Originally posted by: Rajasekhar_Kadambur

Hi,

I am trying to work on a bilinear problem which taking longer time. What i am thing is to solve the LP without bilinear constraint in the CPLEX and pass the decision variables as a warm start for the CP with bilinear Constraint or vice versa. 

Could someone suggest the best way to work on CPLEX to CP or linking CP to CPLEX. Please explain the pros and cons of each approach. 

Thanks in advance.

Hi

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

useful ?

And are your decision variables binary ?

regards

Originally posted by: Rajasekhar_Kadambur

Hi,

Yes my decision variables are binaries and integers and one of my constraint is bilinear.

suggest a way to decompose the model, either CPLEX to CP or CP to CPLEX

Hi,

if your decision variables are binary, you can rewrite

x*y = z

into something like

z<=x
z<=y
z>=x+y-1

You may also use logical constraints :

dvar int x;
dvar boolean b;
dvar int bx;

subject to
{
(b==0) == (bx==0);
(b==1) == (bx==x);
}

And then eveything will work fine both with CPLEX and CPO

regards