[Dip] bounds as explicit constraints?

[Matthew Galati]

CPM is the cutting plane method (branch and cut).

PC is price and cut.

Using those settings (below) you are randomly selecting the relaxation to be
1/3 of the original problem. I am not surprised it is much slower than
Cplex. This has very little chance of performing well and is only there for
the sake of illustration and testing.

If you know the problem has some structure, you should provide that
structure using the MILPBlock application.



To answer your other questions:

> I am wondering if Dip expects that all the bound on variables being
explicitly declared as constraints or?

No.


> Is there any specific strategy to deal with bigM model in DIP, in
particular?

Nothing specific. In fact, DIP does not have a presolver activated (this is
on Ted's ToDo List). So, a poorly formulated bigM model will likely perform
badly - since bigM is typically taken care of in the presolver.








On Fri, Sep 16, 2011 at 8:58 AM, Shahin Gelareh <shahin.gelareh at gmail.com>wrote:
The setting in param of MILP example project is:
I Assume it is Price and cut (PC) but I dont know what you mean
by CPM (Cutting plance only?)?

#{0,1}: use DECOMP for price and cut
doPriceCut    = 1

<snip>

# Name of MPS file for instance (assumed to be in DataDir).
Instance        = Scheduling

# Seed for the random number generator (for choosing rows).
RandomSeed   = 1

# Percentage of rows (randomly chosen) for relaxation.
RelaxPercent = 0.33



On Fri, Sep 16, 2011 at 2:42 PM, Matthew Galati <matthew.galati at gmail.com>wrote:
What algorithm are you using?

CPM or PC?






On Fri, Sep 16, 2011 at 6:08 AM, Matthew Galati <matthew.galati at gmail.com>wrote:

> Please send me the example.
>
> Sent from my iPhone
>
> On Sep 16, 2011, at 2:32 AM, Shahin Gelareh <shahin.gelareh at gmail.com>
> wrote:
>
> > Hi,
> >
> > I am wondering if Dip expects that all the bound on variables being
> explicitly declared as constraints or?
> > I have got a Big-M model which is solved fine on Cplex in 5 seconds for
> an small instance but when I feed it into the MILP example takes time for
> ever and my application also fails .
> > Adding explicit did not make things better but sill the question remains.
> > Is there any specific strategy to deal with bigM model in DIP, in
> particular?
> >
> > regards,
> > _______________________________________________
> > Dip mailing list
> > Dip at list.coin-or.org
> > http://list.coin-or.org/mailman/listinfo/dip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/dip/attachments/20110916/941d6532/attachment.html>