TY - JOUR

T1 - Superconductor Amoeba-Inspired Problem Solvers for Combinatorial Optimization

AU - Takeuchi, Naoki

AU - Aono, Masashi

AU - Yoshikawa, Nobuyuki

N1 - Funding Information:
Superconductor logic can perform logic operations in an energy efficient manner by taking advantage of its physical features: zero dc resistance, magnetic flux quanta, and the Josephson effect. Over the last several decades, many types of superconductor logic families have been proposed and demonstrated . Rapid single flux quantum (RSFQ) logic is one of the most developed logic families; RSFQ microprocessor prototypes have been designed and demonstrated by Tanaka et al. Moreover, the extensive study of RSFQ logic has contributed to the invention of very energy efficient logic families, such as energy efficient RSFQ (ERSFQ) logic , reciprocal quantum logic (RQL) , and low-voltage RSFQ (LV RSFQ) logic . The Cryogenic Computing Complexity (C3) project funded by the Intelligence Advanced Research Projects Activity (IARPA) has recently been developing very low-power microprocessors using ERSFQ logic and RQL while developing submicron fabrication technology .
Funding Information:
The present study was supported by PRESTO of the Japan Science and Technology Agency (JST) (Grants No. JPMJPR1528 and No. JPMJPR1321). The circuits were fabricated in the Clean Room for Analog-digital superconductiVITY (CRAVITY) of the National Institute of Advanced Industrial Science and Technology (AIST) with the standard process (STP2). The authors would like to thank C. J. Fourie for providing the 3D inductance extractor (InductEx), Y. Abe for supporting measurements, and C. L. Ayala for proofreading the manuscript.
Publisher Copyright:
© 2019 American Physical Society.

PY - 2019/4/22

Y1 - 2019/4/22

N2 - Adiabatic quantum-flux-parametron (AQFP) logic is an energy-efficient superconductor logic family; the energy dissipation of an AQFP gate can be arbitrarily reduced through adiabatic switching. In addition to high energy efficiency, AQFP logic has the advantage that it can easily introduce stochastic processes by exploiting naturally occurring thermal fluctuations. We propose using AQFP logic to implement an amoeba-inspired problem solver (APS), which is a stochastic local search method to explore solutions to combinatorial optimization problems such as the Boolean satisfiability problem (SAT). We designed a superconductor amoeba-inspired problem solver (SAPS) using AQFP logic, which finds solutions to a simple logical constraint satisfaction problem in the manner of APS, and fabricate it using a Nb integrated circuit fabrication process. Experimental results show that the probability distribution of the stochastic processes in AQFP logic can be controlled by the magnitude of bias current and that SAPS finds solutions using a small number of iterations when a moderate bias current is applied. The present results indicate the possibility of using AQFP logic to build hardware dedicated to the implementation of stochastic local search algorithms to solve combinatorial optimization problems such as SAT.

AB - Adiabatic quantum-flux-parametron (AQFP) logic is an energy-efficient superconductor logic family; the energy dissipation of an AQFP gate can be arbitrarily reduced through adiabatic switching. In addition to high energy efficiency, AQFP logic has the advantage that it can easily introduce stochastic processes by exploiting naturally occurring thermal fluctuations. We propose using AQFP logic to implement an amoeba-inspired problem solver (APS), which is a stochastic local search method to explore solutions to combinatorial optimization problems such as the Boolean satisfiability problem (SAT). We designed a superconductor amoeba-inspired problem solver (SAPS) using AQFP logic, which finds solutions to a simple logical constraint satisfaction problem in the manner of APS, and fabricate it using a Nb integrated circuit fabrication process. Experimental results show that the probability distribution of the stochastic processes in AQFP logic can be controlled by the magnitude of bias current and that SAPS finds solutions using a small number of iterations when a moderate bias current is applied. The present results indicate the possibility of using AQFP logic to build hardware dedicated to the implementation of stochastic local search algorithms to solve combinatorial optimization problems such as SAT.

UR - http://www.scopus.com/inward/record.url?scp=85064882382&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064882382&partnerID=8YFLogxK

U2 - 10.1103/PhysRevApplied.11.044069

DO - 10.1103/PhysRevApplied.11.044069

M3 - Article

AN - SCOPUS:85064882382

VL - 11

JO - Physical Review Applied

JF - Physical Review Applied

SN - 2331-7019

IS - 4

M1 - 044069

ER -