Does anyone have experience in moving from Matlab’s LMI toolbox to the various JuliaOpt packages? I’m taking a non-linear dynamics course (AAE666 at Purdue) and we’ve just been introduced to linear matrix inequalities in the context of quadratic stability. As such, I’m still trying to grapple with the toolbox and LMIs in general and I’m not sure how to translate such a problem to a JuliaOPT package. Any resources or thoughts would be much appreciated.
Did you have a look at JuMP, especially the Section Semidefinite Constraints or the Example Example minimal Ellipse?
I have no experience so far in LMI’s but I think this would be a good starting point.
For example solving the Lyapunov Inequality for a continuous system could look like this:
using JuMP using MosekTools using LinearAlgebra A = [-0.5 1; 0 -0.3]; n = size(A,1); Q = [1 0; 0 1] model = Model(with_optimizer(Mosek.Optimizer)) @variable(model, P[i=1:n, j=1:n], PSD) @SDconstraint(model, P ⪰ 0) @SDconstraint(model, A'*P + P*A ⪯ -Q) @SDconstraint(model, A'*P + P*A ⪰ -Q) optimize!(model) P = value.(P) # Test result: A'*P + P*A ≈ -Q