Originally posted by: Stoepsel

Hello,

 

I don't have a lot of experience in working with CPLEX so please forgive any missing information or inaccurate descriptions. :)

I'm currently working on a quite small optimization problem using the IBM ILOG CPLEX Optimization Studio.

The problem has 24 decision variables and about 25 constraints. Compared to other problems this seems to be really small. The arrays however have a range of 34944.

Since I added a constraint I'm getting the "CPLEX Error 1001: out of memory"-error after about 40 minutes.

I'm guessing this is happening before the optimization itself starts since there are no messages in the engine protocoll.

In the task manager it says that "Java (TM) Platform SE Binary" is using most of the RAM. At some point the RAM usage of this process increases drastically until the optimization aborts.

I have already set the available memory to various different values and setting the memory emphasis to 1 with no results.

 

Can anyone help me?

Thank you in advance!

 

Andy

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Laci Ladanyi

About not having a log: make sure that logging is turned on (though it is on by default...). In java you need to use the cplex.setOut() method (http://www-01.ibm.com/support/knowledgecenter/SSSA5P_12.6.2/ilog.odms.cplex.help/CPLEX/Parameters/topics/SCRIND.html?lang=en) to change how logging is done.

As far as why cplex runs out of memory... hard to say without seeing the log :-). But if you have 24 general integer variables with that wide feasible region and you have 25 constraints then it is quite possible that there are very few integer feasible solutions and cplex just keeps branching eventually running out of memory. Can you share the problem instance with us? Just export the instance (preferably as a sav file) and attach it to this thread or email it to ladanyi at us dot ibm dot com.

 

Thanks,

--Laci

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Stoepsel

Hi Laci,

thanks for your reply!

Sorry english is not my first language. What do you mean by instance? The code?

After some adjustments I got the optimiziation running. However it stopped after 2 hours. 

Here is the enginge log:

Presolve has eliminated 212228 rows and 214787 columns...

Aggregator has done 145301 substitutions...

Aggregator has done 151401 substitutions...

Aggregator has done 157501 substitutions...

Aggregator has done 163601 substitutions...

Aggregator has done 169701 substitutions...

Aggregator has done 175801 substitutions...

Aggregator has done 181901 substitutions...

Aggregator has done 188001 substitutions...

Aggregator has done 194101 substitutions...

Aggregator has done 200201 substitutions...

Aggregator has done 206301 substitutions...

Aggregator has done 212401 substitutions...

Aggregator has done 218501 substitutions...

Aggregator has done 224601 substitutions...

Aggregator has done 230701 substitutions...

Aggregator has done 236801 substitutions...

Aggregator has done 242901 substitutions...

Tried aggregator 2 times.

MIP Presolve eliminated 701301 rows and 423168 columns.

MIP Presolve modified 34943 coefficients.

Aggregator did 296919 substitutions.

Reduced MIP has 1410483 rows, 1376542 columns, and 3483609 nonzeros.

Reduced MIP has 905974 binaries, 0 generals, 0 SOSs, and 591472 indicators.

Presolve time = 1160.39 sec. (638038.94 ticks)

Warning: MIP starts not constructed because of out-of-memory status.

 

Root node processing (before b&c):

  Real time             = 1168.52 sec. (638746.09 ticks)

Parallel b&c, 4 threads:

  Real time             =    0.00 sec. (0.00 ticks)

  Sync time (average)   =    0.00 sec.

  Wait time (average)   =    0.00 sec.

                          ------------

Total (root+branch&cut) = 1168.52 sec. (638746.09 ticks)

 

Is this any help?

 

Thank you

Andy

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Laci Ladanyi

Yes, this helps. Well, your problem certainly has more than 24 variables... :-). By "instance" I meant that problem itself that you are trying to solve. Can you post the code that you use to generate the problem?

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Stoepsel

I've sent you an e-mail with the code.

#CPLEXOptimizers
#DecisionOptimization