With 4 minicourses this year it won’t be possible to schedule contributed talks. Participants who are willing to contribute by presenting their own work, have the opportunity to do so in a “virtual” poster session. It will be fun to discuss each other posters at any time outside the lectures hours. 

Quantitative mixing for locally monotone stochastic evolution equations slides

Jonas Tölle Aalto University, joint work with Gerardo Barrera IST Lisboa.

Potentials of occupations measures of Gaussian fields and fractal minimizers slides

Jonas Tölle Aalto University, joint work with Michael Hinz Bielefeld and Lauri Viitasaari Aalto.

Concentration of large random Gram matrices, slides

Ian Välimaa, Aalto University

Abstract: We consider the concentration of a Gram matrix XX^T and its off-diagonal part, where X is a large random matrix with independent and centered entries bounded by one. Such matrices play a fundamental role in high-dimensional statistics including covariance matrix estimation, dimension reduction and tensor clustering. It is well-known that, under suitable assumptions, the empirical eigenvalue distribution of the Gram matrix converges to the Marchenko–Pastur distribution. Furthermore, the extremal eigenvalues converge to the edges of the limiting distribution. In contrast, less is known about the behavior when the diagonal is removed, especially when the matrix X is particularly wide. This presentation focuses on the extremal eigenvalues of the off-diagonal part of the Gram matrix.