Georgia Institute of Technology College of Sciences

We will prove the key lemma underlying the probabilistic unique continuation result of Ding-Smart, namely that for "thin" tilted rectangles, boundedness on all of one of the long edges and on a 1-\varepsilon proportion of the opposite long edge implies a bound (in terms of the dimensions of the rectangle) on the whole rectangle (with high probability). 

The estimation of distributions of complex objects from high-dimensional data with low-dimensional structures is an important topic in statistics and machine learning. Deep generative models achieve this by encoding and decoding data to generate synthetic realistic images and texts. A key aspect of these models is the extraction of low-dimensional latent features, assuming data lies on a low-dimensional manifold. We study this by developing a minimax framework for distribution estimation on unknown submanifolds with smoothness assumptions on the target distribution and the manifold. The framework highlights how problem characteristics, such as intrinsic dimensionality and smoothness, impact the limits of high-dimensional distribution estimation. Our estimator, which is a mixture of locally fitted generative models, is motivated by differential geometry techniques and covers cases where the data manifold lacks a global parametrization. 

The Fermi variety plays a crucial role in the study of    periodic operators.  In this talk, I will  first discuss recent works on the irreducibility of  the Fermi variety  for discrete periodic Schr\"odinger  operators.   I will then  discuss the applications to  solve  problems of embedded eigenvalues, isospectrality and quantum ergodicity.