optimal with unscaled infeasibilities
Iglesias Lopez, Araceli
My model is optimal with unscaled infeasibilities in CPLEX 12.1.0 but optimal CPLEX 8.0.0 as you can see. What can I do?
|
|
|
|
|
|
|||||||||||
|
IBM ILOG AMPL 12.1.0 (5724-Y45) AMPL Version 20090327 (x86_win32) |
|
||||||||||||||
|
|
|
|
|
|
|
|
|
|
|||||||
|
|
CPLEX 12.1.0: timing=1 |
|
|
|
|
|
|||||||||
|
|
prestats=1 |
|
|
|
|
|
|
|
|
||||||
|
|
mipdisplay=4 |
|
|
|
|
|
|||||||||
|
|
display=1 |
|
|
|
|
|
|
|
|
||||||
|
|
absmipgap=1e-4 |
|
|
|
|
|
|||||||||
|
|
mipgap=1e-4 |
|
|
|
|
|
|
|
|
||||||
|
|
time=1200 |
|
|
|
|
|
|
|
|
||||||
|
|
integrality=1e-6 |
|
|
|
|
|
|||||||||
|
|
startalgorithm=6 |
|
|
|
|
|
|||||||||
|
|
mipemphasis=1 |
|
|
|
|
|
|||||||||
|
|
mipalgorithm=1 |
|
|
|
|
|
|||||||||
|
|
mipcrossover=2 |
|
|
|
|
|
|||||||||
|
|
varselect=4 |
|
|
|
|
|
|
|
|
||||||
|
|
MIP Presolve eliminated 3935 rows and 5253 columns. |
|
|
||||||||||||
|
|
MIP Presolve modified 3106 coefficients. |
|
|
|
|
||||||||||
|
|
Aggregator did 33425 substitutions. |
|
|
|
|
||||||||||
|
|
Reduced MIP has 8940 rows, 15382 columns, and 96382 nonzeros. |
|
|
||||||||||||
|
|
Reduced MIP has 31 binaries, 0 generals, 0 SOSs, and 0 indicators. |
|
|
||||||||||||
|
|
Presolve time = 0.91 sec. |
|
|
|
|
||||||||||
|
|
MIP emphasis: integer feasibility. |
|
|
|
|
||||||||||
|
|
MIP search method: dynamic search. |
|
|
|
|
||||||||||
|
|
Parallel mode: deterministic, using up to 2 threads. |
|
|
|
|||||||||||
|
|
Initializing dual steep norms . . . |
|
|
|
|
||||||||||
|
|
|
|
|
|
|
|
|
|
|
||||||
|
|
Nodes Cuts/ |
|
|
|
|||||||||||
|
|
Node Left Objective IInf Best Integer Best Node ItCnt Gap |
|
|||||||||||||
|
|
|
|
|
|
|
|
|
||||||||
|
|
0 0 332251.0836 31 332251.0836 30480 |
|
|||||||||||||
|
|
* 0+ 0 331341.6153 332251.0836 30480 0.27% |
|
|||||||||||||
|
|
0 0 332241.2743 29 331341.6153 Cuts: 39 32325 0.27% |
|
|||||||||||||
|
|
* 0+ 0 332161.7224 332241.2743 32325 0.02% |
|
|||||||||||||
|
|
0 0 332226.3792 31 332161.7224 Flowcuts: 20 34067 0.02% |
|
|||||||||||||
|
|
* 0+ 0 332190.9978 332226.3792 34067 0.01% |
|
|||||||||||||
|
|
0 0 332217.1010 26 332190.9978 Flowcuts: 11 34932 0.01% |
|
|||||||||||||
|
|
|
|
|
|
|
|
|
||||||||
|
|
Flow cuts applied: 50 |
|
|
|
|
|
|||||||||
|
|
Gomory fractional cuts applied: 6 |
|
|
|
|
||||||||||
|
|
|
|
|
|
|
|
|
||||||||
|
|
Root node processing (before b&c): |
|
|
|
|
||||||||||
|
|
Real time = 112.02 |
|
|
|
|
||||||||||
|
|
Parallel b&c, 2 threads: |
|
|
|
|
|
|||||||||
|
|
Real time = 0.00 |
|
|
|
|
||||||||||
|
|
Sync time (average) = 0.00 |
|
|
|
|
||||||||||
|
|
Wait time (average) = 0.00 |
|
|
|
|
||||||||||
|
|
------- |
|
|
|
|
|
|||||||||
|
|
Total (root+branch&cut) = 112.02 sec. |
|
|
|
|
||||||||||
|
|
Using devex. |
|
|
|
|
|
|
||||||||
|
|
|
|
|
|
|
|
|
||||||||
|
|
Iteration log . . . |
|
|
|
|
|
|||||||||
|
|
Iteration: 1 Scaled infeas = 0.000171 |
|
|
|
|||||||||||
|
|
Iteration: 140 Objective = 332190.995525 |
|
|
|
|||||||||||
|
|
Iteration: 928 Objective = 332190.995927 |
|
|
|
|||||||||||
|
|
Iteration: 1138 Objective = 332191.693179 |
