Quantum Annealing from Keio's Faculty of Science and Technology Challenging the Frontier of Physics and Information Science

The Joy of Pioneering the Frontier between Physics and Information Science

Perfectly harnessing the essence of quantum physics

From the moment we wake up in the morning, we make a series of choices on what clothes to wear, what to eat for breakfast, what route to take to school or work, and more. We make these choices based on various criteria such as how something looks, how efficient it is, and how safe it is. As Associate Professor Shu Tanaka explains, “The task of selecting the most optimal choice from a large number of options based on a set criteria is called a ‘combinatorial optimization problem.’”

There are a countless number of these types of problems in industry and in one’s own work, such as figuring out what is the most efficient order for carrying out a task. These problems can arise when making deliveries, operating machinery and equipment, and many other situations. One of the difficulties of this type of problem is that the number of combinatorial options explodes exponentially when there are, for example, more places to deliver a shipment. “It is there,” he says, “that we needed to devise a smarter way, a better algorithm, to solve these problems. This is where a quantum computing technique called quantum annealing comes in. It holds great potential in solving combinatorial optimization problems at high speeds, and has been in the spotlight recently" (Fig. 1).

Quantum annealing is a computing technique where one can efficiently arrive at the solution of an optimization problem. This type of quantum computing is currently garnering attention and can be positioned at the intersection of physics and information science.
“To optimize something is to ‘maximize’ or ‘minimize’ it,” says Tanaka. “In either case, the optimization is handled the same way mathematically. Here, we can think in terms of minimization. In physics the lower an energy state something has, the more stable it is said to be. Quantum annealing is designed to apply this idea to combinatorial optimization problems. In other words, you can think of it as finding the most efficient solution to a combinatorial optimization problem by seeking out a stable state with low energy.”
“Annealing” is a term that comes from metallurgy and materials science. To form an alloy, the metal atoms that make up the alloy are broken apart by heat, and the temperature is slowly lowered. This produces a stable molecular arrangement. The method of making computations scientists call “simulated annealing” takes its cue from this process. However, while simulated annealing uses heat, quantum annealing uses quantum fluctuations instead. With quantum fluctuations, the values written to a qubit are quantum superpositions of 0s and 1s. As the quantum fluctuation is weakened from that state, those 0s and 1s settle and become stable, and the values generated at that time mark the solution to the combinatorial optimization problem.
As Tanaka explains, “Quantum annealing is a computing method that was published in a 1998 paper by Professor Hidetoshi Nishimori of the Nishimori Group at the Tokyo Institute of Technology (now the Institute of Science Tokyo as of October 2024), which I joined for my research in 2002. The Nishimori Group was truly a laboratory that approached information science with physics.”
Methods such as quantum annealing that use the mechanisms of natural phenomena to compute something are called “natural computing.” Quantum annealing, in particular, has become popular among scientists, as it was the method used by the Canadian quantum computing company D-Wave to deploy the world’s first commercial quantum annealer in 2011.

To use the quantum annealing computing technique, the problem needs to be formulated as a function for the Ising model or Quadratic Unconstrained Binary Optimization (QUBO) model. However, while some problems can be simple to solve using such functions, others can be difficult. Some problems cannot even be formulated as functions in the first place.
“When tackling a combinatorial optimization problem related to a social issue, the first step is to formulate a function by referring to past research cases and looking for similar types or extrapolating from what existed at the time, since each problem already has some kind of fixed type. I think that the ability to think about how to formulate functions when encountering completely new types of combinatorial optimization problems is one of our strengths as professional researchers,” he said. With his wealth of experience and a keen intuition, Tanaka has worked with a variety of companies to implement optimization strategies in the distribution of advertisements, travel planning, transportation, and integrated circuit design. Now, he is focusing on combining AI and quantum annealing. It was during his doctoral program that he decided to fix his gaze on this field in particular.
“It all started when I was talking with a friend who was working on research in informational science, which got me wondering if I could do something interesting by combining machine learning and quantum annealing. In 2009, I presented a paper on this at an international conference on artificial intelligence. For a while after I received my degree, I did very little quantum annealing research, but then I was contacted by a company that expressed interest in this paper, and that led to a return to that topic,” he said.
After repeated discussions with researchers at the University of Tokyo and the National Institute for Materials Science, he came up with and developed a tool called Factorization Machine with Quantum Annealing (FMQA), which uses machine learning to convert problems into the Ising model or a QUBO format. Currently, he is hard at work developing various applications using FMQA, as well as working on R&D to advance it further (Fig. 2).

“For example,” explains Tanaka, “If you wanted to develop a new material that conducts electricity well by mixing several substances, you would first mix those substances in various proportions, measure the current passing through, and produce data showing that the ratio of this mixture conducts this much electricity. With FMQA, you would use machine learning based on those sets of data to create functions to be utilized with the Ising model or QUBO. A quantum annealing machine would then be used to search for the optimal solution to this function. The solution that emerges at this time would indicate the mixing ratio that should be tried next in the experiment.”
In other words, quantum annealing replaces the “intuition” of experimental scientists who look at the results of an experiment and link them to the next experiment. FMQA is run again based on the results of the experiments conducted according to this. As this process is repeated, the accuracy of the functions gradually improves, and it will eventually be possible to find a ratio with the desired conductivity. “We are working with companies and universities to explore the materials we can use and develop other applications using this method,” he adds (Fig. 3).

