Speaker: Kathryn Beck, Tufts University

Date: June 2, 2026

Abstract: The graph short-time Fourier transform was recently defined as an extension of the discrete short-time Fourier transform to the graph setting. The construction utilizes the heat kernel as a window function in order to localize the graph Fourier transform, leading to a time-dependent graph Gabor system. When the underlying graphs are vertex-transitive or strongly regular, the Gabor system was shown to be a tight frame for all time. We focus our attention on random graph models, particularly Erdős–Rényi graphs. We present preliminary results in which we investigate the tightness of the graph Gabor system as time evolves.