Originally posted by: JDalal

I am solving a LP model in Concert (Java), and extracting the dual variable values using getDuals() function. 

In a separate implementation, I am solving the Dual of my LP model explicitly and printing out the variable values.

Now, I observe several variable values are quite different. Is it normal, assuming possibility of alternate optima, or something is wrong with my implementation?

By the way, the objective function value is matching in both implementations.

Any suggestion is appreciated.

Originally posted by: T_O

If both solutions are dual feasible for the original problem, this is ok. Like primal solutions, dual solutions don't have to be unique.

Best regards,
Thomas

Originally posted by: JDalal

Thank you for the clarification.