[Cbc] Need advices for properly using CBC
Karl St-Cyr
karl at patrikdufresne.com
Sun May 18 20:48:32 EDT 2014
The problem is not the feasibility of the problem but the time it takes to
solve it. 130 sec is way above what I would consider acceptable for the
implementation I want to do. As I stated before, a problem may be comprised
of 300+ of these dumps. Most of them will be solved in a fraction of a
second, but if every complex ones take 130 sec I will not be able to use
this implementation and will be forced revert to a simpler approach to the
problem that gives me lesser results.
I will keep looking for another solution.
On Sat, May 17, 2014 at 1:22 AM, Rod Frowd <rod at frowd.net> wrote:
> This problem seems to solve ok with cplex:
>
> Microsoft Windows [Version 6.1.7601]
> Copyright (c) 2009 Microsoft Corporation. All rights reserved.
>
> C:\Users\Rod Frowd>cd downloads
>
> C:\Users\Rod Frowd\Downloads>cplex
>
> Welcome to CPLEX Interactive Optimizer 12.1.0
> with Simplex, Mixed Integer & Barrier Optimizers
> 5724-Y48 (c) Copyright IBM Corp. 1988, 2009. All Rights Reserved.
> CPLEX is a registered trademark of IBM Corp.
>
> Type 'help' for a list of available commands.
> Type 'help' followed by a command name for more
> information on commands.
>
>
> CPLEX> r dump-5-20121111.lp
> Warning, line 33970: Name 'tolerance_0' does not exist.
> Warning, line 33971: Name 'tolerance_0' does not exist.
> Warning, line 33972: Name 'tolerance_1' does not exist.
> Warning, line 33973: Name 'tolerance_1' does not exist.
> Warning, line 33986: Name 'tolerance_4' does not exist.
> Warning, line 33987: Name 'tolerance_4' does not exist.
> Warning, line 33988: Name 'tolerance_5' does not exist.
> Warning, line 33989: Name 'tolerance_5' does not exist.
> Warning, line 33996: Name 'tolerance_7' does not exist.
> Warning, line 33997: Name 'tolerance_7' does not exist.
> Warning, line 36312: Name 'tolerance_0' does not exist.
> Warning, line 36312: Name 'tolerance_1' does not exist.
> Warning, line 36313: Name 'tolerance_4' does not exist.
> Warning, line 36313: Name 'tolerance_5' does not exist.
> Warning, line 36314: Name 'tolerance_7' does not exist.
> Problem 'dump-5-20121111.lp' read.
> Read time = 0.08 sec.
> CPLEX> o
> Tried aggregator 1 time.
> MIP Presolve eliminated 367 rows and 2385 columns.
> Aggregator did 5 substitutions.
> Reduced MIP has 3344 rows, 20712 columns, and 69531 nonzeros.
> Reduced MIP has 20708 binaries, 4 generals, 0 SOSs, and 0 indicators.
> Presolve time = 0.33 sec.
> Clique table members: 3322.
> MIP emphasis: balance optimality and feasibility.
> MIP search method: dynamic search.
> Parallel mode: deterministic, using up to 8 threads.
> Root relaxation solution time = 1.34 sec.
>
> Nodes Cuts/
> Node Left Objective IInf Best Integer Best Node ItCnt
> Gap
>
> 0 0 0.0000 856 0.0000 4698
> 0 0 0.0000 4 Fract: 2 4698
> 0 0 0.0000 9 Cuts: 2 4702
> 0 2 0.0000 9 0.0000 4702
> * 83 70 integral 0 5.0000 0.0000 58637
> 100.00%
> 100 80 infeasible 5.0000 0.0000 66047
> 100.00%
> * 128 80 integral 0 3.0000 0.0000 89612
> 100.00%
> 200 86 1.0000 4 3.0000 0.0000 126676
> 100.00%
> 300 104 1.0000 4 3.0000 0.0000 176476
> 100.00%
> 400 120 0.5000 59 3.0000 0.5000 225409
> 83.33%
> 500 87 0.0000 33 3.0000 0.5000 284694
> 83.33%
> 600 30 infeasible 3.0000 0.5000 345371
> 83.33%
> 700 24 infeasible 3.0000 0.6667 409641
> 77.78%
> 800 4 1.3333 100 3.0000 0.6667 439745
> 77.78%
>
> Zero-half cuts applied: 3
> Gomory fractional cuts applied: 5
>
> Root node processing (before b&c):
> Real time = 6.60
> Parallel b&c, 8 threads:
> Real time = 122.65
> Sync time (average) = 42.83
> Wait time (average) = 53.99
> -------
> Total (root+branch&cut) = 129.25 sec.
>
> Solution pool: 2 solutions saved.
>
> MIP - Integer optimal solution: Objective = 3.0000000000e+000
> Solution time = 129.65 sec. Iterations = 442232 Nodes = 823
>
> CPLEX>
>
> On 14/05/2014 5:14 AM, Karl St-Cyr wrote:
>
> Hello again,
>
>
> First of all, I would like to thanks the mailing list for the support. I
> realize I am quite new to this and your support has made the whole
> experience much more positive and has pointed me in the right direction.
>
>
> I realized I had some bugs in the creation of my linear problem and I
> hopefully fixed them. Even with the tips I got, the "new" problem is
> unsolvable in a decent amount of time. The linear problem take over 5 000
> seconds to solve. I've read the documentation and tried different
> combination of arguments without any luck. So, once again, I'm asking for
> some help.
>
>
> Here is an instance I have issues with.
>
>
> https://www.dropbox.com/sh/qh311l5gqt1mhcx/AABervHj6hcTmrIvyTFbsOxfa
>
>
>
>
>
> On Thu, May 8, 2014 at 8:39 AM, Karl St-Cyr <karl at patrikdufresne.com>wrote:
>
>> Hi,
>>
>> I'm trying to use CBC to solve my linear problem. After implementing a
>> version of BigM I manage to find a better solution but the time to solve it
>> is a bit too high for my needs. This is my first time using CBC and I want
>> to make sure I'm not doing something wrong or missing a way to speed it up.
>>
>> You'll find 3 dump examples I have problems with in the link bellow (take
>> note that 1 problem can be composed of 300+ of theses dumps and would need
>> to execute all of them in more or less 7 min, some of the dumps are also
>> taking close to no time).
>>
>> https://www.dropbox.com/sh/dff6h98qxhf7svy/AAC60y10SORx3b25P7bT2bAOa
>>
>> Thank you for your help
>> Karl
>>
>>
>
>
> --
> Karl St-Cyr
> Développeur logiciel
> Patrik Dufresne Service Logiciel
>
>
> _______________________________________________
> Cbc mailing listCbc at list.coin-or.orghttp://list.coin-or.org/mailman/listinfo/cbc
>
>
>
>
> ------------------------------
> <http://www.avast.com/>
>
> This email is free from viruses and malware because avast! Antivirus<http://www.avast.com/>protection is active.
>
>
--
Karl St-Cyr
Développeur logiciel
Patrik Dufresne Service Logiciel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20140518/b7153d2e/attachment.html>
More information about the Cbc
mailing list