Quantum Universe EXPO

Yuma Matsumoto — Master's student (2nd year at the time), Department of Physics, School of Science, Institute of Science Tokyo

A century has passed since quantum mechanics was born. The "quantum" that once lived only in the minds of theoretical physicists is now something we can hold in the palm of our hand and actually put to work.

In this milestone year, hoping to share the fun and accessibility of "quantum" — and the world it opens up — with people in Korea, we held a problem-solving hackathon built around quantum annealing at Pusan National University. It was the second installment of the QI4U Expo series, following Italy, and I was responsible for the event's direction, coordinating with everyone involved, and serving as the MC on the day.

Running a workshop across the barriers of language and culture was a serious challenge. Despite running into all sorts of trouble, I believe we managed to create a truly exciting event together with more than 60 participants. I've tried to write this in a way that conveys what it was actually like, so I hope you'll read on.

Here we go.

"What even is quantum mechanics?" is a hard question to answer. But if you instead ask, "Could we build something fun on top of 'quantum'?", the ideas turn out to come surprisingly easily. Quantum annealing (QA) in particular was a perfect fit for a hackathon. QA is drawing attention as a solver that uses quantum effects to tackle "combinatorial optimization problems" — and these problems are hiding everywhere in our daily lives.

In a nutshell, a combinatorial optimization problem asks: "Among many possible choices, which way of choosing is the best?" For example, deciding the shortest route to visit several delivery destinations, scheduling tasks efficiently within a limited time, or working out how to allocate limited resources for the greatest effect.

A more familiar example: choosing snacks for a school trip so that you stay within a 300-yen budget while maximizing your own satisfaction is itself a fine piece of combinatorial optimization. Within a limited budget, you have to weigh the combination of "which snacks to pick" and find the best one. And of course personal demands creep in too — "I absolutely have to bring potato chips!" The more such conditions (constraints) you pile on, the harder it suddenly becomes to find the optimal combination.

And quantum annealing can use the unique properties of quantum to search for good answers to exactly these problems. The invisible "quantum" turns out to be connected to surprisingly familiar problems — from picking snacks for a school trip to city-wide logistics. Sharing that fascination with people in Korea was the starting point for this event.

Fired up to run a quantum-annealing-based problem-solving hackathon in Busan — Korea's largest port city, cradled by mountains and sea — we set out, together with the participants, to turn locally grounded ideas and approaches into something real.

Pusan National University holds the distinction of being Korea's first national university, and it boasts some of the finest educational and research facilities in the Busan area. The participants were not limited to PNU students. People came from neighboring universities, employees from Korean companies interested in quantum technology, and people from the Seoul area (Seoul National University, Korea University, Yonsei University, KAIST) who joined either by staying over or online — even a Korean resident of the United States took part remotely. The lineup was wonderfully diverse: from computer science and physics to political science, social science, business administration, and philosophy — people from across the humanities and sciences who shared an interest in quantum technology or in connecting with Japanese students.

The first day opened with an invited lecture by Specially Appointed Professor Hidetoshi Nishimori of Institute of Science Tokyo. He shared precious behind-the-scenes stories from the time he wrote the Kadowaki–Nishimori paper (published in 1998, this is the landmark paper that first proposed the method of quantum annealing; our speaker, Prof. Nishimori, is one of its authors, making it truly the origin of this field), the fundamentals of quantum annealing woven together with quantum mechanics and statistical physics, and research stories around the D-Wave machine — all pitched for newcomers. The participants' background knowledge varied widely, but they listened intently to a talk that was both fascinating and powerful.

During the Q&A, I took advantage of being an MC who can only speak plain English and asked questions that surely no one else would dare to — like "Do you like quantum mechanics?" and "What is your hope or dream for quantum annealing?" I'll keep Prof. Nishimori's answers a secret.

Early in the afternoon came a talk by our very own Ohzeki-san, titled "Quantum Annealing and its challenges," introducing the social problems we have tackled with quantum annealing along with the concrete topics and impact involved. Beyond themes like logistics and traffic control, he shared applications unique to the Ohzeki Lab with the Korean audience — estimating the sugar content of fruit, photomosaic art, matching for childcare support, and optimizing signboard placement based on crowd-flow prediction at university open-campus events.

