15558

Learning Many-Body Quantum Systems From Local Measurements

APA

(2020). Learning Many-Body Quantum Systems From Local Measurements. The Simons Institute for the Theory of Computing. https://simons.berkeley.edu/talks/self-testing-approach-certifying-quantum-systems

MLA

Learning Many-Body Quantum Systems From Local Measurements. The Simons Institute for the Theory of Computing, Mar. 30, 2020, https://simons.berkeley.edu/talks/self-testing-approach-certifying-quantum-systems

BibTex

          @misc{ scivideos_15558,
            doi = {},
            url = {https://simons.berkeley.edu/talks/self-testing-approach-certifying-quantum-systems},
            author = {},
            keywords = {},
            language = {en},
            title = {Learning Many-Body Quantum Systems From Local Measurements},
            publisher = {The Simons Institute for the Theory of Computing},
            year = {2020},
            month = {mar},
            note = {15558 see, \url{https://scivideos.org/Simons-Institute/15558}}
          }
          
Itai Arad (Technion)
Talk number15558
Source RepositorySimons Institute

Abstract

Recovering an unknown Hamiltonian or a Linbladian from measurements is an increasingly important task for certification of noisy quantum devices and simulators. Recent works have succeeded in recovering the Hamiltonian of an isolated quantum system with local interactions from long-ranged correlators of a single eigenstate. Here we generalize these works to allow for the recovery of the Hamlitonian or a Linbladian from their steady states, including, for example, the Gibbs state at a finite temperature. Our approach takes advantage of the non-commuteness of the underlying dynamics to derive non-trivial local constraints between the local reduced density matrices and the local generators of dynamics. This enables us to learn a local patch of the system from local observations only on that patch, even though the overall state might be globally entangled. Surprisingly, there are cases in which our algorithm can be exponentially faster than the classical problem of learning a Boltzmann machine, or, more generally, a graphical model. Finally, our approach can also be easily extended to the case of learning a system from its dynamics. Based on joint works with E. Bairy, N. Lindner, Chu Guo and D. Poletti: 1. Phys. Rev. Lett. 122, 020504 (2019), arXiv:1907.11154 2. New J. Phys. 22 032001 (2020), arXiv:1807.04564