Kronecker product constraints

[Javier Gutiérrez Oliet]

I am trying to implement a minimization problem of the following form:

objective = cvxpy.Minimize(cvxpy.norm(vx, 1))

constraints = [A@vx == b]

where the array A gets huge as it is the kronecker product of two arrays and therefore I get memory errors as I try to get to bigger dimensions.

I was wondering if there is any way I could use cvxpy in order to use the array A without declaring it as a variable before.

Thank you in advance!

J.

[Steven Diamond]

Unfortunately there is no special efficient way to do this in cvxpy that comes to mind. I did research on the topic of special linear operators in the past and made a system that handles many special operators from computational imaging (https://github.com/comp-imaging/ProxImaL) but we didn't implement kronecker product.