Speaker
Description
The problem of reconstructing spectral densities from noisy
Euclidean-time Monte-Carlo data provides a valuable test-bed for
investigating real-time and inclusive observables, and sign problems
more broadly. In this talk, I will discuss new methods for spectral
reconstructions that unify several approaches, including
analyticity-based approaches (Nevanlinna-Pick interpolation, moment
problems), convex programming approaches, and the commonly used
Hansen, Lupo, and Tantalo (HLT) method. These methods use tools
originally developed for the conformal bootstrap in order to provide
rigorous bounds on smeared spectral functions, and are directly
applicable to noisy Monte-Carlo data. I will review the methods and
their relations, and also discuss potential future directions.