Originally posted by: Maigha

Hello,

In case of linear programs, situations arise where we have multiple optimal solutions. I know that certain parameters can be set to populate alternate optimal solutions using API classes in MATLAB. By this I mean using the class parameters instead of the toolbox options. Can I obtain alternate solutions using the CPLEX toolbox by some way?

Are you talking about pure linear programs or about (mixed) integer linear programs here?

For (mixed) integer linear programs there is the populate() function in the class API. This functionality is not available through toolbox functions. The toolbox functions only optimize problems. For anything more advanced you will have to use the class API.

Originally posted by: Maigha

Thanks Daniel!

I meant both linear and mixed integer linear programs. I guess you have answered it for both. While solving a problem I realized that I might have multiple optimal solutions and so was wondering if I need to use API classes for obtaining them.

Yes, if there are multiple alternate solutions then you will have to use the class API in case you want to see more than one of them. Note however that populate() only applies to models with integer variables.

Originally posted by: Maigha

Thanks Daniel! 

Originally posted by: WalterNeek

@DanielJunglas I am using cplexlp in Matlab API and want to select whether I want basic or non-basic solution types. Is there a way to do that. My issue is that unlike "linprog" I am getting zero ​value decision variables in "Cplexlp" which are not the solution type i am looking for. My problem seems to have multiple solutions.

Thanks

Have you tried CPX_PARAM_SOLUTIONTYPE ? (see here for how to set parameters in matlab)

In any case, I am not clear what you mean by "not the solution type I am looking for". Could you elaborate on that? Would it be possible to add additional constraints that exclude these solutions?

Originally posted by: WalterNeek

Hi @DanielJunglas 548c249b-bb6b-42f6-99b8-f0376c80829e  Thank you for your reply. I am looking for solutions not to be at extreme points. When i solve the problem using linprog in MATLAB for a flux problem I get lots of small "non-zero" solutions representing the small fluxes in my network but when I use CPLEXLP I get some extreme large solutions and zero values for other which is not representing the physics I am trying to model. I am not able to put extra constraints. ​