chore(main): release 0.15.0

[denehoffman]

🤖 I have created a release beep boop

0.15.0 (2026-02-10)

⚠ BREAKING CHANGES

• change Variable trait methods to operate on &Dataset rather than &Arc<Dataset>

Features

• Add 'threads' argument to pretty much everything in Python API (ed5502a)
• Add __radd__ implementations wherever __add__ is implemented (150c068)
• Add __sub__, __rsub__, and __neg__ implementations to Python Vector3 and Vector4 (59367c0)
• Add BinnedGuideTerm under new experimental module (8d0c626)
• Add conj operator to Amplitudes and Expressions (62dfe28)
• Add Dataset::weighted_bootstrap (9f638fb)
• Add Debug and Display to every Variable (and require them for new ones) (ab8ec68)
• Add evaluate(Variable) method to Dataset and Event (764bd7f)
• Add NLL::project_with to do projecting and isolation in one step (38a495c)
• Add PhaseSpaceFactor amplitude (a439b99)
• Add PiecewiseScalar, PiecewiseComplexScalar, and PiecewisePolarComplexScalar amplitudes (dd1f657)
• Add PolPhase amplitude (07ebd06)
• Add project_gradient and project_gradient_with methods to NLL (98327dd)
• Add available_parallelism function to python API (b06d620)
• Add basic implementation to read directly from ROOT files (0e3a4a6)
• Add evaluate_gradient functions where evaluate is available in python API (092fc1f)
• Add experimental Regularizer likelihood term (de4ae47)
• Add gamma factor calculation to 4-momentum (1975fee)
• Add gradient evaluation to NLL and LikelihoodExpression in Python API (3c45181)
• Add intermediate Model and serialization updates (561753b)
• Add MCMC samplers from ganesh to laddu (5cb213a)
• Add method for boosting an event or a dataset to a given rest frame (1efab27)
• Add method for opening a dataset boosted to the rest frame of the given p4 indices (2d10626)
• Add methods to serialize/deserialize fit results (ffd1b03)
• Add no-op implementations for adding 0 to add-able types (a8f6323)
• Add perf commands to justfile (3c27f69)
• Add python versions of Point, Particle, SwarmObserver, and Swarm from ganesh (59a84eb)
• Add PyVariable to simplify python functions which need to take generic Variables (348deb2)
• Add subtraction, division, and negation operations for all Amplitudes and Expressions (c9b3b3f)
• Add Swarm methods to Python API and update other algorithm initialization methods (ddc6813)
• Add the ability to name likelihood terms and convenience methods for null and unit likelihood terms, sums, and products (c5352d7)
• Add unit-valued Expression and define convenience methods for summing and multiplying lists of Amplitudes (c1297eb)
• Add VariableExpressions to handle Dataset filtering (3f01968)
• Add vector indexing and Event.get_p4_sum to python API (8b97865)
• Allow for the use of sum(list[Dataset]) in Python code (631fe49)
• Allow users to add Datasets together to form a new Dataset (ef6f80e)
• amplitudes: Add From impl for AmplitudeID to Expression conversion (a0cd045)
• bench: Updated benchmark to run over available parallelism (0b62180)
• Bump ganesh to add "skip_hessian" minimization option to skip calculation of Hessian matrix (fd1eea8)
• Bump MSRV (for bincode) and bump all dependency versions (5c7aab8)
• Bump versions on ganesh, pyo3, and (rust) numpy and make appropriate updates for each (3f2d67a)
• Change MSRV to 1.84.0 and resolver to 3 (31e9172)
• Change swarm repr if the swarm is uninitialized to not confuse people (aaaf144)
• Change the way NLLs are constructed to allow the user to specify a generated dataset (7755ff6)
• Change the way NLLs are constructed to allow the user to specify a generated dataset (a210106)
• Create example_2, a moment analysis (7dd7c01)
• data: Create test_dataset method for testing purposes as well as add Default impl to Event (1aff54c)
• data: Make Event::get_p4_sum generic over its argument (4f93af6)
• Display the AmplitudeID's name and ID together (d1a0f1c)
• enums: Add Channel enum (417eb99)
• enums: Add equality comparison to enums and convert to lowercase before string conversion (de3b205)
• enums: Add serde to enums (5ab8b13)
• First pass implementation of MPI interface (16d8391)
• Implement custom gradient for BinnedGuideTerm (0945647)
• Improve error handling by getting rid of unwraps wherever possible (7968904)
• Improvements to Dataset conversions and opening methods (1264ef2)
• Make mpi a feature in py-laddu to allow people to build the python package without it (ae23bb2)
• Move binning functions to utils (f7fe352)
• Redefine eps->aux in Event definition (45df457)
• Restructure the minimizer/mcmc methods to no longer take kwargs (c88feea)
• Separate parameter logic into a new struct and unify fixing/freeing/renaming (834a4e7)
• Split laddu python package into two, with and without MPI support (e488b46)
• Switch API for acceptance correction to not process genmc till projection (1d404a5)
• Switch the MPI implementation to use safe Rust via a RwLock (58ecc24)
• Unify binning functions and add (f914e58)
• Update ganesh version and add Global move to ESS (a5bcf4c)
• Update ganesh, numpy, and pyo3 (ebf42ad)
• Update MPI code to use root-node-agnostic methods (109846e)
• Update type hints with ropts and add magic methods to easily pickle Status (e4bd21b)
• Use absolute value of mass and width in Breit-Wigner amplitude (1d07b92)
• variables: Add Mandelstam variables (29f0b19)