Through a live-coding demo, Ohzeki-san also let everyone experience how easily quantum annealing and optimization problems can be handled in Python. This part was actually a last-minute addition we asked for, for various reasons, and he improvised it on the spot. Yet it turned out to be invaluable: it let us gauge the overall level and enthusiasm of the room, and it became one of the keys to the workshop's success.

After that came a string of varied talks: a presentation from the corporate side by LG CNS, which is based in Korea; a talk by Seon-Geun Jeong on improving logistics at the Port of Busan by applying the D-Wave machine; and a talk by Gilhan Kim of Yonsei University on machine learning and quantum annealing. With Prof. Nishimori and Ohzeki-san joining in, lively discussions broke out and the room was thoroughly energized.

As it happens, a student reporter who took part actually wrote the event up in an article (the piece on Japan Korea Daily is here). It even caught small moments from the day, like the surprise birthday party we threw for Ohzeki-san between lectures. But what the reporter singled out as the most striking thing was exactly this Day 1 atmosphere — professors and students talking as equals, without any barriers between them. "Seeing scholars and students debate on an equal footing is something you can rarely experience in Korea; it showed me another possible form of academic community," the article read.

Since this day was mostly lecture-based, the participants surely learned a lot — but I also figured they must be pretty worn out. I was wondering what the attendance rate would be from Day 2 onward… and to my amazement, 100% of the registered participants came back for the workshop on Day 2 and beyond!

On Day 2 we held no presentations and devoted the entire day to group work. Each group was made up of four to five Korean participants plus one Japanese supporter, and they worked on brainstorming and program development. We also had each group choose a group leader, responsible for submitting the presentation materials and steering the group's direction. Past QI4U events hadn't used this kind of leader structure, but this time I thought, on the spot, "It might be better to appoint leaders," and ran with the idea on the fly.

Here's a little behind-the-scenes story. In fact, on Day 1 we hadn't given a single lecture on how to actually build a QUBO (a mathematical formulation that rewrites a combinatorial optimization problem into a form quantum annealing can solve — building this is the crux of the programming). Our hosts were concerned about this and asked me to give a follow-up lecture at the opening of Day 2.

However, I had never intended to give a comprehensive talk on QUBO in the first place, and judging from how Day 1 had gone and the participants' eagerness — reflected in the Day 2 attendance — I felt this concern (the "how-to" of QUBO) could be absorbed within the group work itself. So in the end we began the workshop without any follow-up. This was a call based on the mindset I've cultivated in the Ohzeki Lab, the on-the-ground feel I picked up firsthand as a local supporter at last year's QI4U in Taiwan, and the experience and intuition built up over my time as a student. In hindsight I think it was the right one, though I'm honestly not sure it would be reproducible — looking back, I may have been walking a bit of a tightrope.

That said, leaving everything entirely up to the groups isn't good either. So — and this was actually Honda-san's idea — instead of a follow-up lecture, we set up individual mentoring sessions with Ohzeki-san at the start of the afternoon. The plan was to focus on brainstorming in the morning, then start shaping things into concrete form in the afternoon based on his advice. I went around interviewing each group's Japanese supporter to compile their morning's work into materials for Ohzeki-san to review — and the sheets came back packed with an astonishing number of ideas. I'd never expected so many to pour out. A happy miscalculation, as they say.

From the afternoon on, the groups settled on their direction and worked on concrete formulations, simulations, and building programs and mock-up apps in time for the next day's presentations. I myself didn't actually belong to any one group; I roamed the venue now and then, giving advice to groups that looked stuck and suggesting promising landing points whenever the Japanese supporters were struggling. Rather than becoming a player myself, I kept a bird's-eye view of the overall level and mood of the room, and used that to keep the whole event moving.

At night, all the Japanese supporters gathered at the hotel for some fun, focused development (a "mokumoku-kai," a quiet heads-down coding session). As the elite crew who had taken part in the December hackathon, everyone's drive to build was sky-high. That said, if the Japanese supporters' work were to become the group's results outright, that would run counter to the spirit of the event. We kept that in mind and made sure our development served only to help each group achieve its single shared goal.

On the morning of Day 3, each group worked on wrapping up loose ends and preparing their presentations. After everyone had a proper lunch, the group presentations finally got underway in the afternoon.

