21
1

Fitting an ellipsoid to random points: predictions using the replica method

Abstract

We consider the problem of fitting a centered ellipsoid to nn standard Gaussian random vectors in Rd\mathbb{R}^d, as n,dn, d \to \infty with n/d2α>0n/d^2 \to \alpha > 0. It has been conjectured that this problem is, with high probability, satisfiable (SAT; that is, there exists an ellipsoid passing through all nn points) for α<1/4\alpha < 1/4, and unsatisfiable (UNSAT) for α>1/4\alpha > 1/4. In this work we give a precise analytical argument, based on the non-rigorous replica method of statistical physics, that indeed predicts a SAT/UNSAT transition at α=1/4\alpha = 1/4, as well as the shape of a typical fitting ellipsoid in the SAT phase (i.e., the lengths of its principal axes). Besides the replica method, our main tool is the dilute limit of extensive-rank "HCIZ integrals" of random matrix theory. We further study different explicit algorithmic constructions of the matrix characterizing the ellipsoid. In particular, we show that a procedure based on minimizing its nuclear norm yields a solution in the whole SAT phase. Finally, we characterize the SAT/UNSAT transition for ellipsoid fitting of a large class of rotationally-invariant random vectors. Our work suggests mathematically rigorous ways to analyze fitting ellipsoids to random vectors, which is the topic of a companion work.

View on arXiv
Comments on this paper