Fix MDE_pBX [sf#20] about pagmo

[darioizzo]

Originally posted by mmarcusx on 2013-04-23 10:00:54.

• Description has changed:

Diff:

--- old
+++ new
@@ -1 +1,3 @@
 Using MDE_pBX on Schwefel or Rosenbrock creates a weird behavior (fitness is jumping up and down). We have to debug the algorithm thoroughly and try to recreate the literature results.
+
+The algorithm runs also slower than expected.

[darioizzo]

Originally posted by darioizzo on 2013-04-27 17:30:33.

The code has been updated significantly (check the development branch) it now runs much faster and a number of bugs were fixed.

The algorithm performances are still poor compared to those reported in the literature. I contacted the original authors of the algorithm and asked them to send us the code used in their papers (thanks to Rainer Storn for contacting them).

Now we wait :) I will anyway continue to try and find a solution myself ....

[darioizzo]

Originally posted by darioizzo on 2013-05-06 20:00:15.

I received the matlab code from Suganthan (one of the author from the original paper), which I attach here for future reference.

While the performances are still not those quoted in the paper, I have now made sure PaGMO implementation matches that sent by the authors.

Major improvement in speed was also obtained.

Ticket closed. New code in development branch (will ship in V.1.1.5)

[darioizzo]

Originally posted by darioizzo on 2013-05-28 12:00:24.

• status: open --> closed
• assigned_to: Marcus Märtens --> Dario Izzo