µ synthesis is a generalization of the controller design done using H infinity control. Is there a way to do µ synthesis using the RobustAndOptimalControl package?
Currently not, no. We support limited forms of mu analysis, in particular with complex structured uncertainties.
Mu synthesis using the DK iteration appears a bit finnecky to implement, in particular the fitting of the scaling functions. If anyone is aware of another approach I’d be interested in learning!
Not so clear to me what you mean. Do you mean difficult?
Yeah, easy in theory, but likely to require a lot of engineering and tweaking to get to work well.
There are some papers that describe how to do it…
https://www.researchgate.net/publication/286976986_D-K_iteration_algorithm_for_m-synthesis
I’m well aware of these papers, but I still don’t find the method particularly attractive. I’d be happy for you to prove me wrong 
My collegues in Delft (NL) seam to like it a lot: https://www.researchgate.net/publication/224782520_Design_of_Robust_Autoland_Control_Laws_Using_Mu-Synthesis/link/546b017f0cf2397f783031db/download
I will discuss it with Jan-Willem van Wingerden from Delft Center for Systems and Control…
Created an issue: Please add mu syntesis · Issue #76 · JuliaControl/RobustAndOptimalControl.jl · GitHub
I think this cannot be too difficult, because Hinfinity calculation is already included in the package, and optimizing some parameters to minimize the robust H∞ performance of the closed-loop system can also not be too difficult because a lot of optimizers are already available in Julia.
I am unlikely to be spending time on this any time soon, but I would be warmly welcoming your contribution!
I double checked, and there is no open source implementation of this algorithm yet, not in scilab, octave or python-control. I start to work at the control department of TU Delft soon, perhaps this gives me the opportunity to work on this topic.
2 Likes