wiki:ES_TSP

Version 4 (modified by abeham, 13 years ago) (diff)

--

(μ+λ)-Evolution Strategy (TSP)

Description

This workbench demonstrates how to realize a simple (μ+λ)-ES. It is not an optimized and tuned workbench, but rather shows how to build an ES and apply it on a problem. For the TSP problem mutation strength adjustment is not used.

Parameters

  • TSP
    • Instance = berlin52
    • Evaluation = Euclidean Rounded
  • Algorithm
    • Mu = 1
    • Lambda = 1
    • MaximumGenerations = 50,000
    • Mutation = Inversion

Implementation & Notes

Works with r60

The workbench uses the ThreadParallelEngine and can make use of a multicore machine if lambda is increased.

Results & Conclusions

Achieves a result in the 10% range to the global optimum.


Attachments (1)

Download all attachments as: .zip