[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Help-glpk] Entering Feasible Solutions and setting optimality toler

From: Brady Hunsaker
Subject: Re: [Help-glpk] Entering Feasible Solutions and setting optimality tolerance for MIPs
Date: Fri, 11 Mar 2005 12:21:16 -0500
User-agent: Mozilla Thunderbird 0.9 (X11/20041124)

Michael Hennebry wrote:
On Wed, 9 Mar 2005, David Stephenson wrote:

I'm just in the process of exploring how to use GLPK for MIPs and was
wondering if there is a simple way to do either of the following:

1.    Setting the optimality threshold for early termination.  (E.g.,
terminate when best recorded feasible is within 5% of optimal)

Yes, but it wasn't intended for that.
In the distributed version you can set the parameter,
but its upper bound is only .1%.
You'll have to edit the code to set a higher value.
You might get somewhat the same effect by adding a constant term
to the objective function.

2.    Specify a feasible integer solution for use in fathoming branches.

No, but you can get about the same result by adding a constraint.

Also, does GLPK implement any sort of cutting plane methodology?

Yes, but I don't remember how to use it.

The IOS (Integer Optimization Suite) part of GLPK has this capability. I don't believe it appears in the documentation yet. You can see the functions in the file glpios.h. I don't think there are specific cuts implemented in IOS, yet, however, so you'd have to implement them yourself for now.

I do believe that cuts are on the horizon for GLPK, though.


Brady Hunsaker
Assistant Professor
Industrial Engineering
University of Pittsburgh

reply via email to

[Prev in Thread] Current Thread [Next in Thread]