ryojitanabe / reproblems Goto Github PK

Hi, in the supplementary file, the input domain of x2 is [4, 50], which is inconsistent with the input domain written in your code [0.5, 50].

According to the RE21 formulation in the supp file,
the lower bound of RE21 for x2 and x3 should be sqrt(2)*a instead of sqrt(a).

That is, in the main.m file,
lowerBounds = [tmpVar sqrt(tmpVar) sqrt(tmpVar) tmpVar];
should be replace by
lowerBounds = [tmpVar sqrt(2)*tmpVar sqrt(2)*tmpVar tmpVar];

After this modification, the obtained results are consistent with the reference_points_RE21.dat file.

The values in reference_points_RE21.dat don't match what was shown in the Figure 2a in the paper.

In RE22, the lower bound of x2 is 0.
However, x2=0 makes two constraints g1 and g2 become -inf.
This happens when I use NSGA-II to solve the current matlab implementation of RE22,

It may be better to replace x2 in the denominators of two constraints by (x2+1e-6),
or to set a small number (e.g., 1e-6) as the lower bound of x2, instead of 0.

Hey @ryojitanabe. Thank you a lot for having made this collection of real analytical problems available.

There is an typo error for problem Re3-7-5. In constraint g11, the bar of sqrt should englobe the 1.575 * 1e8 value. You fixed it for example in the c implementation but the supplementary material pdf was not updated.

In the supplementary file and the source codes, the constraint on x4 is miscalculated as 11 <= x4 <= 20, but it should be 2 <= x4 <= 11.

Hi @ryojitanabe
Thanks for providing the source code of the RE problems in jMetal 4.5.

I have just adapted them to the last stable version (jMetal 5.10) in GitHub (https://github.com/jMetal/jMetal). The repository also includes the reference fronts in CSV format (https://github.com/jMetal/jMetal/tree/master/resources/referenceFrontsCSV).

Regards,

Antonio

Hey @ryojitanabe. Thank you for having coded the examples in different languages, it helped me a lot to debug my code which uses these examples.

There is a difference between the description of constraint g7 = 500000 - DWT in the supplementary material pdf and the way it is implemented for example in the c version or the python version. In this case, the constraint is g7 = 50000.0 -DWT. I took a look in the original implementation described in the original paper

Parsons, M. G., & Scott, R. L. (2004). Formulation of Multicriterion Design optimization Problems For Solution With Scalar Numerical Optimization Methods. Journal of Ship Research, 48(1), 61-76.

which corresponds to g7 = 500000 - DWT. Is there a reason for which you removed a 0 ?

Maybe another error of typo in the same problem. When I look at the equation S-100, I understand it as the fact that the annual_cargo variable is negative as annual_cargo = - CARGO_DWT * RTPA. Consequently, one user, when looking at the problem can think (it was my case) that f1 = annual_cost / annual_cargo is negative, which is not the case, as it is clarified by your implementation. The simplest will be to make the following correction: f3 = - annual_cargo = - CARGO_DWT * RTPA or -f3 = annual_cargo = cargo_dwt * RTPA.

Thank you again for having collected in the litterature this great set of real benchmarks !