#350 closed task (done)
Evaluate bias in certain permutation operators
Reported by: | abeham | Owned by: | |
---|---|---|---|
Priority: | medium | Milestone: | HeuristicLab 3.3.0 |
Component: | ZZZ OBSOLETE: Permutation | Version: | 3.2 |
Keywords: | Cc: |
Description
Several permutation modification operators are biased in their selection of breakpoints, as the following code example shows:
breakPoint1 = random.Next(result.Length - 1); breakPoint2 = random.Next(breakPoint1 + 1, result.Length);
Assuming N = result.Length - 1, the probability of selecting breakPoint1 = N-1 and breakPoint2 = N is higher than breakPoint1 = 0 and breakPoint2 = 1, namely 1/N vs (1/N)*(1/N-1)
The influence of this bias on the optimization should be evaluated.
Change History (6)
comment:1 Changed 16 years ago by abeham
- Status changed from new to assigned
comment:2 Changed 15 years ago by abeham
- Owner abeham deleted
- Status changed from assigned to new
comment:3 Changed 15 years ago by swagner
- Milestone changed from 3.2 to Iteration 0
Milestone 3.2 deleted
comment:4 Changed 14 years ago by abeham
- Resolution set to fixed
- Status changed from new to closed
Permutation operators have all been "reimplemented" using literature references. This ticket can be closed.
comment:5 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
first results with crossover operators show that the bias does not have negative effects, on the contrary, solution quality was better with bias.
Mutation operators have yet to be tested.