The boundary for quantum advantage in Gaussian boson sampling

被引:59
作者
Bulmer, Jacob F. F. [1 ]
Bell, Bryn A. [2 ,9 ]
Chadwick, Rachel S. [1 ,3 ]
Jones, Alex E. [1 ]
Moise, Diana [4 ]
Rigazzi, Alessandro [4 ]
Thorbecke, Jan [5 ]
Haus, Utz-Uwe [6 ]
Van Vaerenbergh, Thomas [7 ]
Patel, Raj B. [2 ,8 ]
Walmsley, Ian A. [2 ]
Laing, Anthony [1 ]
机构
[1] Univ Bristol, Quantum Engn Technol Labs, Bristol, Avon, England
[2] Imperial Coll London, Dept Phys, Ultrafast Quantum Opt Grp, London, England
[3] Univ Bristol, Quantum Engn Ctr Doctoral Training, Bristol, Avon, England
[4] Hewlett Packard Enterprise, Zurich, Switzerland
[5] Hewlett Packard Enterprise, Amstelveen, Netherlands
[6] HPE HPC AI EMEA Res Lab, Wallisellen, Switzerland
[7] HPE Belgium, Hewlett Packard Labs, Diegem, Belgium
[8] Univ Oxford, Dept Phys, Oxford, England
[9] Oxford Quantum Circuits, Thames Valley Sci Pk, 3 Collegiate Sq,South Ave, Reading RG2 9LH, Berks, England
基金
英国工程与自然科学研究理事会;
关键词
PERFORMANCE;
D O I
10.1126/sciadv.abl9236
中图分类号
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号
07 ; 0710 ; 09 ;
摘要
Identifying the boundary beyond which quantum machines provide a computational advantage over their classical counterparts is a crucial step in charting their usefulness. Gaussian boson sampling (GBS), in which photons are measured from a highly entangled Gaussian state, is a leading approach in pursuing quantum advantage. State-of-the-art GBS experiments that run in minutes would require 600 million years to simulate using the best preexisting classical algorithms. Here, we present faster classical GBS simulation methods, including speed and accuracy improvements to the calculation of loop hafnians. We test these on a similar to 100,000-core supercomputer to emulate GBS experiments with up to 100 modes and up to 92 photons. This reduces the simulation time for state-of-the-art GBS experiments to several months, a nine-orders of magnitude improvement over previous estimates. Last, we introduce a distribution that is efficient to sample from classically and that passes a variety of GBS validation methods.
引用
收藏
页数:8
相关论文
共 48 条
[1]  
Aaronson S, 2011, ACM S THEORY COMPUT, P333
[2]  
[Anonymous], 2008, MONTE CARLO STRATEGI
[3]  
Balasubramanian K., 1980, Ph.D. thesis
[4]   A finite-difference sieve to count paths and cycles by length [J].
Bax, E ;
Franklin, J .
INFORMATION PROCESSING LETTERS, 1996, 60 (04) :171-176
[5]   A faster hafnian formula for complex matrices and its benchmarking on a supercomputer [J].
Björklund A. ;
Gupt B. ;
Quesada N. .
ACM Journal of Experimental Algorithmics, 2019, 24 (01)
[6]   Classical simulation of linear optics subject to nonuniform losses [J].
Brod, Daniel J. ;
Oszmaniec, Michal .
QUANTUM, 2020, 4
[7]   Generalized concurrence in boson sampling [J].
Chin, Seungbeom ;
Huh, Joonsuk .
SCIENTIFIC REPORTS, 2018, 8
[8]  
Clifford P., 2020, ARXIV PREPRINT ARXIV
[9]  
Clifford P, 2018, SODA'18: PROCEEDINGS OF THE TWENTY-NINTH ANNUAL ACM-SIAM SYMPOSIUM ON DISCRETE ALGORITHMS, P146
[10]   Faster exponential-time algorithms in graphs of bounded average degree [J].
Cygan, Marek ;
Pilipczuk, Marcin .
INFORMATION AND COMPUTATION, 2015, 243 :75-85