C# conversion

[Stedak]

I am attempting to convert the tiny_gp.java to c#.

Are there any test cases available (i.e. fitness data with a know
result set)?

Thanks!

Steve K.

[rp...@essex.ac.uk]

We don't have test data of the type you mention. Because of the use of
random number generators, it is difficult to compare results across
implementations in different languages. So, unless you had exactly the
same random number generator implementation, there would be
differences in the output produced (even if we gave both generators
the same seed).

Sorry.

Riccardo

[nojhan]

However, there exists several benchmarks that can be used to validate
evolutionary computation, but most of them are problem-centered.
Hopefully, the CEC benchmarks is specifically designed to compare
evolutionary algorithms, thus being a nice tool to start
experimentations :
http://www3.ntu.edu.sg/home/EPNSugan/

Note that, in order to ensure a fair comparison, it is mandatory to
use statistical tools over data gathered from rigorous experiments.
Thus, you can use statistical tests to validate the fact that your own
implementation achieve similar performances than the one you target.

--
Johann Dréo
http://metah.nojhan.net
http://www.nojhan.net/pro

[Alexei Eleusis Díaz Vera]

If what you need is to run or reference tiny_gp in .Net  or mono, you can also use IKVM.NET. I'm using it to run ECJ in .Net and works fine.
Best wishes
Alexei

2008/9/22 Stedak <st...@theketchams.com>