Decision Optimization

 View Only
  • 1.  subproblem error when solving mip with empty callback

    Posted Tue November 29, 2022 07:43 AM
    Edited by System Fri January 20, 2023 04:43 PM
    hi
    I have an empty branch callback and trying to solve seymour-disj_10 presolved version.
    https://miplib2010.zib.de/miplib2010/seymour-disj-10.php
    Error log is attached.
    Works ok with mip empahsis changed to 3.

    Version identifier: 12.10.0.0 | 2019-11-26 | 843d4de
    CPXPARAM_Preprocessing_Reduce 1
    CPXPARAM_Preprocessing_Linear 0
    CPXPARAM_Threads 32
    CPXPARAM_MIP_Strategy_File 3
    CPXPARAM_Emphasis_MIP 2
    CPXPARAM_MIP_Strategy_HeuristicFreq -1
    CPXPARAM_TimeLimit 3600
    Warning: Control callbacks may disable some MIP features.
    Tried aggregator 1 time.
    MIP Presolve eliminated 1 rows and 0 columns.
    MIP Presolve modified 2 coefficients.
    Reduced MIP has 4788 rows, 1030 columns, and 62118 nonzeros.
    Reduced MIP has 995 binaries, 35 generals, 0 SOSs, and 0 indicators.
    Presolve time = 0.03 sec. (19.05 ticks)
    Probing time = 0.01 sec. (1.24 ticks)
    Tried aggregator 1 time.
    Reduced MIP has 4788 rows, 1030 columns, and 62118 nonzeros.
    Reduced MIP has 995 binaries, 35 generals, 0 SOSs, and 0 indicators.
    Presolve time = 0.02 sec. (16.13 ticks)
    Probing time = 0.00 sec. (1.24 ticks)
    Clique table members: 216.
    MIP emphasis: optimality.
    MIP search method: traditional branch-and-cut.
    Parallel mode: deterministic, using up to 32 threads.
    Root relaxation solution time = 1.60 sec. (1178.27 ticks)

    Root node processing (before b&c):
    Real time = 1.69 sec. (1241.92 ticks)
    Parallel b&c, 32 threads:
    Real time = 0.00 sec. (0.00 ticks)
    Sync time (average) = 0.00 sec.
    Wait time (average) = 0.00 sec.
    ------------
    Total (root+branch&cut) = 1.69 sec. (1241.92 ticks)
    Exception in thread "main" ilog.cplex.CpxException: CPLEX Error 3019: Failure to solve MIP subproblem.

    at ilog.cplex.CplexI.CALL(CplexI.java:5204)
    at ilog.cplex.CplexI.setStatus(CplexI.java:6708)
    at ilog.cplex.CplexI.access$600(CplexI.java:190)
    at ilog.cplex.CplexI$SolveHandle.stop(CplexI.java:3162)
    at ilog.cplex.CplexI.solve(CplexI.java:3212)
    at ilog.cplex.CplexI.solve(CplexI.java:3204)
    at ilog.cplex.IloCplex.solve(IloCplex.java:13062)

    ------------------------------
    srinivas tamvada
    ------------------------------
    #DecisionOptimization


  • 2.  RE: subproblem error when solving mip with empty callback

    IBM Champion
    Posted Thu December 08, 2022 04:50 PM
    Something does not track here. According to the MIPLIB page, the seymour-disj-10 instance has 1209 binary variables and no general integer variables. I ran it (part way) in the interactive optimizer and confirmed that, but your log indicates 995 binary variables and 35 general integer variables. You said something about a "presolved version". Is that different from the one linked in the MIPLIB page and, if so, where can one find it?

    ------------------------------
    Paul Rubin
    Professor Emeritus
    Michigan State University
    ------------------------------



  • 3.  RE: subproblem error when solving mip with empty callback

    Posted Sun December 11, 2022 01:35 PM
    hi
    basically read the miplib file which is mps, and write out the .pre file. I also read the pre file and then saved it as a .sav file. See attached
     It does not happen for me when I change the emphasis to 3, so not sure if it will reproduce for you.

    ------------------------------
    srinivas tamvada
    ------------------------------



  • 4.  RE: subproblem error when solving mip with empty callback

    IBM Champion
    Posted Wed December 14, 2022 02:03 PM
    I tried this using the latest version of CPLEX (22.1.1), working with both the MPS file from the web site and the SAV file you provided. I used your parameter settings, with two exceptions: the thread limit (my system is too small for 32 threads); and the time limit (I went with 30 seconds). I included an empty (legacy) branch callback. No exceptions were thrown.

    ------------------------------
    Paul Rubin
    Professor Emeritus
    Michigan State University
    ------------------------------