User-Friendly Tail Bounds for Sums of Random Matrices

This paper presents new probability inequalities for sums of independent, random, self-adjoint matrices. These results place simple and easily verifiable hypotheses on the summands, and they deliver strong conclusions about the large-deviation behavior of the maximum eigenvalue of the sum. Tail bounds for the norm of a sum of random rectangular matrices follow as an immediate corollary. The proof techniques also yield some information about matrix-valued martingales. In other words, this paper provides noncommutative generalizations of the classical bounds associated with the names Azuma, Bennett, Bernstein, Chernoff, Hoeffding, and McDiarmid. The matrix inequalities promise the same diversity of application,…

• DA
  Defense Advanced Research Projects Agency

  Awards: N66001-08-1-2065, FA9550-09-1-0643
• OO
  Office of Naval Research

  Awards: FA9550, N00014
• AF
  Air Force Office of Scientific Research

  Awards: FA9550-09-1-0643, FA9550-, FA9550