Converting GAMS script to JuMP: Marginal values



Hi all,

Hopefully just a quick question here: I’m new to optimisation solvers in general so this may something I’ve missed.

I’m converting a GAMS script into a JuMP solution (using Ipopt). After the optimisation step, my GAMS script sometimes uses the ‘level’ values of a time dependent variable var.l(t), and at other times the ‘marginal’ value var.m(t).

Is this marginal value equivalent to the ‘reduced cost’ available via getdual(var) in JuMP parlance?

Not all of my implementation is matching up, so I can’t quite test the answer for myself at the moment.


Yes. And similarly for shadow costs (dual values of constraints): constraint.m in GAMS is the same as getdual(constraint) in JuMP.


Excellent! Thanks Niclas.