Free cookie consent management tool by TermsFeed Policy Generator

Opened 15 years ago

Closed 15 years ago

Last modified 14 years ago

#932 closed enhancement (done)

Implement Algorithms: Evolution Strategy

Reported by: abeham Owned by: abeham
Priority: high Milestone: HeuristicLab 3.3.0
Component: Algorithms.EvolutionStrategy Version: 3.3
Keywords: Cc:

Description

We need an evolution strategy

Change History (16)

comment:1 Changed 15 years ago by abeham

  • Priority changed from major to critical
  • Status changed from new to assigned

comment:2 Changed 15 years ago by abeham

Added first version of ES in r3130

comment:3 Changed 15 years ago by abeham

  • Component changed from ### Undefined ### to Algorithms.EvolutionStrategy

comment:4 Changed 15 years ago by abeham

fixed a bug when there is no visualizer in r3146

comment:5 Changed 15 years ago by abeham

fixed a bug in ES in r3148 Also included some parameter sanity checks, still missing is a check that determines if the currently selected ParentsPerChild are valid with the currently selected recombinator. Especially the GA crossovers just allow 2 parents.

Maybe we need a property in ICrossover that an implementing crossover can override which says something about the maximum number of parents (bounded by int.MaxValue).

comment:6 Changed 15 years ago by abeham

added self adaptive crossover and mutation strength adjustment in r3183

comment:7 Changed 15 years ago by abeham

fixed selection of normal all positions adaptive mutation and crossover other than adaptive crossover in r3185

The SelfAdaptiveNormalAllPositionsManipulator is pretty ES specific, should it be moved to the ES namespace?

comment:8 Changed 15 years ago by abeham

TODO: Use the ProblemSize variable instead of the ES specific counterpart.

comment:9 Changed 15 years ago by abeham

fixed serialization bug in r3326

comment:10 Changed 15 years ago by abeham

r3336

  • The ProblemSize variable cannot be easily used from the algorithm, because it is not in IProblem and because not every problem has a problem size
  • Thus there have been some more extensive changes... take a look at the change set

comment:11 Changed 15 years ago by abeham

r3450

  • changed handling of strategy parameters

comment:12 Changed 15 years ago by abeham

r3678

  • updated evolution strategy

comment:13 Changed 15 years ago by abeham

r3782

  • changed some ES defaults
  • fixed ISelfAdaptiveManipulator to derive from IManipulator

comment:14 Changed 15 years ago by abeham

  • Resolution set to fixed
  • Status changed from assigned to closed

A sigma self adaptive ES has been implemented in 3.3

comment:15 Changed 14 years ago by swagner

  • Milestone changed from Iteration 4 to Current

Milestone Iteration 4 deleted

comment:11 Changed 14 years ago by swagner

  • Milestone changed from Current to HeuristicLab 3.3.0

Milestone Current deleted

Note: See TracTickets for help on using tickets.