Forum Discussion

FICO's avatar
FICO
Community Manager
3 years ago

Solver (xpress) returned non-zero return code (3221225477) and EXCEPTION_ACCESS_VIOLATION

Has anyone run into issues with "ERROR: Solver (xpress) returned non-zero return code (3221225477)" and "ERROR within optimization The optimizer has encountered an unexpected exception: 0xc0000005 (EXCEPTION_ACCESS_VIOLATION)"?

I am not too familiar with Xpress, so I appreciate any advice.

Here is the output log:

ERROR: Solver (xpress) returned non-zero return code (3221225477)

ERROR: Solver log: FICO Xpress Solver 64bit v8.8.0 Nov 21 2019 (c) Copyright

   Fair Isaac Corporation 1983-2019. All rights reserved

    Optimizer v35.01.01   

   Reading Problem \tmpo9loh692.pyomo Problem Statistics

          76029 (     0 spare) rows 80380 (     0 spare) structural columns

       12670668 ( 2220102 spare) non-zero elements

   Global Statistics

           9099 entities       0 sets       0 set members

   FICO Xpress v8.8.0, Hyper, solve started 16:29:58, Sep 13, 2023 Heap

   usage: 194MB (peak 194MB, 2MB system) Minimizing MILP \tmpo9loh692.pyomo

   Original problem has:

        76029 rows       80380 cols    12670668 elements     9099 globals

   Presolved problem has:

        38413 rows       51984 cols     7744184 elements     6271 globals

   LP relaxation tightened Presolve finished in 39 seconds Heap usage: 418MB

   (peak 902MB, 2MB system)

   Coefficient range                   original                solved

     Coefficients  [min,max] : [ 1.00e-06, 1.00e+08] / [ 1.00e-06,

     2.00e+00] RHS and bounds [min,max] : [ 5.68e-14, 1.00e+08] / [

     3.08e-03, 1.68e+07] Objective     [min,max] : [ 1.00e+00, 1.00e+00] /

     [ 1.00e-03, 1.00e+08]

   Symmetric problem: generators: 26, support set: 231

    Number of orbits: 77, largest orbit: 3 Row orbits: 148, row support: 348

   Will try to keep branch and bound tree memory usage below 486.0GB Fixed

   231 (of 231) columns in 3 (of 3) subproblems

   Starting concurrent solve with dual, primal and barrier (98 threads)

 ***REMOVED FOR SPACE SAVING***

       -------- Concurrent statistics:

         Dual: 34209 simplex iterations, 30.29s

       Primal: 16367 simplex iterations, 30.30s

      Barrier: 23 barrier and 2136 simplex iterations, 62.46s

               Barrier used 98 threads 56 cores, L1\L2 cache: 32K\39424K

               Barrier used AVX support, crossover used 8 threads

   Optimal solution found

      Its        Obj Value     S  Ninf Nneg       Sum Inf Time

        1      65934801.78     P     0    0       .000000  103

   Barrier solved problem

     23 barrier and 1 simplex iterations in 103s

   Final objective                      : 6.593480177943809e+07

     Max primal violation     (abs/rel) : 8.481e-11 / 8.481e-11 Max dual

     violation       (abs/rel) : 7.105e-15 / 7.105e-15 Max complementarity

     viol. (abs/rel) :      0.0 /      0.0

   All values within tolerances

   Starting root cutting & heuristics

    Its Type   BestSoln   BestBound  Sols   Add   Del    Gap    GInf

    Time

   k        1666201569. 65934801.78     1                96.04%      0

   106 c        564036618.3 65934801.78     2                88.31%

   0   108 In vectored exception handler

   **********************************************************************

   ERROR within optimization The optimizer has encountered an unexpected

   exception:

    0xc0000005 (EXCEPTION_ACCESS_VIOLATION) at 0x00007FFDDFC4E899 in thread

    ID# 30064

   The following diagnostic information may be of use to the FICO Xpress

   support team

   Command string :

     mipoptimize

   **********************************************************************

Thanks in advance!

2 Replies

  • FICO's avatar
    FICO
    Community Manager

    It looks like the solver crashed, which would be a bug in the solver. How exactly are you invoking the solver? From the console? From Mosel? From an application? In another way? Can you share the model that triggers this problem?

    You are using version 8.8 which is rather old. Can you upgrade to a more recent version? There are chances that the bug is already fixed there.

  • FICO's avatar
    FICO
    Community Manager

    It seems you also submitted this problem to the support channel. Let's continue discussing there. Most important question is whether you can upgrade to a more recent version of Xpress and whether the issue persists.