Something that doesn’t get as much attention in the quantum world as it should is the significant amount of work in Japan to develop quantum and quantum-inspired systems to solve various types of combinatorial optimizing problems. These problems have a tremendous number of potential real world use cases and could provide a very valuable advantage over classical computers with a specialized machine that has a high number of qubits (or spins) with high quality and high connectivity. Part of the reason Japanese researchers have been so active in this area is because the original concept was proposed in 1998 by Tadashi Kadowaki and Hidetoshi Nishimori from the Tokyo Institute of Technology in a paper titled Quantum Annealing in the Transverse Ising Model. Whereas the majority of quantum computing research we track in the U.S., Europe, Australia, and China, is based around gate-based computing, it appears that the research in Japan is more heavily weighted towards quantum annealing and other optimization approaches.

Here is a summary of some of the activity we have seen:

  • Fujitsu has introduced a quantum-inspired product called a digital annealer. This product uses classical computing technology, but supports up to 8192 spins with all-to-all coupling. We reported on their second generation device here and you can see additional details about this offering on their website here.
  • NEC invested $10 million in D-Wave in 2019 and signed an agreement to be an authorized reseller of D-Wave’s cloud computing services. However, they are also planning to develop their own quantum annealer with by 2023 and reached a partnership deal with ParityQC to utilize an architecture that ParityQC has developed. The planned machine will included qubits with long coherence times and also all-to-all connectivity between qubits. Additional information regarding NEC’s quantum computing initiatives can be found on their website here.
  • NTT has been developing a technology called a Coherent Ising Machine (CIM) that uses photonics with pulses of light that circulate in a 1 km fiber optic cable ring. They have established a Silicon Valley Research group and are collaborating with many researchers around the world including Caltech which is helping them develop an on-chip 100 GHz pulsed pump laser source and on-chip parametric oscillator device. The goal is to develop with world’s fastest Coherent Ising Machine device. They had previously developed a 2000 spin device and made it available on the cloud briefly in 2018/2019. Their new effort is aiming to have a machine that has a great many more qubits, more highly integrated and more reliable. You can view our report on this latest announcement here.
  • Toshiba is also developing another quantum-inspired solution called a Simulated Bifurcation Machine (SBA). You can read a technical paper in Science Advantages about a 2000 spin all-to-all coupling implementation of this machine here. Toshiba recently announced that they would become a partner of the Microsoft Azure Quantum cloud offering and we expect some version of a Toshiba SBM to available there shortly.
  • Hitachi has been performing research on something they call a CMOS Annealing Machine. Although they have not publicly released any machine of this type yet, you can view some information about it that they published on their website here and here.
  • Finally, besides the large companies listed above developing hardware, there are a number of software startups in Japan that are developing software for quantum annealing machines. A*Quantum is developing an upstream layer platform that can map problems to lower level languages and run on an annealer. blueqat (formerly MDR) has had several projects, including a Coronavirus research project, that use the D-Wave quantum annealers. Jij is a company whose founders all came from the Nishimori Lab at the University of Tokyo and specializes in developing technologies for computationally difficult optimization problems. Groovenauts has released a business optimization service that runs on a D-Wave computer over the cloud. Sigma-i has signed a cloud access agreement with D-Wave. As part of the agreement, Sigma-i will promote the use of quantum annealing machines at companies, universities, research institutes, etc., in order to contribute to the development of quantum computing technology in Japan. And Tokyo Quantum Computing is a software startup specifically working on quantum annealing computing.

So we don’t think anyone should ignore these efforts. Some of the hardware roadmaps we have heard about extend to the ten’s or hundred’s of thousands of spins (or qubits) which could far exceed the roadmaps we have seen from IBM, Google or any of the other gate based companies. In addition, many of these implementations include all-to-all coupling which helps when embedding a problem onto the hardware. Although the architectures built for optimization are not universal and would not support some of the applications being eyed for quantum computing, there are a great many places where this technology could be used with great success given a large enough machine that has a low error rate. So in the race to provide technologies that can solve difficult computational problems better or faster than standard classical computing solutions, these approaches may show a lot of success in the future.

February 20, 2021