Ushering in practical applications of the dream quantum computer through quantum annealing
January 09, 2019
A quantum computer is a dream machine, which, if realized, would be capable of instantaneously performing massive computations beyond what conventional computers could do. NEC was the first to demonstrate a solid-state quantum bit (qubit) operation that serves as a key technology for quantum computers. Going forward, NEC is conducting further research aimed at the practical application of a “total-coupling quantum annealing machine,” which is the easiest to use among quantum computers, by 2023.
NEC aims to stabilize the “quantum superposition state,” which is essential for solving problems through quantum computing
Conventional computers perform calculations by writing “bit values” as either “0” or “1” based on the on-and-off switching of electric current. Quantum computers, on the other hand, make use of the mysterious state of being both “0” and “1” at the same time (so-called quantum superposition state) to simultaneously perform parallel computations for when the “qubit value” is “0” and when it is “1.” A quantum computer that is capable of fully exhibiting its theoretical potential has so far not been developed. This is because the superposition state, which is something like being in between the particle and wave forms, is in fact very unstable and dissipates within a very short time, making it difficult to gain time needed for solving problems.
The new method originally developed by NEC prevents the disruption of the quantum superposition state by heat, vibrations, and other forms of noise, making it possible to stably maintain the superposition state 1000 times longer than through existing technologies.
Further, performing computations using the quantum annealing machine described below requires a mapping process where real-world combinational optimization problems are converted into the coupling of qubits in the superposition state, thereby making it necessary to stably couple multiple qubits. NEC has overcome this problem using original two-dimensional arrangement and total coupling between qubits and eventually aims to develop a “total-coupling quantum annealing machine.”
Instantaneous computation of complex combinational optimization problems, such as finding the most optimum route for home delivery services
Artificial intelligence (AI) is increasingly being used to solve various real-world problems. One method for solving problems is selecting the best combination from an infinite number of choices generated through machine learning. A typical example for this is determining which cargo to be carried by which truck and which route to take in order to finish the delivery of goods within the shortest time, given an X number of trucks and Y number of items to deliver.
These kinds of complex combinational optimization problems take even supercomputers an enormous amount of time to solve, making it impossible to obtain accurate answers within the required time. The possible number of routes increases exponentially with the increase in the number of delivery points. Moreover, other factors such as adjusting to changes in conditions brought about by the continuously changing traffic situation and the need for re-delivery due to absence of recipients must also be considered. The number of combinations involved in determining the shortest delivery time with the minimum use of fuel and minimum working hours for delivery staff, therefore, becomes extremely large. It is thus impossible for current computers to find the best combination even for years on end. Even the quantum computers currently offered for commercial use cannot solve such complex and large-scale problems. Indeed, we are still a long way from realizing practical quantum computers. Business, however, requires making the best decisions within at most a few tens of minutes.
Combinational optimization problems related to the search for routes is a common problem that is true not only for home delivery services, but for large-scale distribution of goods as well. Likewise, the problem of finding the best allocation and combination of thousands of financial instruments such as shares and investment trust, and assets such as bonds, cash, and savings, etc. to achieve the best investment results is also a difficult problem in the finance industry. There are in fact many areas and applications where the elucidation of high-quality combinations can lead to the creation of better products and services, such as in design of circuits, synthesis of metal products, and development of drugs.
As societies mature, their problems also increase in complexity, leading to an ever-increasing number of factors to consider in solving problems. In the future, the evolution of quantum computer technologies capable of solving complex problems will lead to significant social changes across many fronts: time savings, energy and cost reduction, proper manpower allocation, labor shortage resolution, industrial optimization, and value creation.
A quantum annealing machine capable of handling computations for combinational optimization problems
Among quantum computers, quantum annealing machines, which use weights among qubits in performing computations, has received wide attention as a tool for solving complex combinational optimization problems within a short period of time.
Quantum computers are divided mainly into gate quantum computers and quantum annealers. Although gate quantum computers offer the same versatility as ordinary computers, there are various challenges that need to be overcome before they can achieve practical application, which could therefore take many years. The practical application of quantum annealers, on the other hand, despite hardware-imposed restrictions on what they can do, is now within reach, and research institutions and companies around the world are racing to develop commercial quantum annealers.
Central Research Laboratories Vice President Yuichi Nakamura, who is in-charge of R&D of quantum computers at NEC, says, “Quantum annealing machines are excellent at rapidly solving combinational optimization problems. Complex real-world problems are solved by using AI machine learning to generate as many choices as possible in the “first half,” and then using quantum annealing to solve combinational optimization problems based on those choices in the “second half” of the process. This collaboration makes it possible to find answers at ultrafast speed.”
In 2011, the world’s first commercial quantum annealing machine was developed by a venture company in Canada, enabling the solution of a certain level of combinational optimization problems under a fixed set of conditions. VP Nakamura, however, claims that: “Rapidly responding to the complexity of the real world requires a machine that can stably operate at a scale of at least 10,000 qubits. Although the technology NEC is working on right now is only for one qubit, it is of pivotal importance because this new method lays the groundwork for coupling multiple, significantly “superior-quality” qubits. This will in turn make it possible to expand the scale in terms of faster speed, paving the way for advanced practical application of quantum annealing machines. I think we can say that our quantum research efforts have finally come to fruition.”
Backed by more than 20 years of experience in quantum computing research, NEC opens up future possibilities through quantum annealing
In 1999, NEC succeeded in demonstrating solid-state qubit operation in the quantum superposition state for the first time through a superconducting device. As it continued its research on ways to control the superposition state, NEC at some point shifted to quantum annealing as the main focus for generating a suite of outcomes from its research efforts. Last October 2018, the industry-academia collaboration on “R&D of quantum annealing technologies using superconducting parametron devices” between NEC, Tokyo Institute of Technology, Waseda University, and Yokohama National University has been selected for implementation under the New Energy and Industrial Technology Development Organization (NEDO)’s project on next-generation computing technology development.
Also, in regard to applications and software for the quantum annealing machine, NEC is closely involved in the “R&D project on common software platforms for Ising machines” led by Waseda University, as part of NEC’s efforts to continue research towards the mutual optimization and robust integration of software design and hardware design.
Anticipating further progress in research on both hardware and software aspects, NEC has decided to increase its research workforce in this area. Moreover, it plans to offer the suite of outcomes from the research by embedding them to a dedicated engine under NEC the WISE, the company’s portfolio of cutting-edge AI technologies. NEC aims to contribute to society through the solution of various problems in the real world through open innovation in a wide range of disciplines with institutions within and outside Japan.
Through ultrafast analysis of combinational optimization problems, quantum annealing machines will make it possible to carry out simultaneous and parallel analysis of combinations at unprecedented orders of magnitude, even for problems in unexplored realms that researchers and engineers have thus far considered as “intractable.” They therefore hold great potential for solving a wide array of real-world problems.
VP Nakamura concluded by saying, “In other words, it would become possible to shoot an infinite number of “poor shots” to “hit” at answers we have not expected to arrive at before. Definitely, some of these answers would bring about novel problem-solving methods and breakthroughs in value creation. What are the new areas that we should focus our initiatives on in the unknown world that the quantum annealing machine opens up? And how can we leverage the results of those initiatives to create a better society? NEC hopes to work together with its customers in the quest for such new “realizations.”