User Tools

Site Tools


solver:model_which_worked_with_an_old_version_of_conopt_fails_after_upgrading_to_the_current_version

Model, which worked with an old version of CONOPT, fails after upgrading to the current version

Q: I discovered that my model runs fine under the old GAMS 22.2, but won't run under the new 22.5 Are there 22.5 defaults I should set to make it run?

Here is the log from 22.2:

   
     C O N O P T 3   x86/MS Windows version 3.14P-016-057
     Copyright (C)   ARKI Consulting and Development A/S
                     Bagsvaerdvej 246 A
                     DK-2880 Bagsvaerd, Denmark

  Using default options.

  Reading Data

    Iter Phase Ninf   Infeasibility   RGmax    NSB   Step InItr MX OK
       0   0        6.9253942619E+07 (Input point)
                                 Pre-triangular equations:        4
                                 Post-triangular equations:      36
       1   0        1.3685290592E+05 (After pre-processing)
       2   0        2.0347053587E+01 (After scaling)
      11   1    12  7.6610890193E+00 8.8E-01     9 3.4E-01    6 F  F
      21   1     6  5.9949884664E-01 5.8E+01     6 0.0E+00      T  T
      26   1     3  3.1722986066E-01 1.0E+00     3 3.4E-01    2 F  F
      31   2     2  2.6599540328E-02 1.8E-03     5 1.0E+00    1 T  T
      36   2     2  3.4630766574E-03 3.1E-05     5 1.0E+00    1 T  T

  ** Feasible solution. Value of objective =   -52536.8756047

and with 22.5:

   
     C O N O P T 3   x86/MS Windows version 3.14R-017-061
     Copyright (C)   ARKI Consulting and Development A/S
                     Bagsvaerdvej 246 A
                     DK-2880 Bagsvaerd, Denmark

  Using default options.

  Reading Data

    Iter Phase Ninf   Infeasibility   RGmax    NSB   Step InItr MX OK
       0   0        6.9253942619E+07 (Input point)
                                 Pre-triangular equations:        4
                                 Post-triangular equations:      36
       1   0        1.3685290592E+05 (After pre-processing)
       2   0        2.0347053587E+01 (After scaling)
      11   1    12  7.6610890193E+00 8.8E-01     9 3.4E-01    6 F  F
      21   1     6  5.9949884664E-01 5.8E+01     6 0.0E+00      T  T
      26   1     3  3.1722986066E-01 1.0E+00     3 3.4E-01    2 F  F
      31   2     2  2.6599540328E-02 1.8E-03     5 1.0E+00    1 T  T
      34   2     2  1.1796589723E-02 3.5E-04     5

  ** Infeasible solution. A variable has reached 'Infinity'.
     Largest legal value (Rtmaxv) is 1.00E+10

The lst file will show, which variable reached 'Infinity'. Search for something similar to:

  ** Infeasible solution. A variable has reached 'Infinity'.
     Largest legal value (Rtmaxv) is 1.00E+10

     The allowable range can be changed with option:

     Rtmaxv=x.xxe+xx

     EV_v: The variable is unbounded 

and put a high upper bound on that variable (in this case EV_v).
Run the model with this bound and check the solution to make sure that bound wasn't active at the solution. If so, you have to increase the upper bound and run the model again.

Here is a comment from Arne Drud on that issue: While trying to find a feasible solution CONOPT ignores the objective function and sometimes it moves in 'crazy' directions and a variable can become very large. If CONOPT keeps the variable at the large value and continue it is likely to get in serious trouble, and I do not know what to do in this situation and this is the reason I stop. The limit for 'very large', rtmaxv, has been modified in 3.14r and it is probably the cause of the changed behavior.

solver/model_which_worked_with_an_old_version_of_conopt_fails_after_upgrading_to_the_current_version.txt · Last modified: 2007/10/11 08:01 (external edit)