Papers
Topics
Authors
Recent
Search
2000 character limit reached

On d-stochastic measures with fractal support and uniform (d-1)-marginals, and related results

Published 9 Apr 2026 in math.PR | (2604.08505v1)

Abstract: The family P<em>d<sup>λ</sup></em>d1\mathcal{P}<em>{d}<sup>{λ</sup></em>{d-1}} of all probability measures on [0,1]<sup>d[0,1]<sup>d whose (d1)(d-1)-dimensional marginals are all equal to the Lebesgue measure λ<em>d1λ<em>{d-1} on [0,1]<sup>d1[0,1]<sup>{d-1} contains remarkably pathological elements: Working with Iterated Function Systems with Probabi-lities (IFSPs) we construct measures μP</em>d<sup>λd1μ\in \mathcal{P}</em>{d}<sup>{λ_{d-1}} of the following two types: (i) μμ has self-similar fractal support; (ii) μμ has self-similar support and models the situation of complete/functional dependence in each direction.As our main results concerning type (i) we prove, firstly, that for every d3d\geq 3 the set D<em>d\mathcal{D}<em>d of Hausdorff dimensions of the supports of elements in P</em>d<sup>λd1\mathcal{P}</em>{d}<sup>{λ_{d-1}} is dense in [d1,d][d-1,d]; and, secondly, that the subset of elements in P<em>d<sup>λ</sup></em>d1\mathcal{P}<em>{d}<sup>{λ</sup></em>{d-1}} having fractal support is dense in P<em>d<sup>λ</sup></em>d1\mathcal{P}<em>{d}<sup>{λ</sup></em>{d-1}} with respect to the Wasserstein metric. Moreover, we show the existence of an element in P<em>3<sup>λ</sup></em>2\mathcal{P}<em>{3}<sup>{λ</sup></em>{2}} of type (ii) whose support is a Sierpinski tetrahedron and study some generalizations.

Summary

  • The paper uses iterated function systems with probabilities to construct singular d-stochastic measures whose every (d−1)-dimensional marginal is uniform.
  • It proves that for every d ≥ 3, Hausdorff dimensions of supports are dense in [d−1,d], while conjecturing that every value in this interval occurs.
  • It shows fractal measures with non-integer support dimensions are Wasserstein-dense, including completely dependent examples such as a Sierpiński tetrahedron with uniform bivariate marginals.

Background and motivation

A probability measure μ\mu on [0,1]d[0,1]^d is dd-stochastic if all univariate marginals are uniform; such measures correspond bijectively to dd-dimensional copulas. The paper studies the strictly smaller family Pdλd1\mathcal{P}_d^{\lambda_{d-1}} of dd-stochastic measures whose every (d1)(d-1)-dimensional marginal equals λd1\lambda_{d-1}, the Lebesgue measure on [0,1]d1[0,1]^{d-1}. Since copulas are Lipschitz continuous, one might expect their underlying measures to distribute mass regularly; Fredricks, Nelsen and Rodríguez-Lallena already refuted this by constructing doubly stochastic measures whose support has any prescribed Hausdorff dimension s(1,2)s \in (1,2) [FNRL]. The present work extends that result to uniform [0,1]d[0,1]^d0-marginals — a substantially stronger constraint — using Iterated Function Systems with Probabilities (IFSPs) built from generalized transformation matrices in the spirit of Trutschnig–Fernández Sánchez [TFS].

IFSP construction and preservation of marginals

The authors fix [0,1]d[0,1]^d1 and consider probability distributions [0,1]d[0,1]^d2 on [0,1]d[0,1]^d3 (generalized transformation matrices) satisfying positive column sums in every coordinate. Each such [0,1]d[0,1]^d4 induces a partition of [0,1]d[0,1]^d5 into rectangles [0,1]d[0,1]^d6, affine contractions [0,1]d[0,1]^d7, and an IFSP fulfilling the open set condition. Its Markov operator [0,1]d[0,1]^d8 maps [0,1]d[0,1]^d9 into itself, so Banach's fixed point theorem yields a unique fixed point dd0.

The central technical device is the uniformity condition: for coordinate dd1,

dd2

Lemma 4.2 shows that if dd3 and dd4 satisfies this condition w.r.t. coordinate dd5, then dd6; iterating and passing to the weak limit gives the first main structural theorem: if dd7 fulfills the uniformity condition for every coordinate, then all dd8-dimensional marginals of dd9 equal dd0. A corollary adds singularity: whenever some cell has dd1, the support of dd2 is Lebesgue-null (volume shrinks geometrically under the Hutchinson operator), so dd3 is singular w.r.t. dd4 while retaining uniform dd5-marginals.

Completely dependent measures on self-similar supports

The paper next constructs elements of dd6 exhibiting complete dependence in every direction — each coordinate is almost surely a function of the remaining dd7 coordinates. A simple non-fractal example is dd8 with dd9; the authors note they could not locate this example in the literature despite suspecting it is known.

