For small TSPs (eil51, eil76), both methods produce similar results. The IGA
gets good results (within
of the benchmarks) in less than one minute,
so does the GA. Here we only tried to separate each problem into two sub-groups
and the running time is the same for both methods. Fig 3 show the
best tours obtained by both approaches on the 76 city problem. Note that we got
a tour without crossings with the IGA.