Inconsistencies in lft/feedback/starprod

[baggepinnen]

It appears as if the implementation of the lft

ControlSystems.jl/src/connections.jl

Line 269 in 17e78c1

A = [sys1.A + α*s1_B2*s2_D22*s1_C2 α*s1_B2*s2_C2;

consistently uses DK22 (s2_D22) instead of DK11 as in the reference. Most upper-left blocks should probably have s2_D11 instead

@olof3

[baggepinnen]

There might be some problem with the s2_C2 element in the lower-right corener of the A matrices, should be s2_C1

[baggepinnen]

Actually, it seems that all indices for s2 are consistently switched 1 <-> 2 making me believe it was a deliberate choice. The book is rather vague on how they define K, i.e., is K upside down compared to P having y/u as the upper input-output pair? Could be therefore the indices are switched?

[olof3]

Actually, it seems that all indices for s2 are consistently switched 1 <-> 2 making me believe it was a deliberate choice. The book is rather vague on how they define K, i.e., is K upside down compared to P having y/u as the upper input-output pair? Could be therefore the indices are switched?

Yes, exactly. "See Zhou etc. for similar (somewhat less symmetric) formulas." I believe that the modified version that I implemented makes better sense for us.

Should probably write out the reference a bit better, but seems like you had no problem to find it :)

[a09174079722A]

Ok