Breeder GA

[francastagna]

PR Summary

We implemented the Breeder Genetic Algorithm (BGA) and added comprehensive tests.

---

Breeder GA Pseudocode:

Source: https://www.researchgate.net/publication/327166354_An_Empirical_Evaluation_of_Evolutionary_Algorithms_for_Unit_Test_Suite_Generation

---

Main Changes

• New Algorithm (BreederGeneticAlgorithm):

  • Performs truncation selection to form parents pool P'.
  • At each step, creates 2 offspring from random parents in P', then carries over exactly 1 child.
  • Invokes lifecycle hooks: beginGeneration, endGeneration, beginStep, endStep.
  • Freezes targets each iteration: frozenTargets = archive.notCoveredTargets().

• Config (EMConfig):

  • Added BreederGA to EMConfig.Algorithm.
  • Added breederTruncationFraction (default 0.5, range (0,1]) and breederParentsMin (default 2).

---

Tests

• Breeder GA tests added (BreederGeneticAlgorithmTest.kt):
  • Finds optimum on OneMax (testBreederGeneticAlgorithmFindsOptimum).
  • Next generation shape (testNextGenerationIsElitesPlusTruncationChildren):
    • Preserves population size, carries elites, and adds exactly one child per iteration.
    • Verifies crossover and mutation counts per iteration.
  • Edge cases:
    • xoverProbability = 0.0, fixedRateMutation = 1.0 → no crossovers; exactly 2 * populationSize mutations; size preserved.
    • fixedRateMutation = 0.0, xoverProbability = 1.0 → exactly populationSize crossovers; zero mutations; size preserved.

---

[francastagna]

rp retest