A mathematical analysis has found an exact fractal dimension for the extreme points of a random convex shape. In the model, that dimension is the ambient dimension minus one, multiplied by a stability index called alpha, almost surely. It is a theorem about this random construction, not an estimate from a finite sample with a confidence interval.
That settles the paper's central question about the Hausdorff dimension of the extreme-point set. Hausdorff dimension is a way to describe the size of an irregular or fractal set at very small scales, so the result puts a precise measure on the complexity of these points.
A random shape built from weights and directions
The object is a random countable stable zonotope, a compact convex set built from Poisson-generated weights and independent directions chosen uniformly on a sphere. The assumptions restrict the ambient dimension to at least two and the stability index alpha to values between zero and one.
No participants, observations, or empirical dataset are involved. The analyzed object is the random set itself, with conclusions stated almost surely across the model's possible outcomes.
Behind the construction is a Poisson random measure whose intensity for a segment weight or jump size r and direction x is alpha times r to the power of minus one minus alpha, combined with the uniform directional measure on the sphere.
Matching bounds close the dimension question
The geometry is recast through a random field. Exposed points, boundary points selected by a supporting hyperplane, are identified with one part of the field's image, while the extreme-point set lies between that image and the image of the full sphere. This lets the authors compare the relevant sets through one geometric parametrization.
For the lower bound, the authors study tangential projections of field increments. They combine density estimates and bounds on negative moments with Frostman's energy criterion, a test that turns suitable control of pairwise distances into a dimension lower bound. The resulting theorem gives the extreme-point set a dimension of at least the ambient dimension minus one, multiplied by alpha, almost surely.
The upper-bound argument takes a different route. It separates large jumps from small ones, uses the large jumps to create a finite arrangement of hyperplanes, and covers the resulting cells while controlling how much the small jumps can move their images. This gives an almost-sure upper bound of the same value for both the field image and the extreme-point set.
The two bounds meet at the same value, establishing the exact almost-sure dimension for the uniform-direction model. The exposed-point set, the extreme-point set, and the full spherical field image also share that almost-sure Hausdorff dimension.
Finite does not mean positive
A second result concerns the Hausdorff measure evaluated at the dimension identified above. The paper proves that this critical measure of the extreme-point set is finite almost surely.
Finiteness is not the same as positivity. The authors say the result does not establish that the critical measure is positive, and they state that it is almost surely zero when the ambient dimension is two.
The result has a defined boundary
The direct theorem assumes uniform directions. The paper says the upper-bound and critical-measure arguments extend to directional laws satisfying a uniform belt condition, while the lower-bound argument additionally needs rotational invariance and tangential non-degeneracy.
That leaves the minimal assumptions for the lower bound in anisotropic models unresolved. The paper also leaves packing dimension, Minkowski dimension, and multifractal properties of the field open.
The manuscript is an arXiv preprint dated 28 August 2026.
Paper data and sources
Original title: Hausdorff Dimension of the Set of Extreme Points of a Random Countable Stable Zonotope
Authors: Maksim Kukushkin
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-28
DOI: Not available
Original paper · Full text