[Scip] Binpacking example

[Felipe Serrano]

Hi Emily

Welcome to the SCIP mailing list!

To answer your first question: the binpacking example is an extension of
SCIP, meaning that it just adds a new reader, a new constraint, a new
branching rule and a new pricer.
Therefore, it works just like the normal SCIP. So after you compile you
can enter to SCIP's interactive shell via
bin/binpacking in Binpacking directory
>From there you can read a binpacking problem instance via
read data/u100_00.bpa   (this uses the new reader)
and finally you can solve it using
optimize    (and this uses the pricer, the branching rules, and the
constraint)
You can further see this by displaying information about the solving process:
display statistics
There you can find how much time (and other relevant information) was
spent in pricing, branching, etc

So, the "SCIPsolve(scip)" is hidden in:
SCIP_CALL( SCIPprocessShellArguments(scip, argc, argv, defaultsetname) );

By the way, you can solve directly a particular instance by calling
bin/binpacking -f data/u20_00.bpa


Regarding your second question: SCIP solves the LP relaxation of the
current integer problem. The dual values are the ones obtained by the
solution of the LP

Best,
Felipe

> Dear all,
>
> I have just started using SCIP. I want to implement a B&P algorithm in
> C++.
> I am looking at the Bin Packing example as template. As I think I will
> need
> to have branching rule, pricer and a conshdlr as well.
> I am, however, confused with some point and cant figure out how it works.
> I
> was hoping I can get help here.
>
> First, I cant find where in the example the master problem is being
> solved,
> i.e. I dont see SCIPsolve(scip) anywhere in the code! I am totaly lost
> about it. I was expecting to use it as in the VRP example after creating
> scip and including pricer(and in this case branch rule and conshdlr).
>
> Second, in the master problem, all the variables are given as binary(not
> continuous) and then we are trying to get the dual values of the master
> problem, isnt this wrong? I cant get how this works.
>
> Thanks in advance,
>
> Emily
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>