16
23

Testing Matrix Rank, Optimally

Abstract

We show that for the problem of testing if a matrix AFn×nA \in F^{n \times n} has rank at most dd, or requires changing an ϵ\epsilon-fraction of entries to have rank at most dd, there is a non-adaptive query algorithm making O~(d2/ϵ)\widetilde{O}(d^2/\epsilon) queries. Our algorithm works for any field FF. This improves upon the previous O(d2/ϵ2)O(d^2/\epsilon^2) bound (SODA'03), and bypasses an Ω(d2/ϵ2)\Omega(d^2/\epsilon^2) lower bound of (KDD'14) which holds if the algorithm is required to read a submatrix. Our algorithm is the first such algorithm which does not read a submatrix, and instead reads a carefully selected non-adaptive pattern of entries in rows and columns of AA. We complement our algorithm with a matching query complexity lower bound for non-adaptive testers over any field. We also give tight bounds of Θ~(d2)\widetilde{\Theta}(d^2) queries in the sensing model for which query access comes in the form of Xi,A:=tr(XiA)\langle X_i, A\rangle:=tr(X_i^\top A); perhaps surprisingly these bounds do not depend on ϵ\epsilon. We next develop a novel property testing framework for testing numerical properties of a real-valued matrix AA more generally, which includes the stable rank, Schatten-pp norms, and SVD entropy. Specifically, we propose a bounded entry model, where AA is required to have entries bounded by 11 in absolute value. We give upper and lower bounds for a wide range of problems in this model, and discuss connections to the sensing model above.

View on arXiv
Comments on this paper