The main construction uses matrices Pdλd1\mathcal{P}_d^{\lambda_{d-1}}0 (equal side lengths Pdλd1\mathcal{P}_d^{\lambda_{d-1}}1, uniformity in all coordinates), which induce IFSPs consisting entirely of similarities with factor Pdλd1\mathcal{P}_d^{\lambda_{d-1}}2. For

Pdλd1\mathcal{P}_d^{\lambda_{d-1}}3

the resulting measure Pdλd1\mathcal{P}_d^{\lambda_{d-1}}4 satisfies three properties simultaneously: it lies in Pdλd1\mathcal{P}_d^{\lambda_{d-1}}5; it is singular with self-similar support of Hausdorff dimension exactly Pdλd1\mathcal{P}_d^{\lambda_{d-1}}6 (Pdλd1\mathcal{P}_d^{\lambda_{d-1}}7 similarities of factor Pdλd1\mathcal{P}_d^{\lambda_{d-1}}8); and it is completely dependent in each direction. Complete dependence is established via the metric Pdλd1\mathcal{P}_d^{\lambda_{d-1}}9 based on Markov kernels: the family of completely dependent measures is closed in dd0, dd1 is a contraction there, and each iterate of a completely dependent seed remains completely dependent because for every dd2 exactly one dd3 has dd4.

The case dd5, dd6 is particularly striking: the support of dd7 is a Sierpinski tetrahedron, yet the measure has uniform bivariate marginals and is completely dependent in each direction. This demonstrates that even maximal dependence structure is compatible with highly fractal support under strong marginal constraints. A second family of examples, built from cyclic permutations of dd8, extends the construction to all dd9.

Density of fractal dimensions

Combining convexity of (d1)(d-1)0 with permutation-invariance arguments, the paper proves its first main result:

Theorem. For every (d1)(d-1)1, the set (d1)(d-1)2 is dense in (d1)(d-1)3.

The proof averages permutations (d1)(d-1)4 of the base matrix (d1)(d-1)5: the convex combination (d1)(d-1)6 induces an IFSP whose number of similarities grows from (d1)(d-1)7 to at most (d1)(d-1)8 in steps bounded by (d1)(d-1)9, giving Hausdorff dimensions λd1\lambda_{d-1}0 that hit every interval λd1\lambda_{d-1}1. Since λd1\lambda_{d-1}2 is dense in λd1\lambda_{d-1}3, density follows. An explicit three-dimensional example achieves dimension λd1\lambda_{d-1}4 with uniform bivariate marginals. The authors conjecture the stronger statement λd1\lambda_{d-1}5 but leave it open.

Topological prevalence of pathological copulas

The second main result strengthens the picture from existence to genericity:

Theorem. For every λd1\lambda_{d-1}6, the family λd1\lambda_{d-1}7 of elements of λd1\lambda_{d-1}8 whose support has non-integer Hausdorff dimension is dense in λd1\lambda_{d-1}9, where [0,1]d1[0,1]^{d-1}0 is the Wasserstein-1 (Hutchinson) metric.

In fact, a remark records the stronger statement that for every open interval [0,1]d1[0,1]^{d-1}1, the elements whose support dimension lies in [0,1]d1[0,1]^{d-1}2 are dense. The proof runs the checkerboard approximations of Mikusiński–Taylor in reverse: given [0,1]d1[0,1]^{d-1}3, the discretized matrix [0,1]d1[0,1]^{d-1}4 belongs to [0,1]d1[0,1]^{d-1}5, and [0,1]d1[0,1]^{d-1}6 converges to [0,1]d1[0,1]^{d-1}7 uniformly on copulas. Replacing the seed [0,1]d1[0,1]^{d-1}8 by any [0,1]d1[0,1]^{d-1}9 with fractal support preserves both membership in s(1,2)s \in (1,2)0 (by the main structural theorem) and the Hausdorff dimension of the support (bi-Lipschitz invariance plus countable stability), while Lipschitz continuity of copulas controls the discretization error within s(1,2)s \in (1,2)1 per term. Consequently, arbitrarily close to any element of s(1,2)s \in (1,2)2 — including the product measure s(1,2)s \in (1,2)3 itself — sit singular measures with fractal, even completely-dependent-type, structure.

Limitations and open questions

Several caveats qualify these results. First, the density claim for s(1,2)s \in (1,2)4 falls short of surjectivity onto s(1,2)s \in (1,2)5; whether every value in the interval is attained is explicitly conjectured but unresolved. Second, the constructions require the uniformity condition and, for the similarity-based results, the symmetric setting s(1,2)s \in (1,2)6; nothing is claimed about supports of non-self-similar type or about the extreme points of s(1,2)s \in (1,2)7, whose characterization (even for s(1,2)s \in (1,2)8) remains open. Third, the mod-s(1,2)s \in (1,2)9 sum example of complete dependence is presented without a literature reference, and the authors concede they could not verify whether it has appeared before. Finally, all results concern [0,1]d[0,1]^d00; the two-dimensional analogue with prescribed one-dimensional marginals is just [0,1]d[0,1]^d01 itself, where the FNR result applies but the [0,1]d[0,1]^d02-marginal machinery does not.

Conclusion

The paper establishes that the constrained family [0,1]d[0,1]^d03 is as pathological as unconstrained [0,1]d[0,1]^d04: it contains singular measures with self-similar supports of dense-set-many Hausdorff dimensions in [0,1]d[0,1]^d05, contains completely dependent elements supported on fractals — including a Sierpinski-tetrahedron-supported [0,1]d[0,1]^d06-stochastic measure with uniform bivariate marginals — and its fractal members are Wasserstein-dense. These findings sharpen the FNR phenomenon under much stronger marginal constraints and provide an effective IFSP toolkit for constructing explicit examples.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.