gpt 6 astra ultra attempt on Math-1031

Average diameter of a bounded cell of a simple arrangement

Definitions and mathematical statement

Fix integers $d\ge1$ and $n\ge d+1$. A simple affine hyperplane arrangement in $\mathbb R^d$ is a family $\mathcal A$ of $n$ hyperplanes in general position: every $j\le d$ distinct hyperplanes meet in an affine subspace of dimension $d-j$, and no $d+1$ have a common point. The components of their complement are open polyhedral regions. The closures of the bounded regions are full-dimensional convex polytopes, called the bounded cells.

For such a cell $P$, let $\delta(P)$ be the diameter of its edge graph: the maximum, over pairs of vertices, of the smallest number of polytope edges in a path joining them. This is a combinatorial graph distance, not a Euclidean distance between points. List all bounded cells as $P_1,\ldots,P_I$ and define \[ \overline\delta(\mathcal A)=\frac1I\sum_{j=1}^I\delta(P_j). \] The general-position assumptions give $I=\binom{n-1}{d}>0$, so the average is defined and weights every bounded cell equally.

The conjecture states that $\overline\delta(\mathcal A)\le d$ for every such arrangement, independently of $n$. Individual cells are allowed to have diameter greater than $d$; the requirement bounds only the mean over all bounded cells. Unbounded regions are omitted, and neither cell volume nor number of vertices changes its weight in the average. Establishing a bound for just one selected polytope therefore does not settle the arrangement statement.

Short English statement

Is the average graph diameter of all bounded cells of a simple hyperplane arrangement at most the ambient dimension?

Research attempt: planar incidence accounting and its dimensional limit

Scope and literature check.

GPT-6 Astra Ultra, 1 October 2026. Diameters are graph distances and the average gives each bounded cell equal weight. Deza--Xie [S3] prove the conjecture in dimension two and determine its exact extremal value there; they also give higher-dimensional estimates and constructions. Thus the planar bound below is a reconstruction of a weaker elementary estimate, not an improvement or a newly solved case. The Garden URL [S2] did not load in this check, so the research paper is used for the literature qualification.

1. An elementary bound tending to two in the plane.

Let $d=2$ and $n\ge3$. Each of the $n$ lines has $n-1$ distinct intersection points, because no two lines are parallel and no three are concurrent. These divide the line into two rays and exactly $n-2$ bounded elementary segments. Thus the arrangement has \[ E_b=n(n-2) \] bounded edges. Every side of a bounded polygonal cell is one of these edges, and each edge borders at most two bounded cells. If the bounded cells have $m_1,\ldots,m_I$ sides, then $\sum_jm_j\le2E_b$.

The edge graph of a convex $m$-gon is a cycle, of diameter $\lfloor m/2\rfloor$. To check this formula, the two routes between any pair of vertices have lengths summing to $m$, so the shorter is at most $\lfloor m/2\rfloor$; a pair separated by that many cycle edges attains it. Using $I=(n-1)(n-2)/2$, we obtain \[ \overline\delta\le\frac1{2I}\sum_jm_j \le\frac{n(n-2)}I=\frac{2n}{n-1} =2+\frac2{n-1}. \] This controls the whole average; no bound on the largest individual polygon diameter was assumed. It remains too large by $2/(n-1)$ to prove the exact desired inequality.

2. Isolate the precise missing incidence saving.

For a bounded edge $e$, let $b(e)\in\{0,1,2\}$ be the number of bounded cells incident to it. Define $D=\sum_e(2-b(e))=2E_b-\sum_jm_j$, a nonnegative deficit. Let $O$ be the number of bounded cells with an odd number of sides. The cycle formula gives the exact identity \[ \overline\delta =\frac{\sum_jm_j-O}{2I} =\frac{2n(n-2)-D-O}{(n-1)(n-2)}. \] Consequently the target $\overline\delta\le2$ is equivalent, for these planar arrangements, to \[ D+O\ge2(n-2). \] This exposes the information erased by the crude double counting: edges not counted twice and the half-unit diameter savings from odd polygons. It is a checkable equivalent condition, not a proof of its lower bound. The cited planar theorem provides the stronger established answer by using additional arrangement structure.

3. A boundary case and the failed extension.

In dimension one, the hyperplanes are $n$ distinct points on a line. The $n-1$ bounded cells are intervals, each with a two-vertex, one-edge graph; their average diameter is exactly one. This checks the dimension and cell-count conventions.

In higher dimensions it is invalid to replace a cell graph by a cycle or identify diameter with half its number of facets. For example a $d$-simplex has $d+1$ facets but complete graph on $d+1$ vertices and diameter one. General simple polytopes have more complex graphs. A count of facet incidences therefore needs an independently proved relationship with shortest paths; the planar identity supplies none. This is the first obstruction to extending this calculation.

Verification.

For three lines there is one triangle, $E_b=3$, $D=3$, $O=1$, and the exact formula gives $(6-3-1)/2=1$, as required. The denominator is positive for $n\ge3$, the permitted planar range. All sums omit unbounded cells, but bounded edges adjacent to them are retained in $D$ with the correct deficit. No Euclidean distances, cell volumes, or unsupported general diameter theorem enter the proof.

Final status

UNRESOLVED as an independent proof in arbitrary dimension. The one-dimensional case, planar asymptotic estimate and exact planar deficit reformulation are verified. The full planar result is already known from [S3]; no new solution or global percentage is claimed.

Sources

• S1 ULAM.AI and the UnsolvedMath Contributors, supplied problems.json, record 3075 (OPG-605), OpenGarden collection. Catalogue identity and source transcription; CC BY 4.0. https://huggingface.co/datasets/ulamai/UnsolvedMath.
• S2 Open Problem Garden, Average diameter of a bounded cell of a simple arrangement. Original problem and discussion; the mathematical formulation below is expanded from the supplied source record. http://www.openproblemgarden.org/op/average_diameter_of_a_bounded_cell_of_a_simple_arrangement.
• S3 A. Deza and F. Xie, Hyperplane Arrangements with Large Average Diameter, arXiv:0710.0328, 2007. Original research checked 1 October 2026. https://arxiv.org/abs/0710.0328.