Originally posted by: SystemAdmin

[shadi said:]

hi
im using a column generation in order to find a set o duties for our busses

i use a cp model in order to find all the duties for each trip, (for my specific problem i find about 920000 duties)  after that i choose i minimum duties which cover all my trips, using master model,
when finish the slave phase the opl show me an error, i think that the error it become form the size of the tupleset (in this case 920000), is there is a limitation in the size of the tupleset or somthing else?

thank you
Kshieboun Shadi
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[alain.chabrier said:]

There is no limit on sets size, only limits on the memory in your computer.

Alain
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[shadi said:]

Dear Alain

when i added a new condition as following:

maximumPairings=500;
colCnt=0;
while (cpConfig.cp.next() && colCnt<maximumPairings) {<br />...
colCnt++;
}

in  order to decrease  the number of the duties  in the slave model, i achieve about 250000 duties, then the master model gives a solution.

i will check this again, becuase the slave model take a lot of time (more than 5 hours) i dont know the reason (i hope that the computer memory is the problem)

thank you
Kshieboun Shadi
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[alain.chabrier said:]

Hello,

trying to be clearer : there is no technical limitation on the [b]modeling[/b] side. Of course, the bigger you make the master model (by creating more columns from the slave model) the harder it will be to find a solution.

Seems you are doing column generation. May be you should generate much fewer columns, save the master problem and use dual values ot generate better columns in next iterations.

Alain
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[shadi said:]

Dear Alain
that what im trying to do to generate much fewer columns,
im trying to add more constraints in my slave in order to get fewer columns.

thank you
Kshieboun Shadi
#CPLEXOptimizers
#DecisionOptimization