I think this isn't the right comparison. ECJ and other systems generally take pains to be true to the underlying standard algorithms that they have implemented. This means that, assuming the toolkits have done their jobs right, solution quality should be essentially the same. The primary differences you'll see is algorithmically: whether GE is better than GP or ES is better than DE or whatnot.
There are plenty of comparisons of ECJ and other toolkits in other more relevant aspects, such as http://www.springerlink.com/content/j57p85k614ql8h77/
On May 2, 2012, at 4:29 PM, George Papaioannou wrote:
> Hi all,
> do we have anywhere any paper with benchmarks about the quality of the solution comparing ECJ and other toolkits or the speed of finding the best solution?