Pulp is able to model, but CoinMP can't solve the problem

95 views
Skip to first unread message

JuanDa

unread,
Jul 13, 2011, 11:24:45 AM7/13/11
to pulp-or-discuss
Hi. I am working in an integer linear programming model to solve an
assignation problem in a transportation company. The model considers
the assignation of activities and resources in three layers: orders,
vehicles, trailers. When the model is running Pulp is able to model
the problem but when the solver is solving the problem (I am using
CoinMP), it seems that the solver stays in an infinite loop. The
results while the solver is solving the problem are shown next:

Modelation time: 8.69015698923 secs
Clp0006I 1 Obj 52.9
Clp0000I Optimal - objective value 52.9
Coin0511I After Postsolve, objective 52.9, infeasibilities - dual 0
(0), primal 0 (0)
Clp0032I Optimal objective 52.9 - 1 iterations time 0.002, Presolve
0.00
Coin0506I Presolve 2 (-2) rows, 4 (-3) columns and 4 (-6) elements
Clp0006I 0 Obj 51.9 Primal inf 2.1 (1) Dual inf 0.0099999 (1) w.o.
free dual inf (0)
Clp0006I 2 Obj 53
Clp0000I Optimal - objective value 53
Coin0511I After Postsolve, objective 53, infeasibilities - dual 0 (0),
primal 0 (0)
Clp0032I Optimal objective 53 - 2 iterations time 0.002, Presolve 0.00
Coin0506I Presolve 2 (-2) rows, 4 (-3) columns and 4 (-6) elements
Clp0006I 0 Obj 51.9 Primal inf 2.1 (1) Dual inf 0.0099999 (1) w.o.
free dual inf (0)
Clp0006I 2 Obj -1e+009
Clp0006I 2 Obj -5e+009
...

Cbc0010I After 44800 nodes, 10145 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (490.92 seconds)
Cbc0010I After 44900 nodes, 10156 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (491.36 seconds)
Cbc0010I After 45000 nodes, 10149 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (491.76 seconds)
Cbc0010I After 45100 nodes, 10151 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (492.22 seconds)
Cbc0010I After 45200 nodes, 10151 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (492.65 seconds)
Cbc0010I After 45300 nodes, 10172 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (493.19 seconds)
Cbc0010I After 45400 nodes, 10156 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (493.61 seconds)
Cbc0010I After 45500 nodes, 10160 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (494.03 seconds)
Cbc0010I After 45600 nodes, 10160 on tree, -8.68519e+007 best
solution, best possible -1.02063e+008 (494.55 seconds)

What do you think about this? I don't understand what is the meaning
of part of this information, Is the solver finding the solution? Do
you think a more powerful solver could reach the solution of this
problem (Gurobi, CPLEX)? Thank you very much for your info...

Stuart Mitchell

unread,
Jul 16, 2011, 6:56:57 AM7/16/11
to pulp-or...@googlegroups.com

Hmm im not and expert on how to read the cbc results. But if you press ctrl-c you should get an early termination. Also you can use the maxSecs parameter or adjust the bound gap.

Gurobi or cplex should also improve the solve time. Perhaps you can also improve your formulation.

> --
> You received this message because you are subscribed to the Google Groups "pulp-or-discuss" group.
> To post to this group, send email to pulp-or...@googlegroups.com.
> To unsubscribe from this group, send email to pulp-or-discu...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/pulp-or-discuss?hl=en.
>
Reply all
Reply to author
Forward
0 new messages