题目：Quantum Annealing of a Programmable Spin Glass

报告人：Anders W. Sandvik

邀请人：汪玲

时间：2024年4月18日（周四）15:30

地点：紫金港校区海纳苑8幢215报告厅

摘要：

In quantum annealing, a system with two non-commuting time dependent interactions is slowly evolved from an initial trivial state to a final complex classical state. If the evolution is slow enough, the final state can represent the solution of a difficult optimization problem encoded in the Hamiltonian. Practical implementations of quantum annealing are limited by short coherence times, however, and there are also still fundamental questions about the speedup that can be achieved relative to classical optimization. I will discuss experiments [1] on a device with more than 5000 programmable superconducting qubits (the D-Wave Advantage system) with couplings programmed to one of the prototypical hard optimization problems - the 3D Ising spin glass. While the coherence times are not long enough to reach the ground state, the results and comparisons with numerical simulations show that the system traverses the spin glass transition, and the final state reached for different system sizes and annealing times correctly reproduces expectations from scaling theory. The state of the art is now approaching the point where quantum annealing can address problems beyond the reach of classical computers.

[1] A. King et al., Nature 617, 61 (2023).

个人简介：

Anders W. Sandvik is a Professor of Physics at Boston University. He completed his MSc at Abo Akademi University in Finland in 1989 and his PhD in Physics at the University of California, Santa Barbara, in 1993. He carried out postdoctoral work at Florida State University and the University of Illinois at Urbana-Champaign before returning to Finland as a Senior Fellow of the Academy of Finland in 2000. He joined the faculty of Boston University in 2004. From the beginning of his research career, he has been interested in quantum lattice models and how to study them with unbiased computational techniques. He developed the versatile Stochastic Series Expansion algorithm, a quantum Monte Carlo method that is particularly efficient for a broad range of quantum magnets. During the past several years, he has designed a class of quantum spin models (J-Q models) which can realize many phenomena of interest in contemporary condensed matter physics and quantum field theory, while being amenable to efficient numerical simulations. His detailed studies of these models have provided unique insights into collective quantum many-body states and their quantum phase transitions. Dr. Sandvik received a Fulbright Fellowship for his PhD studies in 1990 and was elected Fellow of the American Physical Society in 2007. In 2017 he was appointed by the Simons Foundation as a Simons Investigator in Physics. He won Aneesur Rahman Prize for Computational Physics in 2021.