Approximate graph matching

In this paper, vectors are column vectors. Unless otherwise specified, the norm of a matrix ‖M‖ is the Frobenius norm, defined by ‖M‖ = √∑ i,jM 2 ij . The all ones vector is denoted 1 and the all zeros vector is denoted 0. The all ones matrix is denoted J and the all zeros matrix is denoted O. The n× n identity matrix is denoted In, or simply I if its size is understood from context. The vectorization of an m × n matrix A, denoted vec(A), is the mn-dimensional vector created by stacking the rows of A as columns on top of one another. The trace of a square matrix A, denoted trA, is the sum of its diagonal entries. The direct sum of square matrices A and B, denoted A⊕B, is the block matrix [ A O