|
|
|
|||||||||||
|
|
Iteration: 1558 Objective = 332192.271875 |
|
|
|
|||||||||||
|
|
Iteration: 2048 Objective = 332196.217383 |
|
|
|
|||||||||||
|
|
Iteration: 2396 Objective = 332205.917854 |
|
|
|
|||||||||||
|
|
Iteration: 2657 Objective = 332241.923135 |
|
|
|
|||||||||||
|
|
Iteration: 2978 Objective = 332250.878506 |
|
|
|
|||||||||||
|
|
Elapsed time = 12.47 sec. (3000 iterations). |
|
|
|
|||||||||||
|
|
Removing shift (3). |
|
|
|
|
|
|||||||||
|
|
Switched to devex. |
|
|
|
|
|
|||||||||
|
|
Iteration: 3198 Infeasibility = 18.864545 |
|
|
|
|||||||||||
|
|
Switched to devex. |
|
|
|
|
|
|||||||||
|
|
|
|
|
|
|
|
|
||||||||
|
|
Times (seconds): |
|
|
|
|
|
|||||||||
|
|
Input = 0.374 |
|
|
|
|
|
|||||||||
|
|
Solve = 112.936 |
|
|
|
|
|
|||||||||
|
|
Output = 15.062 |
|
|
|
|
|
|||||||||
|
|
CPLEX solution status 5 with fixed integers: |
|
|
|
|||||||||||
|
|
|
optimal with unscaled infeasibilities |
|
|
|
||||||||||
|
AMPL Version 20010215 Win32 |
|
|||||||
|
CPLEX 8.0.0: timing=1 |
|
|
||||||
|
prestats=1 |
|
|
|
|||||
|
mipdisplay=4 |
|
|
||||||
|
display=1 |
|
|
|
|||||
|
absmipgap=1e-4 |
|
|
||||||
|
mipgap=1e-4 |
|
|
|
|||||
|
time=1200 |
|
|
|
|||||
|
integrality=1e-6 |
|
|
||||||
|
startalgorithm=6 |
|
|
||||||
|
mipemphasis=1 |
|
|
||||||
|
mipalgorithm=1 |
|
|
||||||
|
mipcrossover=2 |
|
|
||||||
|
varselect=4 |
|
|
|
|||||
|
MIP Presolve eliminated 3927 rows and 5307 columns. |
|
|||||||
|
MIP Presolve modified 4322 coefficients. |
|
|||||||
|
Aggregator did 33370 substitutions. |
|
|||||||
|
Reduced MIP has 9003 rows, 15383 columns, and 96302 nonzeros. |
|
|||||||
|
Presolve time = 3.14 sec. |
|
|||||||
|
MIP emphasis: integer feasibility |
|
|||||||
|
Nodes Cuts/ |
|
|
||||||
|
Node Left Objective IInf Best Integer Best Node ItCnt Gap |
||||||||
|
|
|
|
|
|
|
|||
|
0 0 332251.1095 31 332251.1095 34129 |
||||||||
|
* 0+ 0 0 331212.7402 332251.1095 34129 0.31% |
||||||||
|
332237.5432 25 331212.7402 Cuts: 54 36435 0.31% |
||||||||
|
* 0+ 0 0 332180.7057 332237.5432 36435 0.02% |
||||||||
|
332214.7929 30 332180.7057 Cuts: 35 37803 0.01% |
||||||||
|
* 0+ 0 0 332190.9978 332214.7929 37803 0.01% |
||||||||
|
|
|
|
|
|
|
|||
|
Flow cuts applied: 56 |
|
|
|
|
||||
|
Gomory fractional cuts applied: 33 |
|
|
|
|||||
|
Using devex. |
|
|
|
|
|
|||
|
|
|
|
|
|
|
|||
|
Iteration log . . . |
|
|
|
|
||||
|
Iteration: 1 Scaled infeas = 0.000181 |
|
|
||||||
|
Iteration: 584 Scaled infeas = 0.000034 |
|
|
||||||
|
Iteration: 1460 Objective = 332190.997743 |
|
|
||||||
|
Elapsed time = 11.59 sec. (2000 iterations) |
|
|
||||||
|
Iteration: 2288 Objective = 332190.998142 |
|
|
||||||
|
Iteration: 3095 Objective = 332190.998142 |
|
|
||||||
|
Iteration: 3636 Objective = 332191.003646 |
|
|
||||||
|
Iteration: 4317 Objective = 332191.004134 |
|
|
||||||
|
Iteration: 4829 Objective = 332191.004149 |
|
|
||||||
|
Elapsed time = 25.50 sec. (5000 iterations) |
|
|
||||||
|
Iteration: 5270 Objective = 332196.412187 |
|
|
||||||
|
Iteration: 5612 Objective = 332222.785130 |
|
|
||||||
|
Iteration: 5904 Objective = 332249.655098 |
