In a landmark achievement for artificial intelligence and mathematical research, the proofs behind Ukrainian mathematician Maryna Viazovska’s groundbreaking work on sphere packing have been formally verified. This milestone, announced today, signifies a major leap forward in AI’s ability to not just assist, but to rigorously validate complex mathematical theorems. The implications extend far beyond abstract mathematics, potentially impacting fields like error-correcting codes crucial for modern technology.
“These results are incredibly impressive and demonstrate rapid progress in AI-assisted proof verification,” says Liam Fowl, an AI-reasoning expert and postdoctoral researcher at Princeton University, who was not involved in the project. “The ability of AI to independently confirm these proofs is a game-changer.”
The Sphere-Packing Problem: A Centuries-Old Challenge
The sphere-packing problem, at its core, asks a deceptively simple question: how can identical spheres be arranged to occupy a given space with the highest possible density? In two dimensions, the answer is a honeycomb pattern. In three dimensions, stacking spheres in a pyramid configuration proves optimal. However, as the number of dimensions increases, finding and proving the most efficient arrangement becomes exponentially more difficult. For decades, mathematicians struggled to solve this problem beyond three dimensions.
Viazovska’s Breakthrough and its Real-World Impact
In 2016, Maryna Viazovska revolutionized the field by providing definitive solutions for the 8- and 24-dimensional sphere-packing problems. Utilizing sophisticated mathematical tools known as (quasi-)modular forms, she demonstrated that a highly symmetrical arrangement called E8 achieves the densest packing in eight dimensions, and the Leech lattice does the same in 24 dimensions. While seemingly theoretical, these findings have practical applications. For instance, the principles of sphere packing are leveraged in the design of error-correcting codes, vital for ensuring data integrity in smartphones, space probes, and other critical systems. The formal verification of these proofs solidifies their reliability and opens doors for wider implementation.
Formal verification, unlike traditional peer review, involves a computer rigorously checking every step of a proof for logical consistency. This process, while demanding, provides an unparalleled level of certainty. Since 2022, advancements in AI, particularly in the realm of large language models, have dramatically accelerated the pace of formal proof verification.
From Chance Encounter to Collaborative Project
The journey towards formalizing Viazovska’s proofs began with a serendipitous meeting in Lausanne, Switzerland, between the mathematician and Sidharth Hariharan, then an undergraduate student. Hariharan, already developing expertise in formalizing proofs, shared his approach with Viazovska, sparking her curiosity. This led to the launch of the Formalising Sphere Packing in Lean project in March 2024. Lean is a powerful programming language and “proof assistant” designed to allow mathematicians to create proofs that can be mechanically verified.
The project quickly attracted a team of experts, including Bhavik Mehta (Imperial College London), Christopher Birkbeck (University of East Anglia, England), and Seewoo Lee (University of California, Berkeley). Their initial task involved creating a detailed “blueprint” of the 8-dimensional proof, identifying which components had already been formalized and which required further work.
“We spent about 15 months building the project repository before making it publicly accessible in June 2025,” recalls Hariharan, now a first-year Ph.D. student at Carnegie Mellon University. “Shortly after, we received an unexpected communication from Math, Inc.”
AI Accelerates the Verification Process
Math, Inc., a startup specializing in AI-powered mathematical reasoning, had developed Gauss, an AI agent specifically designed to automate proof formalization. “Gauss is a reasoning agent that combines natural language understanding with fully formalized reasoning,” explains Jesse Han, Math, Inc.’s CEO and cofounder. “It can conduct research, utilize tools, write Lean code, and verify proofs – all autonomously.”
Math, Inc. previously gained recognition for using Gauss to formalize the strong prime number theorem, a feat that had previously occupied leading mathematicians for years. Applying this technology to Viazovska’s work, Gauss initially proved 30 intermediate facts within the sphere-packing project, even identifying a typo in the team’s code.
After a period of silence, Math, Inc. returned with a significantly enhanced version of Gauss. This new iteration not only autoformalized the 8-dimensional proof in a matter of days but also tackled the far more complex 24-dimensional case.
“When they informed us they had completed the 24-dimensional proof, we were astonished,” says Hariharan. “This technology has the potential to revolutionize mathematical research.”
The 24-dimensional proof, comprising over 200,000 lines of code, was completed in just two weeks. While Gauss automated the bulk of the work, both Han and Hariharan emphasize the crucial role of human contributions in laying the groundwork for this success. This achievement represents a true collaboration between humans and AI.
Han believes this is just the beginning. “This technology will ultimately free mathematicians to focus on what they do best: exploring new mathematical frontiers.”
What impact will this level of AI assistance have on the future of mathematical discovery? And how will the relationship between human intuition and artificial intelligence evolve in the pursuit of knowledge?
Frequently Asked Questions About Sphere Packing and AI Verification
What is sphere packing, and why is it important?
Sphere packing is the problem of determining the most efficient way to arrange spheres in a given space. It has applications in diverse fields, including error-correcting codes, materials science, and cryptography.
What did Maryna Viazovska prove regarding sphere packing?
Maryna Viazovska proved the optimal arrangements for sphere packing in 8 and 24 dimensions, solving a long-standing problem in mathematics.
How does AI assist in verifying mathematical proofs?
AI tools like Math, Inc.’s Gauss can automatically formalize proofs, checking every step for logical consistency and identifying potential errors.
What is formal verification, and why is it significant?
Formal verification is the process of using a computer to rigorously check the correctness of a mathematical proof, providing a higher level of certainty than traditional peer review.
What is the Lean programming language used for in this context?
Lean is a “proof assistant” that allows mathematicians to write proofs in a formal language that can be verified by a computer.
Could AI eventually discover new mathematical theorems independently?
While current AI tools primarily assist mathematicians, the rapid advancements in the field suggest that AI may eventually be capable of independent mathematical discovery.
Disclaimer: This article provides information for general knowledge and informational purposes only, and does not constitute professional advice.
Share this groundbreaking story with your network and join the conversation in the comments below! What are your thoughts on the future of AI in mathematics?
Discover more from Archyworldys
Subscribe to get the latest posts sent to your email.
