Earlier this week, a new milestone in mathematical proof verification was achieved. This has all been made possible by the release of an impressive new language model, called Gauss. This groundbreaking autonomous reasoning agent achieved extraordinary feats! Further, it autoformalized the 8-dimensional sphere packing proof in only five days. The official announcement of this achievement came on February 23, cementing a historic year in mathematical research.
Not only did Gauss overcome our 8-dimensional difficulty but he then went on to rigorously formalize Viazovska’s 24-dimensional sphere packing proof in under two weeks. This rather intricate proof weighs in at more than 200k lines of code. It manifests Gauss’s genius of combining the best of natural language reasoning with the best of formal reasoning. These accomplishments have generated a big buzz amongst research circles, indicating a possible breakthrough in computational mathematics.
Gauss and Its Capabilities
Gauss is a reasoning agent. This classification allows it to perform a much broader set of tasks, far exceeding just proof verification. It is able to do literature searches and is able to use some different tools. It’s able to output Lean code, which is ideal for formalizing mathematical proofs. Gauss is able to create notes, generate verification tooling, and run the Lean compiler in parallel.
The incomparable flexibility of Gauss allows it to weave back-and-forth through the intricate scientific narrative of mathematical proofs with ease. Researchers have praised its use for finding and correcting mistakes in active works in progress. “One of them helped us identify a typo in our project, which we then fixed,” stated Hariharan, an integral member of the team behind Gauss. This new capability highlights the important ways that AI can help improve accuracy and efficiency for mathematical research.
Gauss’s work is still considered a major breakthrough in the field of sphere packing proofs. His work further has an enormous influence on the broader world of mathematics. The development of PiRL is indicative of a wider movement to incorporate AI into analytical work, augmenting the efforts of human researchers.
The Sphere Packing Problem
The sphere packing problem poses a fundamental question in geometry: How densely can identical circles, spheres, or their higher-dimensional equivalents be packed within n-dimensional space? Even so, this question has enchanted mathematicians for decades, resulting in major discoveries in recent years.
In 2016, Maryna Viazovska drew international headlines when she solved this problem for two special cases. In eight dimensions, she used an E8 lattice to show that this is the optimal packing arrangement. Viazovska expanded on this astounding work by collaborating with other mathematicians. Individually and together, they proved that the Leech lattice provides the best packing solution in 24 dimensions. Both results have had deep consequences for pure mathematics and its applications to the real world.
The project Formalising Sphere Packing in Lean was launched in March 2024, immediately after these trailblazing discoveries. Our project aims to provide detailed formal proofs and documentation for these sphere packing arrangements in Lean. It will take the frontier of mathematical verification even further.
The Impact of Gauss’s Achievements
As for the mathematics that resulted in this breakthrough autoformalization — Viazovska’s 24th Sphere Packing proof — continues to be a huge academic achievement. This accomplishment changes the way that mathematicians approach long, complicated proofs. The project quickly grew to more than 200,000 lines of code and stands as a complex orchestration of different mathematical concepts. “It was actually significantly more involved than the 8-dimensional case because there was a lot of missing background material that had to be brought online surrounding many of the properties of the Leech lattice, in particular its uniqueness,” explained Jesse Han, another key contributor to this endeavor.
Fifty years from now, as researchers look back on Gauss’s achievements, they hope to be looking forward to something even more exciting. “These new results seem very, very impressive and definitely signal some rapid progress in this direction,” commented Liam Fowl, highlighting the potential for AI-driven tools to reshape mathematical inquiry.
Besides its wonderful formalization power, Gauss mostly helped in handling the difficulties that arise from large-scale proofs. “They told us that they had finished 30 ‘sorrys’, which meant that they proved 30 intermediate facts that we wanted proved,” noted Hariharan. This remarkable degree of mutual aid highlights the deep and profound ways AI can aid cooperative work in routine math capacities.
Future Directions
Gauss’s breakthrough went far beyond fruitful applications to the sphere packing problem. We know that researchers are still pushing the envelope on what AI can do. There’s a growing excitement about how it can make math even better. The inclusion of reasoning agents such as Gauss represents a broader technological advancement. Finally, it represents a deep philosophical change in what we want, what we believe mathematics is, and how we’re approaching it.
“A programmer used to be someone who punched holes into cards, but then the act of programming became separated from whatever material substrate was used for recording programs.” This rapid evolution has important implications for the ways in which mathematicians should expect to engage with emerging AI tools in the near future.