New Inequalities on Fractal Analysis and Their Applications

Der‐Chen Chang, Yong XuJournal of Inequalities and Applications 2007 · DOI

Abstract

Two new fractal measures, $\mathcal{M}^{*s}$ and $\mathcal{M}^s_*$, are constructed from upper and lower Minkowski contents. This paper explores their properties, demonstrating that the modified upper and lower box dimensions can be derived directly from these measures. Furthermore, the work establishes refined inequalities for the dimensions of product sets and product measures in Euclidean space.

Introduction

In fractal geometry, the characterization of sets often relies on various notions of dimension. The Hausdorff dimension ($\dim$) and the Packing dimension ($\text{Dim}$) are well-established because they are based on Borel regular measures that satisfy $\sigma$-stability—the property where the dimension of a countable union of sets equals the supremum of the dimensions of the individual sets.

Conversely, Box dimensions (upper $\Delta$ and lower $\delta$), while computationally attractive for applications in physics and engineering, lack this stability. To address this, mathematicians introduced modified box dimensions ($\hat{\Delta}$ and $\hat{\delta}$). However, a significant theoretical gap remained: unlike the Hausdorff or Packing dimensions, these modified box dimensions were not initially associated with specific, naturally occurring measures. This paper bridges that gap by constructing two new measures derived from Minkowski contents.

Methodology: From Minkowski Contents to Borel Measures

The authors construct the measures $\mathcal{M}^{*s}$ and $\mathcal{M}^s_*$ using the upper and lower Minkowski contents, respectively. The primary challenge is that Minkowski contents are not countably subadditive and thus do not qualify as outer measures.

To resolve this, the authors employ a construction technique analogous to the development of the packing measure. They define the new measures by taking an infimum over all possible partitions of a set into a countable collection of bounded subsets. Specifically, for a set $E \subset \mathbb{R}^n$, the construction ensures that the erratic behavior of Minkowski contents is smoothed into Borel regular measures. This process provides the rigorous foundation necessary to analyze the dimensions of Cartesian products of both sets and probability measures.

Key Theoretical Insights

The Measure-Dimension Bridge

A central result of this research is the proof that the dimension indices derived from these new measures are exactly the modified box dimensions. Specifically, the paper establishes:

  • $D(E) = \text{Dim}(E)$ (where $D$ is the dimension index for $\mathcal{M}^{*s}$)
  • $d(E) = \hat{\delta}(E)$ (where $d$ is the dimension index for $\mathcal{M}^s_*$)

This equivalence (Theorem 4.6) places modified box dimensions on the same measure-theoretic footing as the Hausdorff and Packing dimensions, ensuring they inherit desirable properties like $\sigma$-stability.

Refined Product Inequalities

The paper provides a more granular understanding of how dimensions behave under Cartesian products. For two sets $E$ and $F$, the authors prove a new chain of inequalities: $$\dim(E) + \dim(F) \leq \dim(E \times F) \leq \hat{\delta}(E \times F) \leq \hat{\delta}(E) + \text{Dim}(F) \leq \text{Dim}(E \times F) \leq \text{Dim}(E) + \text{Dim}(F)$$

This result is particularly significant because it highlights the interaction between the modified lower box dimension and the packing dimension, offering tighter bounds than previously known in the literature.

Applications to Product Measures

Extending the analysis from sets to measures, the authors investigate Borel probability measures $\mu$ and $\nu$. They establish a parallel set of inequalities for the product measure $\mu \times \nu$ (Theorem 5.6): $$\hat{\delta}(\mu) + \hat{\delta}(\nu) \leq \hat{\delta}(\mu \times \nu) \leq \hat{\delta}(\mu) + \text{Dim}(\nu) \leq \text{Dim}(\mu \times \nu) \leq \text{Dim}(\mu) + \text{Dim}(\nu)$$

These bounds provide a robust framework for understanding the scaling properties of combined measure spaces, which is critical in dynamical systems and multifractal analysis.

Conclusion

By constructing $\mathcal{M}^{*s}$ and $\mathcal{M}^s_*$, this work resolves a long-standing theoretical deficiency in the study of box dimensions. The resulting inequalities for product sets and measures offer a more precise toolkit for fractal analysis. Future research may extend these constructions to general metric spaces or apply these refined bounds to specific classes of fractals, such as self-affine or random sets, where box dimensions are frequently utilized.

Summary updated

Abstract

Two new fractal measures and are constructed from Minkowski contents and . The properties of these two new measures are studied. We show that the fractal dimensions Dim and can be derived from and , respectively. Moreover, some inequalities about the dimension of product sets and product measures are obtained.
BibTeX
@article{chang2007new,
  author = {Der‐Chen Chang and Yong Xu},
  title = {{New Inequalities on Fractal Analysis and Their Applications}},
  journal = {Journal of Inequalities and Applications},
  year = {2007},
  doi = {10.1155/2007/26249},
}
Figure