Heuristic recurrent algorithms for photonic Ising machines

被引:97
作者
Roques-Carmes, Charles [1 ,2 ]
Shen, Yichen [3 ]
Zanoci, Cristian [3 ]
Prabhu, Mihika [1 ,2 ]
Atieh, Fadi [2 ,3 ]
Jing, Li [3 ]
Dubcek, Tena [3 ]
Mao, Chenkai [2 ,3 ]
Johnson, Miles R. [4 ]
Ceperic, Vladimir [3 ]
Joannopoulos, John D. [3 ,5 ]
Englund, Dirk [1 ,2 ]
Soljacic, Marin [1 ,3 ]
机构
[1] MIT, Res Lab Elect, 50 Vassar St, Cambridge, MA 02139 USA
[2] MIT, Dept Elect Engn & Comp Sci, 77 Massachusetts Ave, Cambridge, MA 02139 USA
[3] MIT, Dept Phys, 77 Massachusetts Ave, Cambridge, MA 02139 USA
[4] MIT, Dept Math, 77 Massachusetts Ave, Cambridge, MA 02139 USA
[5] Inst Soldier Nanotechnol, 500 Technol Sq, Cambridge, MA 02139 USA
基金
美国国家科学基金会;
关键词
NEURAL-NETWORKS; OPTIMIZATION; LIGHT; COMPUTATION; SILICON;
D O I
10.1038/s41467-019-14096-z
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
The inability of conventional electronic architectures to efficiently solve large combinatorial problems motivates the development of novel computational hardware. There has been much effort toward developing application-specific hardware across many different fields of engineering, such as integrated circuits, memristors, and photonics. However, unleashing the potential of such architectures requires the development of algorithms which optimally exploit their fundamental properties. Here, we present the Photonic Recurrent Ising Sampler (PRIS), a heuristic method tailored for parallel architectures allowing fast and efficient sampling from distributions of arbitrary Ising problems. Since the PRIS relies on vector-to-fixed matrix multiplications, we suggest the implementation of the PRIS in photonic parallel networks, which realize these operations at an unprecedented speed. The PRIS provides sample solutions to the ground state of Ising models, by converging in probability to their associated Gibbs distribution. The PRIS also relies on intrinsic dynamic noise and eigenvalue dropout to find ground states more efficiently. Our work suggests speedups in heuristic methods via photonic implementations of the PRIS. Application-specific computational hardware helps to solve the limitations of conventional electronics in solving difficult calculation problems. Here the authors present a general heuristic algorithm to solve NP-Hard Ising problems in a photonics implementation.
引用
收藏
页数:8
相关论文
共 75 条
[51]   Artificial neural networks in hardware A survey of two decades of progress [J].
Misra, Janardari ;
Saha, Indranil .
NEUROCOMPUTING, 2010, 74 (1-3) :239-255
[52]   Crystal statistics I A two-dimensional model with an order-disorder transition [J].
Onsager, L .
PHYSICAL REVIEW, 1944, 65 (3/4) :117-149
[53]   Critical phenomena and renormalization-group theory [J].
Pelissetto, A ;
Vicari, E .
PHYSICS REPORTS-REVIEW SECTION OF PHYSICS LETTERS, 2002, 368 (06) :549-727
[54]   COLLECTIVE PROPERTIES OF NEURAL NETWORKS - A STATISTICAL PHYSICS APPROACH [J].
PERETTO, P .
BIOLOGICAL CYBERNETICS, 1984, 50 (01) :51-62
[55]  
Phare CT, 2015, NAT PHOTONICS, V9, P511, DOI [10.1038/nphoton.2015.122, 10.1038/NPHOTON.2015.122]
[56]  
Pierangeli D, 2018, ARXIV181209311
[57]   EXPERIMENTAL REALIZATION OF ANY DISCRETE UNITARY OPERATOR [J].
RECK, M ;
ZEILINGER, A ;
BERNSTEIN, HJ ;
BERTANI, P .
PHYSICAL REVIEW LETTERS, 1994, 73 (01) :58-61
[58]   Solving Max-Cut to optimality by intersecting semidefinite and polyhedral relaxations [J].
Rendl, Franz ;
Rinaldi, Giovanni ;
Wiegele, Angelika .
MATHEMATICAL PROGRAMMING, 2010, 121 (02) :307-335
[59]   Deep learning [J].
Rusk, Nicole .
NATURE METHODS, 2016, 13 (01) :35-35
[60]  
Saade A, 2016, INT CONF ACOUST SPEE, P6215, DOI 10.1109/ICASSP.2016.7472872