[jORLib] Objective of the solution in JorLib

Joris Kinable jkinable at cs.cmu.edu
Mon Nov 12 17:52:12 EST 2018


Yes, in the branch-and-price implementation, the objective is assumed to be
integer. For many optimization problems, you can round all fractional
values. Alternatively, multiply your fractional values with a constant, say
100, prior to rounding to preserve the last 2 decimals.

I currently have very little time to develop jORLib, as most of my time is
spent on JGraphT (jgrapht.org).

Joris Kinable

On Sun, Nov 11, 2018 at 8:40 AM Ying Zhang <yingzhang8996 at gmail.com> wrote:

> Dear Kinable,
>
> Now I am using JorLib to solve a VRP. The column generation method works
> well. However, since the decision variables are integer in the master
> problem, we need to branch.
>
> In the Branch and Price method, do you imply the *objective *of the
> solution is always *integer*?
>
> See AbstractBranchAndPrice,
>
> /** Stores the objective of the best (integer) solution **/
> protected int objectiveIncumbentSolution;
>
> [image: image.png]
>
> If the objective of incumbent solution is 100.5, and if config.PRECISION=0.000001, then an error throws in method MathProgrammingUtil.doubleToInt
>
> Is the code still in maintenance? I see the latest commit is Sep, 2016. Thank you very much.
>
>
> Ying
>
> Tsinghua University
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/jorlib/attachments/20181112/ce42d2d9/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 70073 bytes
Desc: not available
URL: <http://list.coin-or.org/pipermail/jorlib/attachments/20181112/ce42d2d9/attachment-0001.png>


More information about the jORLib mailing list