Wednesday, April 8, 2009

Unjust benchmark: TSP MIP/Gurobi vs GA/Octave

I was asked to have a look at the Genetic Algorithm from this site: http://www.mathworks.com/matlabcentral/fileexchange/13680. Just to get an idea I benchmark here the TSP model eil51 from TSPLIB against GAMS/Gurobi. I used Octave instead of Matlab. The solution TSP tour is quite different.

  MIP/Gurobi GA/Octave
Source eil51.gms tsp_ga.m
Language GAMS Matlab
Solver Gurobi Octave
Options threads 4
cuts 2
heuristics 0.2
defaults
Objective 426.00
(Proven optimal)
437.23
Time (seconds) 214.255 1939.2

Warning: to a large extent this is not a fair comparison.
TSP Tour from GAMS/Gurobi:


TSP Tour from TSP_GA:


3 comments:

  1. Why do you call this an unfair comparison? I think the GA people do a lot of interesting work, but I have seen a zillion papers on GA for the TSP, most of which seem to believe that 50 node TSPs are difficult to solve to optimality. A reminder that optimality is not hard for this size problems seems completely fair.

    ReplyDelete
  2. I would not be surprised that a high-performance, parallel GA code with some tuning (to make things even) does a little better than what I showed here. Of course the nice thing of a MIP is that they provide bounds and in this case prove optimality.

    ReplyDelete
  3. You can put the code for do the first graphic in .gch?? Please

    ReplyDelete