[Bonmin-tickets] [Bonmin] #26: Wrong result using Bonmin with Ampl

Bonmin coin-trac at coin-or.org
Sun Jun 15 16:29:51 EDT 2008


#26: Wrong result using Bonmin with Ampl
--------------------------+-------------------------------------------------
Reporter:  michael.hecht  |       Owner:  pbonami
    Type:  defect         |      Status:  new    
Priority:  critical       |   Component:  Bonmin 
 Version:  stable/0.1     |    Keywords:         
--------------------------+-------------------------------------------------
 Using the following Ampl example with Bonmin gives the wrong result (all
 x[i] = -0.5). Using it with IpOpt gives the right result (all x[i] ~ 0).
 What's wrong?

 # schwefel.mod  OBR2-AN-5-10
 # Original AMPL coding by Elena Bobrovnikova (summer 1996 at Bell Labs).

 # Ref.: C. Jansson and O. Knueppel, "A Global Minimization Method:
 # the Multi-Dimensional Case", Technische Informatik III,
 # TU Hamburg-Hamburg, Jan. 1992, p. 103 (problem "whs25").

 # Schwefel function

 # Number of variables:  5
 # Number of constraints:  10
 # Objective separable convex
 # Simple bound constraints

 # The global minimum is Fs = 0, x = (0,0,0,0,0).

 set I := {1..5};

 var x{I} <= 0.4 >= -0.5 := 0.3;

 minimize Fs:
      sum {i in I} x[i]^2;

 option solver ipopt; # Choose BONMIN as the solver (assuming that
                       # bonmin is in your PATH)

    solve;             # Solve the model
    display {i in I} x[i];
    display Fs;

-- 
Ticket URL: <https://projects.coin-or.org/Bonmin/ticket/26>
Bonmin <http://projects.coin-or.org/Bonmin>
Basic Open-source Nonlinear Mixed INteger programming



More information about the Bonmin-tickets mailing list