #634
• Formulation and SpuriousFeature Wrappers by @jduerholt in #636
• Add AdditiveSaasModel to BotorchSurrogates by @jduerholt in #637
• New Pydantic Version Adaptions by @jduerholt in #638
• Add nchoosek for forumlation wrapper by @jduerholt in #639
• add clamping by @jduerholt in #641
• Hotfix/stability by @jduerholt in #640
• Refactor Input Preproccesing and Mixed Optimization by @jduerholt in #626
• Feature/ Exclude constraints usable in GA by @LukasHebing in #644
• Fix formulation wrapper by @jduerholt in #647
• change badge-svg to main by @bertiqwerty in #649
• Enable manual trigger of test pipeline by @jduerholt in #654
• Conditional input features by @TobyBoyne in #645
• Hotfix Acqf setup for fixed-features by @jduerholt in #664
• Removed XGBoost and removed scikit-learn version constraint. by @md-121 in #666

New Contributors

• @md-121 made their first contribution in #666

Full Changelog: v0.2.1...v0.3.0

Maintenance release that solves the issue of failing acquisition setup when used with fixed features that have not been used before.

What's Changed

• Fix docu domain.md by @KappatC in #599
• Fix error in domain doc by @jduerholt in #600
• Probabiltiy of Feasibility ACQF by @jduerholt in #580
• 595 bug categorical doe unstable by @dlinzner-bcs in #597
• EDBO Priors by @jduerholt in #602
• Shap by @jduerholt in #592
• Feature/exclude constraints in bo by @jduerholt in #604
• is_fulfilled for Domain and Inputs by @jduerholt in #605
• Feature/modularize ga by @LukasHebing in #596
• Bugfix in Hyperopt Functionality of SingleTaskGPSurrogate by @jduerholt in #607
• Hvarfner Priors for MixedSingleTaskGP by @jduerholt in #609
• fix input/output - mixup by @bertiqwerty in #608
• Feature/relevance pursuit Implemention of RobustRelevancePursuitSingleTaskGP into BoFire by @Jimbo994 in #591
• Additions for Fully Bayesian Models by @jduerholt in #606

New Contributors

• @KappatC made their first contribution in #599

Full Changelog: v0.2.0...v0.2.1

This release contains a major interface simplification. A strategy can now be instantiated via

from bofire.strategies.api import SoboStrategy
sobo = SoboStrategy.make(domain=domain)

Of course, the old way still works to keep the separation between data models and functionality possible:

import bofire.strategies.api as strategies
from bofire.data_models.strategies.api import SoboStrategy as DataModel
data_model = DataModel(domain=domain)
sobo = strategies.map(data_model)

Further note that we have now a GA based optimizer for acquisition functions, see this tutorial.

What's Changed

• Dtype not specified in ActiveLearningStrategy by @jduerholt in #576
• Expose sequential to data model by @jduerholt in #577
• Make the refpoint setting for MOBO more flexible by @jduerholt in #518
• Feature/add ga optimizer by @LukasHebing in #561
• pragmatic support for discrete and categorical variables by @dlinzner-bcs in #575
• New readme and getting_started notebook by @chrihaas in #472
• add jingle by @dlinzner-bcs in #583
• Bofire music jingle by @bertiqwerty in #585
• Fix missing documentation by including subfolders automatically by @rlars in #584
• Create a make method in Strategys by @TobyBoyne in #572
• CategoricalExcludeConstraint by @jduerholt in #582
• bounds can be tuple by @bertiqwerty in #567
• fix edgecase by @dlinzner-bcs in #590
• Working Stepsize Implementation by @jduerholt in #588
• Remove cvxopt by @LukasHebing in #594

