SketchyCGAL.jl

A Julia implementation of the SketchyCGAL SDP solver.

sketchy_cgal(C, As, b; R, iterations=1e3, β=1, info_io=stdout)

Solves a trace constrained SDP:

min.    ⟨C, X⟩
s.t.    ⟨As[i], X⟩ == b[i], ∀i;
        tr(X) = 1;
        X ⪰ 0.

where C, As[1], ..., As[m] are symmetric matrices of size n and b is a vector of length m. Matrices should be scaled such that their spectral norm is 1.

Acknowledgement

This project is based on

Yurtsever, A., Tropp, J., Fercoq, O., Udell, M., & Cevher, V. (2021). Scalable Semidefinite Programming. SIAM Journal on Mathematics of Data Science, 3(1), 171-200.