Many groups combined presentation slides with live app demonstrations, and I was endlessly impressed that such results could come out of so short a preparation period. Some went even further — using free cloud services so that other participants could access their apps, or collecting the data needed for optimization through a live survey during the presentation itself.

The themes were genuinely diverse, too. From optimizing the meat-and-side combinations to maximize satisfaction at a samgyeopsal dinner, to optimizing evacuation routes tailored to a specific university building, to a timetable-suggestion app built around the particular circumstances of PNU — one locally rooted, "only-here" idea after another kept popping up.

From among them, let me introduce just one: the topic presented by the winning team. What this team zeroed in on was the question, "How should a city's roads be arranged to make it easier to prevent crowd accidents during large-scale events?" Because the team included members studying political science and diplomacy, they advanced their discussion from a distinctive angle rooted in concern for social issues. In Korea, population concentration — especially in Seoul — is a social problem, and during events like Halloween, huge crowds gather in particular areas. In fact, in 2022, during a Halloween event in Itaewon, Seoul, a crowd became densely packed and people surged in from both ends of a narrow alley, resulting in a tragic crowd crush that claimed many lives.

This team focused on the "concentration of human flow" behind the accident. They captured the city's road network as a graph structure, modeling intersections as vertices and roads as edges. They then formulated it as an optimization problem of deciding the direction of travel for each road (which way to send the flow of people). In other words, taking the existing road network as a given, they aimed to disperse the flow of people and avoid excessive concentration on particular roads by optimizing "which roads to make one-way" and "which direction to guide people in." However, if you simply optimize, you might end up with a one-way configuration where "you can't get from one point to another." So the team designed a QUBO that built in, as a constraint, the requirement that every location remain mutually reachable, while keeping the flow of people from concentrating on any single road.

Building on these efforts and their presentation results, some groups even earned invitations to AQC (Adiabatic Quantum Computing — an international conference on quantum annealing and adiabatic quantum computation; the 2026 edition will be held at Institute of Science Tokyo). And even now, several groups are continuing their work, aiming to present there. "Oh, AQC 2026 is in Tokyo, huh" — it was an international conference I'd been watching as if it were someone else's affair. Now, the work born from this hackathon is bearing fruit on that very stage, and the participants themselves are about to step onto it. That an event lasting just three days could lead all the way to an academic presentation venue is, as the person who planned it, something that makes me genuinely happy.

This event was genuinely tough in all sorts of ways, but I gained a great deal from it, too. Above all, hearing words of thanks from everyone after the event was what made me happiest. The people of Busan were all so warm. For example—

• Many participants called out to me in all kinds of moments with "Thank you, great work!"
• Someone later emailed me to say, "I was moved by your leadership — I'd love to ask you all sorts of things."
• I became close with participants — getting excited talking about Japanese anime, and being taken out to delicious restaurants in Busan.

This event came together thanks to the cooperation of so many people.

First, my deepest gratitude goes to Professor Won-Joo Hwang, who graciously welcomed an event in this format and arranged a wonderful, well-equipped venue at Pusan National University. I'm also grateful to everyone in the Hwang Lab, who supported us wholeheartedly throughout — from the various tasks that came up on the day, to spot-on local advice, to quick, decisive communication amid a mix of Japanese, Korean, and English.

A total of eleven students from Tohoku University and Institute of Science Tokyo took part on-site. Let me take a moment to thank each of them.

At the core of the operating team, each playing to their own strengths: Yamada, Okizawa, Yoshihara, and Toyoshima of the Ohzeki Lab.

Hyukjin ("Jin") of the Sekizawa Lab at Institute of Science Tokyo, who accepted my sudden invitation without hesitation, brought his talents to bear on-site, and helped make what I wanted to do a reality.

And the undergraduates from Tohoku University and Institute of Science Tokyo — Sakakibara, Terao, Shimazaki, Takemura, and Yanagisawa — who took an interest in this initiative after "QI4U CNG," the joint Tohoku–Science Tokyo event we held last December, and who, even on their first trip abroad, showed outstanding coding and development skills. I truly believe this was an event where everyone came together as one team.

And finally, to Ohzeki-san and Honda-san, who entrusted me with leading QI4U in PNU and watched over me to the very end — thank you so much.