Re: [eclipse-clp-users] Questions about mixed integer linear programming using eplex

From: Kish Shen <kisshen_at_cisco.com>
Date: Thu, 15 May 2008 19:13:14 +0100
Hi Kim,

Kim Lai wrote:
> hi, 2 questions,
>
> 1. when I tried to use to solve Mix Integer linear program,
> for each integer variables, It will warn as follows
> "Eplex warning: Imposing integer bounds on variable(s) _15437 for 
> eplex instance eplex does not impose integer type"
> Is there any ways to stop or ignore this type of warnings ?
This is because you used eplex's (::)/2 and specified the bounds with 
integers, something like:

eplex: (V :: 0..1),

You get a warning because with other constraint solvers in ECLiPSe, e.g. 
ic, if you give integer bounds with (::)/2, this will also constrain the 
variable(s) to take only integer values. This is not the case with eplex.

You can avoid these warning messages by not using integer bounds (e.g. 
V:: 0.0..1.0) , or use ($::)/2 instead.

>
> 2. As I put constraints using Big-M method for each pair of 
> disjunctive constraints,
> with more and more constraints, the following errors will occur many 
> times.
> "Clp6002E 1 bad bound pairs or bad objectives were found - first at C98"
> Although at the end, I still got an answer from eplex.
> But will these errors make the result solved by eplex incorrect ?
> Or not a optimal solution ?
>
These messages comes from Clp -- the COIN linear solver you are using 
with eplex. You need to check with the COIN people to see what these 
messages mean. You can check up on COIN's mailing lists at:

http://list.coin-or.org/mailman/listinfo/

I assume that the solution is correct and optimal -- eplex checks with 
the solver to see if the call to solve the problem returns without 
errors, and also check if the solution is optimal or not.

If you find that the solution is incorrect, please let me know. This 
likely reflects problems with the solver, and I will try to reproduce it 
(and make sure it is not an eplex problem), and report it to the people 
maintaining the solver (CLP/CBC in your case).  I do know of some 
problems with the binary distribution of eplex with CLP/CBC, and they 
seem to be due to us using a relatively old gcc (3.3) to compile our 
code. Apparently these problems are not reproduced when compiling with 
gcc 4. Unfortunately I have had no time to investigate this further yet.

Cheers,

Kish


> thanks
>
> -- 
> ....Best Regards
> by Kim Lai, 賴廣甫
> Welcome to visit http://kimklai.blogspot.com
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft 
> Defy all challenges. Microsoft(R) Visual Studio 2008. 
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> ------------------------------------------------------------------------
>
> _______________________________________________
> ECLiPSe-CLP-Users mailing list
> ECLiPSe-CLP-Users_at_lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/eclipse-clp-users
>   
Received on Thu May 15 2008 - 11:13:48 CEST

This archive was generated by hypermail 2.2.0 : Thu Feb 02 2012 - 02:31:58 CET