Topic
  • 4 replies
  • Latest Post - ‏2008-10-30T19:34:30Z by SystemAdmin
SystemAdmin
SystemAdmin
151 Posts

Pinned topic CP and CPLEX under the same ODM application

‏2008-10-14T23:58:16Z |

[thiago.serra said:]

Hi,

I've been facing some trouble with embedding CP and CPLEX together in the same ODM application.


I have a .mod file with the "main" body, which only calls one model (CP or CPLEX) at a time.

If I put this on ODM, the solving progress window does not show the objective and goal functions progress (even if they are found and shown on the screen) but keep saying that data is being initialized.

For the second model called, the old goal is still shown. If I stop ODM by hand, it says that no solution was found (even if engine log says the opposite).

If processing completes after a while, it says a solution were found, but does not show it on ODM.


Realizing that one of the models (the second one) should be together with the "main" function, so that a solution would be browsed after processing, I tried putting it together.

But now I have a problem even without ODM, since I'm declaring in the mod file to use one engine, but I call another one in the "main" body.

Does anybody has a suggestion for achieving what I want?


TIA,
Updated on 2008-10-30T19:34:30Z at 2008-10-30T19:34:30Z by SystemAdmin
  • SystemAdmin
    SystemAdmin
    151 Posts

    Re: CP and CPLEX under the same ODM application

    ‏2008-10-15T21:41:10Z  

    [thiago.serra said:]

    Hi again,

    Since the last post I did some progress with the manual at the section "ILOG Optimization Decision Manager 3.0.1 > ODM User's Manual > Advanced usage > Multiple models in an ODM application > What is multimodel architecture? > Architecture of a multimodel ODM application".
    Still, I can't figure out how to combine CP and CPLEX on ODM.
    Does anybody did that before?

    TIA,
  • SystemAdmin
    SystemAdmin
    151 Posts

    Re: CP and CPLEX under the same ODM application

    ‏2008-10-16T00:37:41Z  

    [thiago.serra said:]

    Hi again,

    Since the last post I did some progress with the manual at the section "ILOG Optimization Decision Manager 3.0.1 > ODM User's Manual > Advanced usage > Multiple models in an ODM application > What is multimodel architecture? > Architecture of a multimodel ODM application".
    Still, I can't figure out how to combine CP and CPLEX on ODM.
    Does anybody did that before?

    TIA,

    [alain.chabrier said:]

    Hi,

    did you look at :
    C:\ILOG\ODM31\examples\Advanced\MultiModel\vellino

    It uses CPO to generate possible bins and then CPLEX to select the one to use, in a unique ODM application.

    Alain
  • SystemAdmin
    SystemAdmin
    151 Posts

    Re: CP and CPLEX under the same ODM application

    ‏2008-10-17T03:06:04Z  

    [alain.chabrier said:]

    Hi,

    did you look at :
    C:\ILOG\ODM31\examples\Advanced\MultiModel\vellino

    It uses CPO to generate possible bins and then CPLEX to select the one to use, in a unique ODM application.

    Alain

    [thiago.serra said:]

    I've missed that one. Thanks a lot!

    If possible, I'd like to ask you another thing:

    I'm solving two problems and showing them on Solve Progress window. I mean, I use thisOdm.generate(...) and thisOdm.solve(...) twice.
    Is it possible to reset or hidden the Cumulative Objective graphic?

    I mean this because the objective function of both problems are very different, so that I have a gap transition that is meaningless, and goals from both models are much more important than the objective function itself.

    TIA,

  • SystemAdmin
    SystemAdmin
    151 Posts

    Re: CP and CPLEX under the same ODM application

    ‏2008-10-30T19:34:30Z  

    [thiago.serra said:]

    I've missed that one. Thanks a lot!

    If possible, I'd like to ask you another thing:

    I'm solving two problems and showing them on Solve Progress window. I mean, I use thisOdm.generate(...) and thisOdm.solve(...) twice.
    Is it possible to reset or hidden the Cumulative Objective graphic?

    I mean this because the objective function of both problems are very different, so that I have a gap transition that is meaningless, and goals from both models are much more important than the objective function itself.

    TIA,

    [alain.chabrier said:]

    Unfortunately, I don't think this is possible currently.

    Alain