Originally posted by: felycite28

Hello eevryone ,

I am working on a model whose objective function is

Min Z= Ax+ exp(p) // p variable and p=K-L , x variable and A constant

I learned that I can apply SOS2 piecewise linearization . Can anyone explain how can I solve it on Cplex , if possoble, on the same example  I mean (exponential style)?

How will I modify the objective function on cplex and make it ready for debug?

Thank you in advance

Hi

instead of writing your own sos2 you could use pwl 

See 

https://www.ibm.com/developerworks/community/forums/html/topic?id=30714742-aef3-41c8-9e1d-d6ce99a63ce8&ps=25

for an example in opl

regards

Originally posted by: felycite28

Yes, I had already done it (the directed link is my question already)  but in that way the solution even for small instances took too much time.  I had  wanted to try to solve with SOS2

just could manage to make it ready for CPLEX ?

Thank you Alex

hi

then let me quote Paul Rubin in quora

You may want to refine the discretization of  in the vicinity of the solution you got and repeat (using the previous solution as an initial incumbent).

regards