[SCIP] Help with Understanding SCIP Objective function values

[James.Gunning at csiro.au]

Natalia,

      FWIW, if the integer variable spans not too great a range, one possible idea is to write it a a sum of booleans (\sum 2^k x_k type thing). Then you

get a bunch of boolean quadratic terms, each of which can be relaxed using the standard Fortet linearisation (google away). Sometimes it works OK. Often

the relaxation isn't of great quality. If the dimensionality is not too high then SDP might be interesting. Others may have much better ideas.

Cheers

   James.


James Gunning
CSIRO
Clayton, Vic., Australia
________________________________
From: Scip <scip-bounces at zib.de> on behalf of Natalia Perina <nperina at princeton.edu>
Sent: Saturday, 15 April 2017 7:58 AM
To: scip at zib.de
Subject: [SCIP] Help with Understanding SCIP Objective function values

Hi there -

I want to use SCIP to implement a branch and price framework for an MIP problem that I am working on for my senior thesis. I have two questions.

First, I am trying to understand how to set the objective function and variables. How do you set which variables will go into the objective function? I have several variables in my problem but it is only the sum of one of them that I want to be maximized for the objective. How do I select which variables go into the objective? In the createVar() method, how do I use the objective value parameter? Is this supposed to the objective variables coefficient? And for variables that I don't want to include in the objective function - would one just set this to 0 or to null? This seems like there would be a simple explanation but I can't seem to find it in the documentation.

Second, I wanted to know if anyone had any advice for linearizing and objective function that is the sum of a integer variable multiplied by a binary variable. I understand that this is non linear and that scip does not support non linear objective functions but does anyone have advice on how I might be able to do this using constraints?

Thanks,
Natalia

--
Natalia Perina
Princeton University l 2017
nperina at princeton.edu<mailto:nperina at princeton.edu>
202 360 9536<tel:(202)%20360-9536>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20170415/11eb35e3/attachment.html>