help-glpk
[Top][All Lists]
Advanced

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

Re: [Help-glpk] Can I provide a pre-calculated lower bound as an input t


From: Heinrich Schuchardt
Subject: Re: [Help-glpk] Can I provide a pre-calculated lower bound as an input to my MIP?
Date: Mon, 6 Jan 2014 08:50:30 +0100 (CET)

Hello Nazmul,

when calling Glpk from Matlab you probably have no chance.

When working directly with the Glpk library you can provide a heuristic 
solution in the callback routine of the MIP solver.

Best regards

Heinrich Schuchardt

http://www.xypron.de


On 06.01.14 at 06:44, Nazmul Islam wrote:

> Hello,
> 
> 
> 
> I am solving a MIP (minimization problem) using GLPK. I am running the code
> 
> in MATLAB using YALMIP.
> 
> 
> 
> I have roughly 3000 integer variables in my code. I use --mipgap to get an
> 
> approximate solution in a reasonable time. From the particular instance of
> 
> my problem, I can pre-calculate a lower bound of my problem. Can I provide
> 
> this lower bound as an input to GLPK?
> 
> 
> 
> An example:
> 
> 
> 
> Let's assume that the optimal solution of my minimization problem cannot be
> 
> lower than a value of L.
> 
> 
> 
> In an usual setting, GLPK continuously updates the lower bounds {L1, L2,
> 
> ...} through linear relaxation and the upper bounds {U1, U2, ... } and
> 
> checks if the lower bound & upper bound becomes closer than the --mipgap
> 
> value. Can I do any of the following two options?
> 
> 
> 
> 1. I want GLPK to use max{L, L1} ; max{L, L2}, max{L, L3} as the lower
> 
> bounds. In other words, if my provided lower bound L is greater than the
> 
> currently found lower bound Li of GLPK, I want GLPK to use that value. This
> 
> will narrow the gap between the lower bound and upper bound of GLPK and
> 
> save execution time..
> 
> 
> 
> 2. I want GLPK to use L as the constant lower bound. In that case, GLPK
> 
> does not need to solve any linear relaxation. It can just find the feasible
> 
> solution and check with the lower bound that I provided.
> 
> 
> 
> Thanks,
> 
> 
> 
> Nazmul
> 
> _______________________________________________
> 
> Help-glpk mailing list
> 
> address@hidden
> 
> https://lists.gnu.org/mailman/listinfo/help-glpk



reply via email to

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