Bug Fixes

• Add a few more missed pyclasses to the py-laddu exports (d143fb8)
• Add better feature guards so all features can technically be used independently (ff4205d)
• Add fallback overloads to several amplitudes (3559264)
• Add feature flag to laddu-python and update MSRV for mpisys compatibility (5c0fb32)
• Add feature flag to benchmark to allow it to be run with "f32" feature (13ae5f3)
• Add more precision to covariance matrices to ensure positive definiteness (49e43ae)
• Add mpi feature for laddu-python to py-laddu (02dc5ed)
• Add non-MPI failing functions for MPI calls on non-MPI python builds (6be7e24)
• Add python feature gate to likelihood-related methods (5719813)
• Add sentry dependency for which to force the version, not sure what the best fix really is, but this should work for now (36e938a)
• Add type hints to __doc__ and __version__ (0c9ffe2)
• Calling get_world before use_mpi causes errors (54df1af)
• Change LikelihoodTerm naming to happen at registration time (ca3516d)
• Change NLL to always use len(accmc) for n_mc (4a9dc38)
• Change unwrap to print error and panic (c2dfa88)
• Convert range to list (63cc88b)
• Correct arguments to fix open_amptools method (8c1b4d8)
• Correct for out-of-bounds errors in MPI bootstrap (ab173b1)
• Correct gradients of zero and one by adding the number of parameters into GradientValues (d37346e)
• Correct justfile path (6feea91)
• Correct parallelism to allow for proper codspeed benchmarking (27a61a1)
• Correct python method to get energy from 4-vector (fc3be63)
• Correct some issues with conversion scripts (e7a1099)
• Correct the open method and counts/displs methods (d4b3ff5)
• Correct typo AIES->AEIS in python type checking files (88dfbc5)
• Corrected signature in methods that read from AmpTools trees (d751c37)
• Corrected signature in methods that read from AmpTools trees (b93ae60)
• Corrected typo where the VerboseMCMCObserver implemented SwarmObserver<()> rather than the VerboseSwarmObserver (4751377)
• Ensure LikelihoodEvaluators report parameter names in fit/mcmc summaries (b90b230)
• enums: Gottfried-Jackson string conversions were accidentally being redirected to Helicity (d57f078)
• Fixed python examples and readme paths (56f35a4)
• Forgot to bump all versions (1284f2e)
• Forgot to export MCMC moves (da899fd)
• Forgot to remove the error types for data reading (a0b7ea9)
• Forgot to switch one coverage attribute to nightly (100c0b1)
• Forgot to update the laddu-python use_mpi function to have a trigger arg (1a10c3d)
• Get rid of unused variable warning (bd26682)
• Improve summation and product methods to only return a Zero or One if the list is empty (54afa62)
• Major bug/typo in boost method and tests to check for it (efa1a24)
• Make some small corrections to the python example_1 and test out the built-in autocorrelation observer (5e31018)
• Make sure code works if no pol angle/magnitude are provided (9bd054c)
• Minor fixes for building without rayon/pyo3 (9a8f23a)
• Minor modification to allow for MSRV of 1.70 (ac553cd)
• Missed a changed path in some ganesh code hidden behind a feature gate (2816e41)
• Missed AEISMove->AIESMove and condensed the lib.rs file for py-laddu (a6e9afc)
• Modify tests and workflows to new structure (33f456c)
• More fixes for newest ganesh version (4d493e3)
• Move rayon feature bounds inside methods to clean up the code and avoid duplication (ff7c47e)
• Move some imports under the python feature flag (3970a67)
• Remove from the rayon-free calls for and (5bf105e)
• Remove on trait (45a7292)
• Remove PyResult on methods which no longer produce errors (f453907)
• Remove loguru and pandas dependencies and add pyarrow dependency (8048674)
• Remove requirement that pickled files end in ".pickle" or ".pkl" (65b5513)
• Reorganize package structure (6020e1d)
• Replace f64 with Float in generate_bin_edges (26bd863)
• Revert MSRV to 1.70.0 (ef57f9d)
• Sync all versions with a single release (fc7569a)
• Sync versions across crates/libraries (65967cf)
• The last commit fixed the typo the wrong way, it is AIES (bc193d6)
• Unwrap call_method so that it reports the stack trace if the method fails (e673c5c)
• Update CI to properly upload artifacts (58a75ca)
• Update coverage to reflect new layout (dc0378c)
• Update GradientValues in non-default feature branches (0bd9560)
• Update GradientValues in non-default feature branches (missed one) (023dffc)
• Update Python vector names (closes #57) (d0f8ee0)
• Use correct enum in L-BFGS-B error method (2ae0d45)
• Use proper ownership in setting algorithm error mode (001505a)
• Use proper path in laddu-mpi entrypoint (8919d64)

Reverts

• Remove weighted_bootstrap (891926f)

Code Refactoring

• Change Variable trait methods to operate on &Dataset rather than &Arc<Dataset> (e6b2842)

---

This PR was generated with Release Please. See documentation.

[codspeed-hq]

Merging this PR will not alter performance

✅ 4 untouched benchmarks

---

_{Comparing release-please--branches--main (7dd2d26) with main (fc7569a)}

[denehoffman]

🤖 Created releases:

• v0.15.0

🌻