Poster 85: Hybrid Computing Platform for Combinatorial Optimization with the Coherent Ising Machine

Authors: Junya Arai (Nippon Telegraph and Telephone Corporation), Yagi Satoshi (Nippon Telegraph and Telephone Corporation), Hiroyuki Uchiyama (Nippon Telegraph and Telephone Corporation), Toshimori Honjo (Nippon Telegraph and Telephone Corporation), Takahiro Inagaki (Nippon Telegraph and Telephone Corporation), Kensuke Inaba (Nippon Telegraph and Telephone Corporation), Takuya Ikuta (Nippon Telegraph and Telephone Corporation), Hiroki Takesue (Nippon Telegraph and Telephone Corporation), Keitaro Horikawa (Nippon Telegraph and Telephone Corporation)

Abstract: Several institutes are operating cloud platforms that offer Web API access to Ising computers such as quantum annealing machines. Platform users can solve complex combinatorial optimization problems by using hybrid algorithms that utilize both users' conventional digital computers and remote Ising computers. However, communication via the Internet takes an order of magnitude longer time than optimization on Ising computers. This overheads seriously degrade the performance of hybrid algorithms since they involve frequent communication. In this poster, we first state issues in the design of Ising computing platforms, including communication overheads. Then, we answer the issues by introducing the computing platform for the coherent Ising machine (CIM), an Ising computer based on photonics technologies. Our platform offers efficient CIM-digital communication by allowing users to execute their program on digital computers co-located with the CIM. We have released the platform to our research collaborators in this autumn and started the evaluation.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF


Back to Poster Archive Listing