Speaker
Description
Tempered Lefschetz thimble method (TLTM) [Fukuma and Umeda, arXiv:1703.00861] is a (parallel-)tempering algorithm to solve the sign problem in Monte Carlo simulations. It is implemented on the generalized Lefschetz thimble method (GLTM) by using the flow time of the antiholomorphic gradient flow as the tempering parameter. The TLTM is expected to versatilely solve the dilemma between the sign problem and the multimodal problem which inherently exists in the GLTM. In this talk, after briefly reviewing the TLTM, I apply the method to the quantum Monte Carlo simulation of the Hubbard model away from half-filling. I show that the TLTM certainly solves the dilemma and gives results that agree nicely with exact values. This talk is based on a paper in preparation [Fukuma, Matsumoto and Umeda].
