Can Robust Optimization be combined with optimizer()?

[XB G]

I'm trying to find the right size \Gamma for the uncertainty set in RO, one way is to redefine my uncertain constraint (e.g.  [-0.5*Gamma <= w <- 0.5*Gamma] ) again and again with different values of Gamma. This is quite slow since YALMIP re-build the robust counterpart repetitively, although there is only one difference.

Another way (I'm considering) is to define Gamma as a variable, and use optimizer to declare Gamma as a parameter (e.g. ro_problem = optimizer(constr, obj, [], Gamma, x)  ), then solve (e.g. ro_problem{1}, ro_problem{0.5} ) them without reformulating the same problem.

I wrote some simple code, it seems YALMIP will think there is no uncertainty description in the model, the previous uncertainty set will be regarded as F_xw constraint by decomposeUncertainty (correct?)

Please let me know if anything is wrong.

Thanks!

XG

[Johan Löfberg]

https://yalmip.github.io/parameterizeduncertainty