- Research
Study Meeting on Formalization of Mathematics
In recent years, proof assistants such as Lean have come to play an increasingly important role in the very methods of mathematical research. At the same time, technologies that enable AI to generate Lean code automatically are also advancing rapidly, and it is becoming less necessary for mathematicians themselves to write all of the code. However, the ability to read formalized descriptions and understand what they mean mathematically, and what they prove under which assumptions, will, we believe, become all the more important.
In response to this background, the ZEN Mathematics Center at ZEN University has for some time been conducting research on the formalization of mathematics and on AI and mathematics. As part of these efforts, on Thursday, February 4, 2027, we will hold a one-day study meeting on Lean and the formalization of mathematics in Minamisunamachi, Tokyo, entitled “Study Meeting on Formalization of Mathematics (A Study Meeting on the Formalization of Mathematics).”
This study meeting is intended as an opportunity to understand the principles of Lean from the perspective of mathematical researchers and to think together about what the work of formalizing mathematics actually entails. Rather than focusing only on how to manipulate code, we will treat, in a single coherent progression, how Lean represents mathematical concepts, propositions, and proofs; by what mechanism the correctness of such representations is guaranteed; and even how the international mathematical library known as mathlib is constructed.
In the morning, we will offer “Introduction to Lean: Digest Version,” a roughly three-hour reorganization of the material from the study meeting “Introduction to Lean: Reading Code from First Principles and Basic Implementation,” held at ZEN University in September 2026. In the afternoon, we will welcome Johan Commelin, Director of the Mathlib Initiative and a member of Project LANA; Adam Topaz, Senior Manager of the Mathlib Initiative and also a member of Project LANA; and Kiran Kedlaya, a distinguished mathematician and member of Project LANA. They will give reports and take part in discussions on the mathematics inherent in formalization itself, the collaborative development of mathlib, Lean in university education, and current issues surrounding AI-based automated formalization and automated theorem proving.
We would be delighted to welcome not only those who wish to learn Lean, but also those who do not plan to write Lean code themselves yet are interested in the formalization of mathematics and in the changes that AI is bringing to mathematical research.
Event Details
• Date and time: Thursday, February 4, 2027, 9:00–17:00 (tentative)
• Venue: L stay & glow Minamisunamachi, Minamisunamachi, Tokyo (Tokyo Metro Tozai Line)
7-10-14 Minamisuna, Koto-ku, Tokyo 136-0076, Japan
https://lstayandgrow.jp/minamisuna/
• Format: In person
• Intended participants: Mathematicians, mathematical researchers, graduate students, and undergraduate students (participants with no prior experience with Lean are also welcome)
• Participation fee: Free
• Capacity: Approximately 80 participants
• Organizer: ZEN Mathematics Center (ZMC), ZEN University
Tentative Program (subject to change)
• “Introduction to Lean: Digest Version” (Japanese; Naoya Umezaki (ZEN University) et al.)
Unlike an ordinary programming course, this session begins with the “principles and structure” of how Lean takes in mathematics and why it can mechanically check proofs. We will cover, in sequence, type theory, the correspondence between propositions and types, proof terms, checking by the kernel, how definitions are constructed, and simple code examples, and will learn the foundations needed to “read” Lean code accurately.
• What Does It Mean to Formalize Mathematics? (English; Adam Topaz)
Simply transcribing a paper mechanically into Lean syntax does not produce a reusable and reliable formalization. We will consider the mathematical judgments and discoveries involved in the process of formalization, including the choice of definitions, the level of abstraction, connections with existing libraries, and how to generalize.
• How Is mathlib Built? (English; Johan Commelin)
We will explain how Lean’s enormous mathematical library, mathlib, is built and maintained through international collaborative development. We will also cover the role of maintainers, the workflow for contributing new definitions and theorems, and review and quality control.
• Lean in University Education (English; Kiran Kedlaya)
Drawing on Lean courses taught at the University of California, San Diego, we will report on course design, student responses, and learning outcomes, and consider the possibility of incorporating Lean into university education in Japan.
• In addition, we are considering holding a discussion session on AI-based automated formalization and automated theorem proving, as well as on the future of mathematics in the age of AI.
Registration
If you wish to participate, please register using the Registration form below.
Registration form: https://forms.gle/9Sg6q88bs7PsdUsv7
The formalization of mathematics is not simply the task of translating mathematics into the language of computers. It is also a task of reconsidering not only mathematical aspects—such as which definitions to choose, which conclusions to derive from which assumptions, and what to share as mathematical objects—but also our relationship with the community, including how much reusability we should aim for. We hope that this study meeting will create a new forum for discussion between those who already use Lean and those who are now beginning to think about formalization.
We sincerely look forward to your participation.
Fumiharu Kato
ZEN Mathematics Center (ZMC), ZEN University