Originally posted by: Nomykan

Hi, 

i have a MIP (two stage stochastic) problem. i have modeled it in OPL and there are no errors in the program but the problem is it takes too much time to solve the problem as there are more than 40K binary variables. in order to give warm start to the CPLEX i want to solve the first stage separately and them use the results of the first stage as constant and parse it to the main model and solve the problem as a whole. any idea how to do this in OPL ? any help will be highly appreciated. i have gone through Alex posts 

• Change some data and solve again without regenerate
• Change some data and solve again with regenerate

 and other relevant posts but sill confused how to do it. Help will be appreciated. Thanks

Hi,

you could also have a look at https://www.ibm.com/developerworks/community/forums/html/topic?id=471cc451-9dc2-44d8-a0e9-5f146fb14765&ps=25

And

https://www.linkedin.com/pulse/making-decision-optimization-simple-alex-fleischer/

And

https://www.linkedin.com/pulse/how-opl-alex-fleischer/

regards

Originally posted by: Nomykan

Hi Alex, 

Thanks for your response. i have gone through the posts you provided and it helped me understand the problem more efficiently but now my question is if we have to just take the values of one decision variable from one model and one decision variable from a second model and give them to the main model. in this case the main model also have some data for other parameters. so we need to add 3 different data sources in the model or how we can do that? Actually all the three data belongs to one model but just to help solve model in less time i first need to solve the two sub models and then using the results of the sub models as constant and give it to the main model which will help reduce solving time. Any help in this case will be appreciated. Thanks  

Hi,

you can do that in the main part in OPL (flow control)

See also

https://www.ibm.com/developerworks/community/forums/html/topic?id=72359501-c169-4215-9dc0-1c87e653b584&ps=25

regards

Originally posted by: Nomykan

Hi Alex,

Thanks for your rightly guided directions through which i am very quickly learning Optimization through CPLEX. i got the main idea of how to use the results of one model and use as input to the next model. but now the problem is how i can use the results of one model whose decision variable is an array of tuples. this is now making problem for me. for example i have a tuple 

tuple edge { int i; int j; }

{edge} edges=...;

and then the decision variable X[edges][time];

and then this decision variable X is used as an input to the next model. do we need to use looping for this or just use the variable name.solutionValue? also the tuple syntax <i,j> when i write in the main it gives error. Help in this issue will be highly appreciated. Thanks  

Hi

you could have a look at

https://www.ibm.com/developerworks/community/forums/html/topic?id=58d747b0-657a-4539-ac92-b5bea147c1d9&ps=25

And about

also the tuple syntax <i,j> when i write in the main it gives error.

You mix the OPL modeling language and the OPL scripting language. They are different. See https://www.ibm.com/developerworks/community/forums/html/topic?id=6e21279b-db57-4540-a77e-c5f9670988d1&ps=25

regards

Originally posted by: Nomykan

Hi Alex,

Thanks for your help. i just now have one question. what is the difference between using the results of one model and using it as input to the second model and  warm start? apparently both looks same  .. any specific reason for this? 

Hi,

warmstart is a way to use input from a first model into a second model.

But then the engine has all freedomx to go anywhere in the space.

In some cases like https://www.ibm.com/developerworks/community/forums/html/topic?id=193ecefc-757c-4c66-9905-7ac0dd81dc8f&ps=25

you want to forbid cplex to go i some part of the option space.

You choose

Regards