Moving from LMI toolbox (Matlab) to JuliaOpt

I am afraid I am not able to follow you. Now you start talking about some \Lambda variable, which appeared in none of your previously stated problems…

A few general remarks:

  1. I can hardly think of a need to analyze positive (semi) definiteness of a matrix while not assuming a symmetry. At least in control theory (which is where your problems are obviously coming from), these matrices define quadratic matrix forms and even if you somehow start with a nonsymmetrix matrix P, you can symmetrize it through (P+P')/2 with no impact on the value of the quadratic form. Thats why you will barely encounter a discussion of nonsymmetric matrices in the LMI/SDP literature.

  2. Whether you can use the function Positive(), you can easily check by yourself. You do not need discussion forum for that, do you? Just read the documentation: for Convex.jl package at Basic Types · Convex.jl. For JuMP. jl package at Variables · JuMP.

  3. In order to increase your chances of being helped by somebody here, please read and follow the instructions at Please read: make it easier to help you. In particular (but not only) the item 2 on inserting the code into your posts.

4 Likes