Model-Based and Sample-Efficient AI-Assisted Math Discovery in Sphere Packing
Abstract
A model-based approach combining Bayesian optimization and Monte Carlo Tree Search improves upper bounds for sphere packing in dimensions 4-16 by formulating SDP construction as a sequential decision process.
Sphere packing, Hilbert's eighteenth problem, asks for the densest arrangement of congruent spheres in n-dimensional Euclidean space. Although relevant to areas such as cryptography, crystallography, and medical imaging, the problem remains unresolved: beyond a few special dimensions, neither optimal packings nor tight upper bounds are known. Even a major breakthrough in dimension n=8, later recognised with a Fields Medal, underscores its difficulty. A leading technique for upper bounds, the three-point method, reduces the problem to solving large, high-precision semidefinite programs (SDPs). Because each candidate SDP may take days to evaluate, standard data-intensive AI approaches are infeasible. We address this challenge by formulating SDP construction as a sequential decision process, the SDP game, in which a policy assembles SDP formulations from a set of admissible components. Using a sample-efficient model-based framework that combines Bayesian optimisation with Monte Carlo Tree Search, we obtain new state-of-the-art upper bounds in dimensions 4-16, showing that model-based search can advance computational progress in longstanding geometric problems. Together, these results demonstrate that sample-efficient, model-based search can make tangible progress on mathematically rigid, evaluation limited problems, pointing towards a complementary direction for AI-assisted discovery beyond large-scale LLM-driven exploration.
Community
AlphaEvolve is amazing and tackles hard-to-solve and easy-to-evaluate problems!
But, many math problems are actually hard-to-solve and hard-to-evaluate!
Here, we can't do much trial and error; we need something more efficient - because trying one solution can take DAYS!
We propose a framework for that and use it in sphere packing; Hilbert's 18th problem!
We recover new bounds not known before! 🥳🥳
The amazing part is we don't use LLMs! We used model-based search!
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Quantum Alternating Direction Method of Multipliers for Semidefinite Programming (2025)
- Random-Key Metaheuristic and Linearization for the Quadratic Multiple Constraints Variable-Sized Bin Packing Problem (2025)
- On the Strength of Linear Relaxations in Ordered Optimization (2025)
- Sparse Multiple Kernel Learning: Alternating Best Response and Semidefinite Relaxations (2025)
- Exponential Speed-ups for Structured Goemans-Williamson relaxations via Quantum Gibbs States and Pauli Sparsity (2025)
- Advanced Cutting-Plane Algorithms for ACOPF (2025)
- A Markov Decision Process for Variable Selection in Branch & Bound (2025)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment:
@librarian-bot
recommend
Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper



