MILPmodel" Solver did not provide marginals for model "

GAMS
1

Hello All,
I ran a MILP model in CPLEX/GAMS with a lot of binary variables and 4 integer variables.
The results are as follows:
How can I solve the infeasibility of the model.
Thanks

"…
Iteration: 528309 Dual objective = 318887.212370
Iteration: 531815 Dual objective = 318929.604829
Iteration: 534938 Dual objective = 318975.745926
Removing perturbation.
Using devex.

Primal simplex solved model.

Fixed MIP status(3): infeasible
Cplex Time: 189.61sec (det. 73479.65 ticks)
Final solve did not return an optimal solution.
Returning a primal only solution to GAMS (marginals all set to NA).

Solution satisfies tolerances.

MIP Solution: 1264291.458963 (294626 iterations, 18 nodes)
Best possible: 1251920.190266
Absolute gap: 12371.268697
Relative gap: 0.009785

— Restarting execution
— Untitled_18.gms(4952) 148 Mb
— Reading solution for model thesis16

*** Solver did not provide marginals for model thesis16

— Untitled_18.gms(4952) 333 Mb
— Executing after solve: elapsed 0:14:43.191
— Untitled_18.gms(4989) 334 Mb
*** Status: Normal completion
— Job Untitled_18.gms Stop 08/04/14 14:15:03 elapsed 0:14:43.384
"


Azadeh

On Wednesday, October 15, 2014 at 3:29:28 PM UTC-4, hossein wrote:


Dear all

After the identification of my model as the infeasible model by GAMS, Cplex tries to solve it again, which is undesirable to me, as follows:

Removing perturbation.

LP status(3): infeasible

Cplex Time: 0.78sec (det. 372.14 ticks)

Dual infeasible or unbounded. Switching to primal to aid diagnosis.

Tried aggregator 1 time.

LP Presolve eliminated 304797 rows and 302399 columns.

Aggregator did 964 substitutions.

Reduced LP has 1205 rows, 1717 columns, and 6116 nonzeros.

Presolve time = 0.25 sec. (92.91 ticks)

Using devex.

Iteration log . . .

Iteration: 1 Scaled infeas = 38.509142

Iteration: 158 Scaled infeas = 14.912938

LP status(3): infeasible

Cplex Time: 0.81sec (det. 408.69 ticks)

Model has been proven infeasible.



I tied to disable the ‘Switching to primal to aid diagnosis’ step by commands as follows:

putclose opts /‘rerun = NO’ /‘feasopt = NO’;

But noting was changed. Can anyone help me to remove this step?


\

To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/d/optout.

2

When the fixed MIP is reported to be infeasible this usually points to a poorly scaled model. So the best way to fix this is to properly scale you model. The GAMS/Conopt manual has a very nice section about scaling (http://gams.com/help/infocenter/topic/gams.doc.solver/conopt/index.html?cp=1_8_5_3#CONOPT_SCALING). You can also try to let Cplex do some internal “magic” by setting a few options like “numericalemphasis=1” and “scaind=1” that aims at improving the numerics of you model. Perhaps this helps. Moreover, you solve the fixed mip with concurrent LP (message “Primal simplex solved model.”) you might try to solve this with a different LP method (set option lpmethod explicitly).

Hope this helps,
Michael Bussieck - GAMSWorld Coordinator

On Tuesday, March 10, 2015 at 3:01:19 AM UTC-4, Azadeh Maroufmashat wrote:

Hello All,
I ran a MILP model in CPLEX/GAMS with a lot of binary variables and 4 integer variables.
The results are as follows:
How can I solve the infeasibility of the model.
Thanks

"…
Iteration: 528309 Dual objective = 318887.212370
Iteration: 531815 Dual objective = 318929.604829
Iteration: 534938 Dual objective = 318975.745926
Removing perturbation.
Using devex.

Primal simplex solved model.

Fixed MIP status(3): infeasible
Cplex Time: 189.61sec (det. 73479.65 ticks)
Final solve did not return an optimal solution.
Returning a primal only solution to GAMS (marginals all set to NA).

Solution satisfies tolerances.

MIP Solution: 1264291.458963 (294626 iterations, 18 nodes)
Best possible: 1251920.190266
Absolute gap: 12371.268697
Relative gap: 0.009785

— Restarting execution
— Untitled_18.gms(4952) 148 Mb
— Reading solution for model thesis16

*** Solver did not provide marginals for model thesis16

— Untitled_18.gms(4952) 333 Mb
— Executing after solve: elapsed 0:14:43.191
— Untitled_18.gms(4989) 334 Mb
*** Status: Normal completion
— Job Untitled_18.gms Stop 08/04/14 14:15:03 elapsed 0:14:43.384
"


Azadeh

On Wednesday, October 15, 2014 at 3:29:28 PM UTC-4, hossein wrote:


Dear all

After the identification of my model as the infeasible model by GAMS, Cplex tries to solve it again, which is undesirable to me, as follows:

Removing perturbation.

LP status(3): infeasible

Cplex Time: 0.78sec (det. 372.14 ticks)

Dual infeasible or unbounded. Switching to primal to aid diagnosis.

Tried aggregator 1 time.

LP Presolve eliminated 304797 rows and 302399 columns.

Aggregator did 964 substitutions.

Reduced LP has 1205 rows, 1717 columns, and 6116 nonzeros.

Presolve time = 0.25 sec. (92.91 ticks)

Using devex.

Iteration log . . .

Iteration: 1 Scaled infeas = 38.509142

Iteration: 158 Scaled infeas = 14.912938

LP status(3): infeasible

Cplex Time: 0.81sec (det. 408.69 ticks)

Model has been proven infeasible.



I tied to disable the ‘Switching to primal to aid diagnosis’ step by commands as follows:

putclose opts /‘rerun = NO’ /‘feasopt = NO’;

But noting was changed. Can anyone help me to remove this step?


\

To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/d/optout.