Version 3.3.0

• 23 May 2026 version 3.3.0
  
  • New Classes
    • EclipsingStarModel, FlippedDataProblem, PeriodicScout, UniformRatioPrior
    • EclipsingStars explains EclipsingStarModel and PeriodicScout
  • New Examples and Tests
    • heartbeat-045.ipynb, TestEclipsingStarModel, TestPeriodicScout, TestWalker
  • Plotter (addressing pull request #28)
    • Move methods using matplotlib to Plotter
    • Added methods: plotWalker, plotIter, plotOrbit, plotOrbit3D, lineOfNodes,
      plotEclipsingStar, plotEsmSideView, plotEsmEclipseView
  • StellarOrbitModel, Tools
    • Lot of additions for use in EclipsingStarModel, e.g. 3D orbit
    • Methods toRect() and toSpher(), in 2 and 3 dim, moved to Tools
  • NestedSampler, IterativeFitter, LevenbergMarquardtFitter
    • Moved plotting stuff to Plotter
  • GalileanEngine
    • Skip mirroring when partial LogL fails (with warning)
  • Removed splineslab. One methods of splineslab moved to BSplinesModel
  • Removed references to future
    • pyproject.toml, .gitignore, bspline.py
  • Minor additions, documentation etc.
  • Rerun all examples
  • Added to Testharnasses
  • Update documentation
    • manual
    • classdocs

Version 3.2.5

• 20 Oct 2025 version 3.2.5
  
  • Removed unused import statements from source and test (pull request #1)

Version 3.2.4

• 13 Aug 2025 version 3.2.4
  
  • Lauch new web site https://dokester.github.io/BayesicFitting/
    Almost all docstrings adapted for the new site.
  • GalileanEngine
    • add a pertubation attribute: wiggle=0.2. See [Dimensions](./docs/Dimensions$
    • find edge by repeated quadratic interpolation before mirroring
  • NestedSampler
    • add new attribute avoid=0.1. See Dimensions
  • PhantomSampler now uses all phantoms to do the evidence integral
    • Adaptions in Engines, Explorer, WalkerList, NestedSampler.
    • PhantomCollection has one WalkerList also in Dynamic cases.
  • Removed Deprecations, Checks and/or Warnings.
    • Model, ErrorDistributions, Problem
    • Formatter
    • Sample

Version 3.2.3

Issue 27: Remove invalid escape sequences from docstrings.

Version 3.2.2

05 Nov 2024 version 3.2.2
Adapt to numpy-2 and pertaining Scipy, Astropy and MatplotLib

V.3.2.1

• 27 March 2024 version 3.2.1
  
  A lot of fairly small stuff.
  • status changed to Alpha for some newer additions
  • Priors:
    • Adapt to input arrays
    • Made limited and/or circular
    • Proper integral when limited
  • Engines:
    • Add bestCheck and bestBoost
    • Minor restructoring
  • Models:
    • Minor restructoring and renaming
    • setLimits() replaced by setPrior() in RepeatingModel
    • provide for more dimensions of outputs in derivative
  • Problems:
    • toString() method restructured
  • Walker and WalkerList:
    • Add logPrior attribute
    • Change in inheritance reporting
  • PhantomCollection:
    • getList() method
  • NestedSampler:
    • Add repiter attribute: report every repiter when verbose=2
    • Add bestBoost
  • ModelDistribution:
    • put internal sample() method in try-except block
  • Tools:
    • subclassof() method
    • printclass finetuning
  • Test and documentation. Update.

Version 3.2.0

• 31 Aug 2023 version 3.2.0

• The new class, PhantomCollection is part of NestedSampler. It contains a sorted
  WalkerList, in which all valid positions are collected, proper walkers
  and phantoms. Each iteration the phantoms with log likelihood lower than the
  low limit for that iteration, are removed. The PhantomCollection is used
  to get a better estimate on the size of the bounding box for the walkers and
  to obtain starting positions for new walkers. In general the PhantomCollection
  contains an order of magnitude more items than the WalkerList itself.
• NestedSampler has a new stopping criterion. It also stops when the log of the
  relative contribution to the logZ (evidence) integral is less than -tolerance (=12).
• FootballModel: new class. Model to estimate strengths of football teams in
  several key parameters.
• Address PhantomCollection and add **kwargs in all Engines
• Quadratic (in stead of linear) interpolation on edge in GalileanEngine
• Unnormalized Gauss prior changed into a normalized one
• Some documentation issues
• More dimensional arrays in LogFactorial
• str() method in NestedSolver and PhantomSampler
• Avoid numeric instabilities in sqrt in SampleList
• New tools in Tools
• More construction options in WalkerList
• New example: Uefa2022.ipynb
• New tests: TestPhantomCollection, TestFootballModel
• Adaptations in existing tests
• Reran all tests and examples

Version 3.1.1

What's new.

• 18 Jan 2023 version 3.1.1
  • remode GaussPriorNew from init.py

Version 3.1.0

What's new.

• 18 Jan 2023 version 3.1.0 (still working on the same update)

  • Implementing accuracy in fitters and samplers
  • Update of documentation
  • More tests

• 19 Nov 2022 version 3.1.0

  • Gauss2dErrorDistribution: New Class to handle correlated errors in X and Y
  • ErrorDistribution and GaussErrorDistribution : adaptation for covariant errors.
  • Small updates and corrections and removal of unused methods
  • Added / corrected version information and documentation issues
  • Rerun all examples and added tests

Version 3.0.1

05 Apr 2022 version 3.0.1:
Addressing issue #18: UserModel does not work for multiple dimensions.