#84 closed enhancement (done)
Merge Sigma-Self-Adaptive and Success Rule Adaptive ES
| Reported by: | abeham | Owned by: | abeham |
|---|---|---|---|
| Priority: | medium | Milestone: | HeuristicLab 3.3.0 |
| Component: | ZZZ OBSOLETE: ES | Version: | 3.2 |
| Keywords: | Cc: |
Description (last modified by abeham)
Instead of having two different ES, one with Sigma self adaption and the other with Success Rule adaption they could be merged by using a DoubleArrayData ShakingFactor. This array should hold 1 strategy parameter in the success rule case and n strategy parameters in the self adaption case.
Change History (12)
comment:1 Changed 17 years ago by abeham
- Description modified (diff)
- Summary changed from Add Sigma-Self Adaptive ES interface to Merge Sigma-Self-Adaptive and Success Rule Adaptive ES
comment:2 Changed 16 years ago by abeham
- Milestone changed from 3.0 to 3.2
- Version changed from 3.0 to 3.2
comment:3 Changed 16 years ago by abeham
- Status changed from new to assigned
comment:4 Changed 16 years ago by abeham
- Owner abeham deleted
- Status changed from assigned to new
comment:5 Changed 16 years ago by abeham
- Owner set to abeham
- Status changed from new to assigned
There is still a bug in the GUI: Problem Dimension should initialize to the dimensions defined in the initial strength vector
comment:6 follow-up: ↓ 7 Changed 16 years ago by abeham
After the talk with Michael following features should be implemented also:
- Automatic setting of LearningRate / GeneralLearningRate according to the formula given by e.g. Beyer and Schwefel 2002.
- Making sure that a stochastic crossover combination is possible
- Adding different recombination operators implementing the various strategies regarding the crossover of object and strategy vector: discrete / intermediate, intermediate / intermediate and discrete / discrete.
comment:7 in reply to: ↑ 6 Changed 16 years ago by abeham
done 1. in r1112
comment:8 Changed 16 years ago by swagner
- Milestone changed from 3.2 to Iteration 0
Milestone 3.2 deleted
comment:9 Changed 16 years ago by abeham
No 2 and 3 have been fixed with the change to crossover (ticket #470) The GUI bug mentioned earlier has been fixed too
comment:10 Changed 16 years ago by abeham
- Resolution set to fixed
- Status changed from assigned to closed
changed the RealVector mutation operator in r1180 to allow e.g. strategy vectors of size 1, so that one strategy value for all dimensions is possible as well as one value for each dimension.
this closes the ticket
comment:11 Changed 14 years ago by swagner
- Milestone changed from Iteration 1 to Current
Milestone Iteration 1 deleted
comment:11 Changed 14 years ago by swagner
- Milestone changed from Current to HeuristicLab 3.3.0
Milestone Current deleted
updated the interface to reflect the changes in r1091 I'm leaving this ticket open, as I didn't yet guarantee the co-existency of the success rule version. Also some of the other operators might have to be updatde.