|
|
||||||
|
Elapsed time = 41.73 sec. (6000 iterations) |
|
|
||||||
|
Iteration: 6197 Objective = 332251.089589 |
|
|
||||||
|
Removing shift (5). |
|
|
|
|
||||
|
Iteration: 6214 Infeasibility = 18.849834 |
|
|
||||||
|
Iteration: 6552 Infeasibility = 0.966292 |
|
|
||||||
|
Iteration: 6792 Objective = 332023.713057 |
|
|
||||||
|
Elapsed time = 56.66 sec. (7000 iterations) |
|
|
||||||
|
Iteration: 7090 Objective = 332168.454168 |
|
|
||||||
|
Iteration: 7383 Objective = 332186.807481 |
|
|
||||||
|
Iteration: 7682 Objective = 332190.097089 |
|
|
||||||
|
|
|
|
|
|
|
|||
|
Times (seconds): |
|
|
|
|
||||
|
Input = 1.968 |
|
|
|
|
||||
|
Solve = 404.373 |
|
|
|
|
||||
|
Output = 72.14 |
|
|
|
|
||||
|
|
|
|
|
|
|
|||
|
CPLEX 8.0.0: optimal integer solution within mipgap or absmipgap; objective 332190.9978 |
||||||||
|
37803 MIP simplex iterations |
|
|
|
|||||
|
0 branch-and-bound nodes |
|
|
|
|
||||
|
Tried aggregator 2 times |
|
|
|
|
||||
Araceli
+Si ha recibido este correo electrónico por error, le informamos que puede contener información confidencial y que está prohibido su uso. Le rogamos lo comunique a su remitente y lo elimine.
Gracias por su colaboración.
If you receive this e-mail by error, please note that it may contain confidential information,
therefore, the use of this information is strictly forbidden.
Please inform the sender of the error and delete the information received.
Thank you for your collaboration.
Robert Fourer
Probably there were unscaled infeasibilities in the result from CPLEX 8 as well, but that condition was not reflected in the termination message. CPLEX scales problems before solving them, to promote faster and more accurate solutions; then it unscales the results before reporting them back to AMPL. The "unscaled infeasibilities" message indicates that CPLEX successfully solved the scaled problem to within its numerical tolerances for feasibility, but that after the variables were unscaled some of the constraints were no longer satisfied to within the tolerances.
Usually unscaled infeasibilities are the result of some coefficients in the problem being much larger in magnitudes than others. If this is the case in your situation, then it may be worth thinking about how you could modify the formulation to reduce the variation in coefficient magnitudes. In any case, if you are curious as to where the unscaled infeasibilities occur in your problem, you can use AMPL's display command to look for constraints that have a significant negative slack.
Bob Fourer
From: am...@googlegroups.com [mailto:am...@googlegroups.com]
On
Behalf Of Iglesias
Lopez, Araceli [araceli....@cepsa.com]
Sent: Thursday, February 18, 2010 5:22 AM
To: am...@googlegroups.com
Subject: [AMPL 3316] optimal with unscaled infeasibilities
--
You received this message because you are subscribed to the Google Groups
"AMPL Modeling Language" group.
To post to this group, send email to am...@googlegroups.com.
To unsubscribe from this group, send email to ampl+uns...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/ampl?hl=en.