Tselil Schramm

About

I am broadly interested in the theory of algorithms, optimization, and computational complexity for problems arising in statistics. My work aims to develop algorithmic tools for high-dimensional estimation problems and to characterize and explain information/computation tradeoffs. Some specific themes in my research are: understanding the algorithmic power of the sum-of-squares hierarchy of semidefinite programs; developing fast spectral methods; and relating the power of different models of computation for high-dimensional estimation tasks.