ΑΙhub.org
 

Maryna Viazovska’s proofs of sphere packing formalized with AI


by
27 April 2026



share this:

Maryna Viazovska. Credit: EPFL 2026.

The proofs that earned EPFL professor Maryna Viazovska the Fields Medal in 2022 have reached a new milestone: their complete formalization by computer, achieved through a collaboration between mathematicians and artificial intelligence tools.

In 2016, Maryna Viazovska solved the sphere packing problem in dimension 8, proving that the E₈ lattice constitutes the densest possible arrangement. Shortly after, together with collaborators, she established an analogous result in dimension 24 using the Leech lattice. Her method provided an elegant solution to a problem studied for centuries, with close ties to applied fields such as error-correcting codes.

For this major contribution, Viazovska was awarded the Fields Medal in 2022, the highest distinction in mathematics. Her proof was swiftly accepted by the scientific community, but formally verifying it by computer represents a challenge of a different kind: it requires translating every step of the reasoning into a logical language that software can check automatically.

The project Formalising Sphere Packing in Lean, launched in 2024 following a meeting in Lausanne between Viazovska and young mathematician Sidharth Hariharan, took on this task. Working with several international researchers, the team constructed a detailed blueprint of the dimension-8 proof and progressively translated it into Lean, a proof assistant widely used in mathematics. A specialized AI, Gauss, developed by startup Math, Inc., then played a decisive role in the formalization. By helping to complete certain intermediate steps, it accelerated the process and enabled the dimension-8 case to be wrapped up in five days, followed by the considerably larger dimension-24 case — over 200,000 lines of code — in two weeks.

This international project illustrates the rapid progress of formal verification and could mark a turning point in the collaboration between mathematicians and AI systems for verifying and constructing complex mathematical proofs.

Find out more

Watershed Moment for AI–Human Collaboration in Math – Twenty-first-century Fields Medal proof formalized for the first time, IEEE Spectrum.




EPFL

            AUAI is supported by:



Subscribe to AIhub newsletter on substack



Related posts :

#RoboCup2026 – humanoid league day 1

  02 Jul 2026
In the first of our round-ups from the humanoid league we introduce the competition, and report some preliminary results.

Adaptive parallel reasoning: the next paradigm in efficient inference scaling

  02 Jul 2026
A detailed analysis of recent progress in the field of parallel reasoning.

Scientists develop new method to generate protein datasets for training AI

  01 Jul 2026
AI is only as good as the data used to train it, and in some areas of protein engineering, the right data is hard to come by.

What’s coming up at #RoboCup2026?

  29 Jun 2026
Find out what's in store at this year's international competition.

AI model used to generate complete models of proteins in motion

  26 Jun 2026
Researchers have used a neural network to create all-atom models of proteins, as well as the dynamic movements that govern their function.

Three ways to avoid being fooled by AI slop

  24 Jun 2026
Global society makes billions of images and uploads hundreds of thousands of hours of video on the internet every day. The problem is, some of this content is misleading or downright wrong.

Engineering Out Loud: S13E1 – How many robots can a single human supervise?

  22 Jun 2026
Professor Julie Adams describes the research showing that one person can supervise more than 100 autonomous ground and aerial robots.

Everything, eco-where, AI at once?

Laura Martinez Agudelo builds on her research of visual representations of ecology and digitalisation to explore how "AI eco-imagery" is portrayed.



AUAI is supported by:







Subscribe to AIhub newsletter on substack




 















©2026.05 - Association for the Understanding of Artificial Intelligence