His paper on using FMQA was published in 2020, and it seems that people in various fields besides the materials field have expressed interest in it. “There have been quite a few companies who have approached us and said, ‘While we can’t announce it publicly yet, we are conducting research using FMQA,’” he says. FMQA is an optimization tool utilizing repeated trial-and-error experiments that lead us to a scientific discovery; in other words, it is a black box optimization algorithm. “I myself am very much looking forward to seeing the many different examples of ‘black box optimization’ that will emerge through the application of FMQA.”

I was born in Edogawa Ward in Tokyo, and went to the Tokyo Metropolitan Ryogoku High School. It was also where the famous writer Ryunosuke Akutagawa graduated before me. I guess my tendency to not worry about details is what makes me feel I am an Edokko. I went to the Tokyo Institute of Technology (Institute of Science Tokyo from October 2024) for my undergraduate and went on to the University of Tokyo for my graduate studies. Since I did not have a lot of financial leeway, I chose to study at a public educational institution.
What I remember most about my time in university is that all of my friends who came out of private high schools in the center of the city were very knowledgeable and mature. I was anxious and eager to prove myself because I felt so far behind. When I think about it now, I wonder if everyone was desperately trying to do the same. This is probably a feeling that many people have to some degree when they are young. In the end, I realized that there is no point in rushing things, and that first and foremost, it is important to fulfill your role as a university student, for doing so will be a huge boon for the rest of your life.

It was during my doctoral studies. I was working under Professor Seiji Miyashita at the University of Tokyo, and one day I came across a paper on quantum annealing that I thought was very interesting. When I mentioned this to Professor Miyashita, he encouraged me to give researching it a shot. When you look at the word “physics,” you can see that it deals with “the physical.” It was therefore a surprise to learn that non-physical things like information can be handled by a physical phenomenon called “quantum annealing.”

Actually, no. In difficult academic fields, we often hear wonderful stories about how the results of years of research come to fruition, but not in my case. I was unable to get a permanent position for a long time and went through multiple universities. Because I worked on research topics that I encountered at each of those universities, I did not work exclusively on quantum annealing for many years.
I spent two years at the Institute for Solid State Physics in the University of Tokyo, one year at Kinki University, three years at the Department of Chemistry in the University of Tokyo’s Faculty of Science, one year at the Yukawa Institute for Theoretical Physics in Kyoto University, three years at the Waseda Institute for Advanced Study, and two years at the Green Computing Systems Research Organization in Waseda University. Including the laboratories at the Tokyo Institute of Technology and the University of Tokyo, where I spent my time as a student, the Department of Applied Physics and Physico-Informatics in the Faculty of Science and Technology at Keio University counts as my tenth department. In particular, my post at Kinki University from April 2010 was decided around February 15, about 40 days before I was set to start there. Since I was in high school, I had been working part-time to buy books and pay for entrance exams. I kept an optimistic outlook, thinking to myself that as long as I had enough to get by I could manage. However, the fact that I had not been able to find a full-time position by that point was breaking my heart as a 29 year old.
It was around 2014, while I was at Kyoto University, that I began to focus on quantum annealing research, which I had been working on as side work since I received my Ph.D. It was at that time that a company expressed interest in a paper I had published in 2009 on my research combining machine learning and quantum annealing. This helped get me serious about quantum annealing research.

Although each student’s circumstances are different, I believe that Keio University is a very good place to be. The Hiyoshi Campus is home to students from a variety of fields, not just science. Spending a period of time there while still young cultivates a broad perspective. Then as the student progresses in their schooling and moves on to the science-focused Yagami Campus, they turn to concentrate on their own specialty.
I myself have no experience studying abroad. Since arriving at Keio University, many students have told me that they are interested in studying abroad. While I feel that studying abroad is a good thing, after arriving where I am as a researcher, I would like students to first consider whether or not they are making the most of their present environment. Keio University has faculty members and colleagues all unique in their makeup, and people in the administrative department are very supportive. Even if for some reason you cannot go abroad to study or do something special, Keio students are in a good environment to learn and experience a lot. I hope that all students will absorb as much as they can there and use that experience to take a very active role in the future. And I hope I can help in any small way I can to make that a possibility.

●In the beginning, my friend took me to visit Professor Tanaka’s laboratory. The professor told me, “Why not come back again? You can visit as many times as you want until you feel this place is for you.” I did so and through the visits I felt that the vibe with the professor and students was good. I thought, “This is the only place for me!” I would like to go to graduate school and develop myself more in this world (4th year undergraduate student).

● I am currently studying the relationship between quantum annealing and physics to improve the performance of quantum computers. I am from the Shu Tanaka Group’s first cohort of students. Although studying is not my forte, Professor Tanaka taught me about the difference between study and research and showed much kindness in guiding me through the process. I thought about joining the workforce after graduation, but I went on to a doctoral program and am doing research day-to-day. I want to be a great researcher like the professor (1st year doctoral student).

● In 2020, I was in the workforce. I had the opportunity to talk to the professor about my desire to do research, and he invited me to come to Keio University, where he had set up his laboratory. I earned my Ph.D in 2023 through the Shu Tanaka Group while simultaneously working in the private sector. This was my first research endeavor in the field of physics, but my professor eliminated all the reasons why I thought it was impossible and set me on a new path. He is a student-oriented person, and no matter how busy he is, he speaks individually with every student in the lab once a week. He is kind and not only do I consult with him about my research, but also about my career (project assistant professor).