Theory Seminar - Spectral Independence in High-Dimensional Expanders, Kuikui Liu
Explore a theory seminar on spectral independence in high-dimensional expanders and its applications to the hardcore model. Delve into Kuikui Liu's presentation, which demonstrates how spectrally independent probability distributions relate to local spectral expanders in high-dimensional simplicial complexes. Learn about the implications for rapid mixing of Glauber dynamics and efficient sampling from these distributions. Discover the groundbreaking application to the hardcore model, showing polynomial-time mixing of Glauber dynamics up to the uniqueness threshold, improving upon previous quasi-polynomial time algorithms. Gain insights into topics such as Markov chain decomposition, Garland's method for local spectral expansion, weak spatial mixing, and tree recurrence. Engage with open-ended problems and future research directions in this 55-minute lecture from the Paul G. Allen School, complete with closed captions.