ribs.emitters.rankers.ImprovementRanker

class ribs.emitters.rankers.ImprovementRanker(seed=None)[source]

Ranks the solutions based on the improvement in the objective.

This ranker ranks solutions in a single stage. The solutions are ranked by the improvement “value” described in ribs.archives.ArchiveBase.add().

This ranker should not be used with CMA-ME. The improvement values for new solutions in CMA-ME are on a different scale from the improvement values for the other solutions, in that new solutions have an improvement value which is simply their objective, while other solutions have an improvement value which is the difference between their objective and the objective of their corresponding cell in the archive.

Methods

rank(emitter, archive, data, add_info)

Generates a list of indices that represents an ordering of solutions.

reset(emitter, archive)

Resets the internal state of the ranker.

rank(emitter, archive, data, add_info)[source]

Generates a list of indices that represents an ordering of solutions.

Parameters

  • emitter (ribs.emitters.EmitterBase) – Emitter that this ranker object belongs to.

  • archive (ribs.archives.ArchiveBase) – Archive used by emitter when creating and inserting solutions.

  • data (dict) – Dict mapping from field names like solution and objective to arrays with data for the solutions. Rankers at least assume the presence of the solution key.

  • add_info (dict) – Information returned by an archive’s add() method.

Returns

the first array (shape (batch_size,)) is an array of indices representing a ranking of the solutions and the second array (shape (batch_size,) or (batch_size, n_values)``) is an array of values that this ranker used to rank the solutions. batch_size is the number of solutions and n_values is the number of values that the rank function used.

Return type

tuple(numpy.ndarray, numpy.ndarray)

reset(emitter, archive)

Resets the internal state of the ranker.

Parameters