Packing Problems
Pack geometric objects into a container without overlap, making them as large or as numerous as possible. This hub indexes every packing family on the site, with deterministic verification, literature citations, and optimality status for each instance.
What a packing problem is
A packing problem specifies three elements: the objects being placed (equal circles, unequal circles, rotatable squares, lines through the origin), the container (a square, disc, L-shape, cross, semicircle, sector, or triangle), and the objective to maximize (common radius, sum of radii, common side length, or minimum angle). For most parameter values, the optimal arrangement remains an open problem.
Checking a candidate layout is fast and exact, whereas finding the best layout is computationally hard. Anyone can submit coordinates; the verifier scores them in milliseconds using exact integer arithmetic, and any valid improvement over the standing record takes the lead immediately.
Packing, covering, dispersion and energy
Related geometric optimization families differ in their constraints: packing forbids overlap and maximizes size; covering forbids uncovered gaps and minimizes size; dispersion places points to maximize the minimum pairwise distance; energy problems minimize a pairwise potential sum. The site features all four categories; this page focuses on packing.
How machine verification works
Every submission consists of fixed-point decimal coordinates. The verifier converts them into exact integers and checks containment and non-overlap using BigInt arithmetic — no floating-point rounding, no square roots, and no approximate tolerances. Every certificate produces the exact same score on any machine.
External benchmarks fall into two distinct categories: a proven optimum has a mathematical proof that no better arrangement exists, whereas a best-known value is simply the strongest published construction so far. Every benchmark links directly to its source.
Circles
The classic circle-packing problems: place n circles inside a container. Most fix the container and maximize a shared radius; P57 allows unequal radii, while P66 optimizes the enclosing triangle.
Circles of radius 1…n
Circle radii are fixed at 1, 2, …, n; the goal is to arrange them so the enclosing container is as small as possible.
Tilted squares
Allowing each square to rotate freely expands the search space from positions alone to positions and orientations, leaving most instances open.
High-dimensional and abstract
Packings of directions, lines, or subspaces in higher dimensions. Verification still runs in exact integer arithmetic, benchmarked against classical spherical and Grassmannian packing tables.
Latest breakthroughs
Public datasets
Every problem family exposes a stable JSON endpoint containing all sub-problems, authoritative scores, proof statuses, coordinates, citations, and verifier versions:
GET https://minmaxarena.com/data/square-circle-packing.json
Published under CC BY 4.0 for free use with attribution. Because live scores update as records fall, please include the generatedAt timestamp when citing.