New Contributors

• @rlars made their first contribution in #584

Full Changelog: v0.1.0...v0.2.0

What's Changed

• Polynomial feature interaction kernel by @e-dorigatti in #520
• nonlinear constraints without torch by @Osburg in #527
• Proper error message for strategy.predict by @jduerholt in #528
• Fix pyright issues by @CompRhys in #505
• Strategy TidyUp by @jduerholt in #529
• 'added code to plot cv folds with plotly' by @Jimbo994 in #534
• Blocking for Fractional Factorial Designs by @jduerholt in #532
• Skip XGBoost test if uninstalled by @TobyBoyne in #538
• Pull Request Template by @jduerholt in #539
• replace .json() with .model_dump_json() by @R-M-Lee in #543
• fix types in getting started by @bertiqwerty in #541
• Bump image for doc pipelines by @jduerholt in #537
• add note on how to print hyperparameters by @bertiqwerty in #544
• test coverage note by @bertiqwerty in #545
• Docs - maximize acquisition functions by @bertiqwerty in #540
• Comment on usage of seeds by @jduerholt in #547
• Templates for issues by @jduerholt in #549
• docs used deprecated pydantic by @bertiqwerty in #550
• add api examples to docs by @bertiqwerty in #548
• Bump Entmoot and LightGBM by @TobyBoyne in #536
• Solve frequently failing doe test by @jduerholt in #546
• Documentation on using categorical outputs by @jduerholt in #552
• Fix doc building by @jduerholt in #554
• Refactor Installation Guide by @jduerholt in #557
• Lift data-models in docs by @bertiqwerty in #560
• Add chemistry focus to documentation by @TobyBoyne in #555
• Refactor ACQF Optimization by @jduerholt in #535
• Use Seed Sequences for random number generation by @TobyBoyne in #566
• fix broken link in tutorials README.md by @R-M-Lee in #564
• Reintroduce sampling parameter into DoEStrategy to allow for seeding by @dlinzner-bcs in #559
• Python 3.10 support by @bertiqwerty in #570
• DoE objective evaluation in pytorch by @Osburg in #525
• Docs versioning by @bertiqwerty in #573

Full Changelog: v0.0.18...v0.1.0

What's Changed

• pin cyipopt version in conda install by @jduerholt in #516
• Custom Sobo Tutorial by @jduerholt in #508
• added plot_gp_slice.py by @Jimbo994 in #499
• Make pyright a seperate test job by @jduerholt in #524
• get_by_key using regex by @jduerholt in #523
• Fix that BoFire is not working without cyipopt by @jduerholt in #522

New Contributors

• @Jimbo994 made their first contribution in #499

Full Changelog: v0.0.17...v0.0.18

What's Changed

• Fix mapper tests by @jduerholt in #502
• Adding Desirability functions as objectives by @LukasHebing in #497
• Refactoring of DoEStrategy data model class by @jduerholt in #448
• Feature/add desirability function in range by @LukasHebing in #504
• Fix wrong desirabilities in predictive strategies by @jduerholt in #506
• I optimality criterion and nonlinear constraints defined by python function by @Osburg in #485
• Defined OptimalityCriteria two times by accident by @dlinzner-bcs in #507
• Make doe tests fail less by @dlinzner-bcs in #509
• Make BoFire compatible to BoTorch 0.13.0 by @jduerholt in #511
• Randomization for FractionalFactorials by @jduerholt in #510
• Fix failing doe notebooks by @Osburg in #512
• Mark DesirabilityObjective as only abstract by @jduerholt in #513

Full Changelog: v0.0.16...v0.0.17

What's Changed

• Add pandas data manipulation transform by @jduerholt in #460
• get_by_keys with exclude option by @jduerholt in #465
• [QOL] cancel tests on same PR by @CompRhys in #467
• Suppress import warning prints by @chrihaas in #466
• StepwiseStrategy: Make domain comparison less strict by @jduerholt in #468
• Fix PiecewiselinearGPSurrogate by @jduerholt in #469
• Multifidelity Strategy by @TobyBoyne in #453
• Fix NumberOfExperimentsCondition by @chrihaas in #470
• Add example notebook that demonstrates how to use inputs also as outputs by @jduerholt in #475
• Move setup.py to pyproject.toml by @chrihaas in #473
• Categorical deterministic surrogate by @jduerholt in #479
• Add categoricals for FractionalFactorialStrategy by @jduerholt in #480
• pin compatible version of packages by @e-dorigatti in #489
• Multiplicative additive sobo objectives by @LukasHebing in #481
• Progessbar for Hyperopt by @jduerholt in #494
• Make the BoFire Data Models OpenAI compatible by @jduerholt in #495
• Group split kfold by @jkeupp in #484
• Make candidate_count enforcement less strict by @jduerholt in #492
• Drop support for Python 3.9 by @jduerholt in #493
• Slowdown due to Scipy 1.15 by @jduerholt in #501
• kernels working on a given set of features by @e-dorigatti in #476

New Contributors

• @chrihaas made their first contribution in #466
• @e-dorigatti made their first contribution in #489
• @LukasHebing made their first contribution in #481

Full Changelog: v0.0.15...v0.0.16

What's Changed

• sketch of how default kernel will change by @R-M-Lee in #435
• Fix enum serialization problems by @jduerholt in #458
• [Lint] apply ruff format as all safe fixes by @CompRhys in #456

Full Changelog: v0.0.14...v0.0.15