Tag
This paper establishes sharp convergence rates for iteratively reweighted least squares methods in nuclear norm minimization, proving the optimality of harmonic-mean weight operators for low-rank matrix recovery.