Skip to content

Latest commit

 

History

History
121 lines (94 loc) · 5.95 KB

CHANGELOG.md

File metadata and controls

121 lines (94 loc) · 5.95 KB

LEAP CHANGES BY VERSION

Being a terse compilation by version of changes.

0.7.0

0.6.0, 6/13/2021

  • Drop support for Python 3.6

    • This keeps us in sync with numpy and dask that also dropped support for 3.6 this year

  • New features

    • Added landscape_features package with some initial exploratory landscape analysis tools
    • Added elitism
    • Added a new example demonstrating integer representations
    • Added a mutate_binomial() operator for integer representations
    • Added visualization of ANN weights for SimpleNeuralNetworkExecutable phenotypes
    • Added metrics for logging population diversity
    • Added support for lexicographical and Koza-style parsimony pressure
    • Added HistPhenotypePlotProbe
    • Added ops.grouped_evaluate() for evaluating batches of individuals
    • Added ExternalProcessproblem for using external programs as fitness functions

  • Documentation

    • Added documentation on leap_ec.context and updated software development guidelines to encourage its use if tracking persistent state outside of function calls was necessary.

  • CI/CD

    • Added a make test-slow harness
    • Added tests that run the examples/ scripts
    • Organized examples into subdirectories
    • Improved test coverage

  • Bugfixes

    • Fix viz parameter when calling simple.ea_solve()
    • Fix algebra error in real_rep.problems.NoisyQuarticProblem
    • Tell dask that functions are impure by default, to make sure it doesn't cache results
    • Change Makefile to use pip install -e . instead of the deprecated python setup.py develop

  • API changes

    • Significantly refactored the executable_rep.rules package to simplify learning classifier systems
    • Added leap_ec.__version__ attribute
    • Added a hard_bounds flag to ea_solve() to tell it to respect the bounds at all times (rather than just initialization); defaults to True
    • Added the most frequent imports (ex. Individual, Representation) into the top-level package
    • Renamed the generations parameter of generational_ea() to max_generations and added an optional stop parameter for other stopping conditions
    • Added probability parameter for the uniform_crossover operator
    • mutate_gaussian now accepts a list of gene-wise hard bound
    • Added select_worst Boolean parameter to tournament_selection
    • Added notes columns parameter to FitnessStatsCSVProbe
    • Added a pad_inputs parameter to TruthTableProblem to handle varying-dimension inputs
    • Added a pad parameter to CartesianPhenotypePlotProbe to plot 2D projections of higher-D functions
    • Added FitnessPlotProbe as a convenience wrapper for PopulationMetricsPlotProbe
    • Added an x_axis_value parameter to FitnessPlotProbe and PopulationMetricsPlotProbe
    • Renamed PlotTrajectoryProbe to the more descriptive CartesianPhenotypePlotProbe
    • Renamed PopulationPlotProbe to the more descriptive PopulationMetricsPlotProbe
    • Renamed leap_ec.distributed to leap_ec.distrib to reduce name space confusion with dask.distributed
    • Renamed leap_ec.context to leap_ec.global_vars
    • Default behavior changes
      • Individual.decoder and Representation.decoder now uses a phenotypic representation (IdentityDecoder) by default
      • Mutation operators no longer have default mutation rates (they must be explicitly set by the user).
      • Set default p_swap = 0.2 for uniform_crossover, instead of 0.5
      • Set default num_points = 2 for n_ary_crossover, instead of 1
      • Set default value for context parameter on probes, so users needn't set it
      • standardized on making context last function argument that defaults to leap_ec.context.context

0.5.0, 1/9/2021

  • Added probability parameter for the n_ary_crossover operator
  • Greatly improved test coverage
  • Added support for static- and variable-length segments, which are fixed-length "chunks" of values
  • Added a simple neural network representation, executable_rep.neural_network, and made it the default for examples/openai_gym.py
  • Changed the Executable interface to act as a Callable object (rather than using a custom output() method)
  • Added statistical_helpers to assist with writing unit tests for stochastic algorithms
  • Added support for integer representations, via the int_rep package
  • Added a Cartesian genetic programming (CGP) representation, executable_rep.cgp, with example in examples/cgp.py
  • Added support for heterogeneous island models, demoed in examples/multitask_island_model.py

0.4.0, 9/19/2020

  • Significantly added to online documentation
  • Major code reorganization
    • exception management for Individual has been moved to RobustIndividual
    • DistributedIndividual now inherits from RobustIndividual
    • core.py has been broken out to separate modules
      • Individual and RobustIndividual now in individual.py
      • representation specific entities moved to new sub-packages, binary_rep and real_rep
      • Representation now in representation.py
      • Decoder now in decoder.py
    • documentation, doctests, examples, Jupyter notebooks, and unit tests updated accordingly
  • added ability to pass ancillary information during evaluation, such as UUIDs that could be used to name output files and directories, yet do not have a direct impact on fitness

0.3.1

  • Apply Representation consistently throughout LEAP, particularly the top-level monolithic functions
  • Added probe to leap_ec.distributed.asynchronous.steady_state() to take regular snapshots of the population

0.3, 6/14/2020

  • fix how non-viable individuals sort themselves when compared since the prior method of comparing math.nan to math.nan yielded non-ideal behavior
  • minor maintenance tweaks

0.2, 6/14/2020

  • changed package name to leap_ec from leap to mitigate pypi namespace collisions
  • minor maintenance tweaks

0.1

  • first major "mature" release of LEAP