Title: “Modewise methods for tensor compression and recovery”

Abstract: Among the most widely used random matrix measurement models are (a) independent sub-gaussian models and (b) randomized Fourier-based models, allowing for the efficient computation of the measurements. My talk will be focused on their generalization to modewise tensor measurement. Modewise tensor maps are linear operators acting on each tensor dimension (mode) separately rather than on the vectorizations of tensors. They require significantly less memory than the measurements working on vectorized tensors but present additional mathematical challenges such as inherent Kronecker structure (thus, less independence in the model). I will talk about the Johnson-Lindenstrauss and restricted isometry properties of such operators, as well as their applications to the low-rank tensor fitting and recovery. Based on the joint work with M.Iwen, D. Needell, M.Perlmutter, and A.Zare.