🔢 mathematics

On the packing dimension of weighted singular matrices on fractals

This paper establishes the first known upper bounds for the packing dimension of weighted singular and $\omega$ -singular matrices, including their intersections with fractal subsets, by employing homogeneous dynamics methods to analyze points on the space of unimodular lattices whose orbits escape on average under diagonal flows.

Original authors: Gaurav Aggarwal, Anish Ghosh

Published 2026-05-05

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Gaurav Aggarwal, Anish Ghosh

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Finding the "Impossible" in a "Messy" World

Imagine you are trying to hit a moving target with a dart. In the world of mathematics, this target is a specific number or matrix, and your "dart" is an approximation using simple fractions (integers).

Usually, there is a limit to how close you can get. However, some special numbers are "singular." These are the "super-targets" that are so weirdly positioned that you can hit them with infinite precision, getting closer and closer forever, no matter how small your error margin is.

This paper is about two things:

Weighted Singular Matrices: These are the "super-targets" where the rules of the game are slightly different. Instead of treating all parts of the number equally, some parts are weighted more heavily (like a dartboard where the center is worth more points, or harder to hit).
Fractals: These are the "messy" places where we are looking for our targets. A fractal is a shape that looks the same no matter how much you zoom in (like a coastline or a snowflake). The authors are asking: "If we restrict our search for these 'super-targets' to only exist inside these messy, self-repeating shapes, how many of them are there?"

The Core Question: How "Big" is the Set of Impossible Targets?

In math, we measure the "size" of a set in two main ways:

Volume (Lebesgue Measure): Does it take up space? (Like water in a bucket).
Dimension (Hausdorff/Packing Dimension): How complex is the shape? (Is it a line? A surface? Or something in between, like a crumpled piece of paper?)

The authors know that these "singular matrices" are incredibly rare. If you pick a random matrix, the chance of it being singular is zero. But they are not empty; there are infinitely many of them. The big question is: How "thick" or "dense" are they?

The paper provides a new upper bound for the "packing dimension." Think of packing dimension as a measure of how tightly you can pack these singular matrices together. If the dimension is low, they are very sparse (like dust). If it's high, they are more like a cloud.

The Main Discovery: A New Ceiling

The authors prove that if you look for these singular matrices inside a fractal (a self-repeating shape), they are even sparser than we thought.

They provide a formula that acts like a ceiling. No matter how you arrange the fractal or the weights, the "density" of these singular matrices cannot exceed a certain limit.

The Formula: The density is roughly the size of the fractal minus a penalty factor.
The Penalty: This penalty depends on how "singular" the matrix is and how the weights are distributed.

The Analogy: Imagine a giant, intricate lace doily (the fractal). You are looking for tiny, invisible gold dust (the singular matrices) hidden in the holes of the lace.

Previous research told us the gold dust was rare.
This paper says: "Actually, if you look closely at the specific pattern of the lace, the gold dust is even rarer than we guessed. Here is the exact mathematical limit on how much gold dust can possibly fit in the holes."

How They Did It: The "Lattice" and the "Flow"

To find this limit, the authors didn't just look at numbers. They used a clever trick called Dani's Correspondence.

The Lattice: Imagine a grid of points stretching out into infinity (a lattice). A "singular matrix" corresponds to a grid that gets stretched and squashed in a very specific way.
The Flow: Imagine a river flowing through this grid. The authors study the path of a boat (the matrix) moving along this river.
- If the boat stays in the "safe zone" (the middle of the river), it's a normal matrix.
- If the boat gets pushed toward the "edges" or "cliffs" (diverges), it corresponds to a singular matrix.

The authors used tools from homogeneous dynamics (the study of how things move on these grids) to prove that if the boat stays near the cliffs for a long time, it must be following a very specific, restricted path.

The "Fractal" Twist

The real breakthrough here is applying this to fractals.

The Challenge: Fractals are jagged and irregular. Standard math tools often smooth things out, which destroys the fractal's unique shape.
The Solution: The authors built special "height functions" (like measuring sticks) that are tailored specifically to the jagged nature of fractals. They proved that even with these weird shapes, the "escape" of the boat toward the cliffs is still limited.

They also handled weights. In the standard game, every direction is equal. In their "weighted" game, some directions are stretched more than others. They proved that even with this uneven stretching, the "escape" is still limited, and they calculated exactly how much.

Summary of Results

New Upper Bounds: They found the first known "ceiling" for how dense these singular matrices can be when they are forced to live inside fractals.
Generalization: Their method works for:
- Weighted matrices (where some parts matter more).
- Unweighted matrices (the standard case, where they improved existing knowledge).
- General Fractals: They expanded the types of fractals where this is known to work (specifically, products of self-similar shapes like the Cantor set).
The "Packing" Dimension: They focused on "packing dimension," which is a slightly more generous measure of size than the standard "Hausdorff dimension." Since they proved a limit on the packing dimension, they automatically proved a limit on the Hausdorff dimension as well.

In a Nutshell

This paper is a mathematical detective story. The detectives (the authors) went to a very complex, jagged crime scene (a fractal) to find a rare type of criminal (a singular matrix). They used a special tracking system (homogeneous dynamics) to prove that these criminals are even more elusive than previously thought. They provided a precise mathematical "fence" that says, "No matter how you try, you cannot pack more than this amount of these criminals into this specific type of fractal."

This is a theoretical result in pure mathematics (Number Theory and Dynamical Systems), and the paper does not claim any direct applications to engineering, medicine, or finance. It is purely about understanding the fundamental geometry of numbers.

Technical Summary: On the Packing Dimension of Weighted Singular Matrices on Fractals

Problem Statement
The paper addresses the problem of estimating the packing dimension of sets of weighted singular matrices, specifically when these sets are intersected with fractal subsets of the space of matrices $M_{m \times n}(\mathbb{R})$ .

In the context of Diophantine approximation, a matrix $\theta$ is $(a, b)$ -singular if Dirichlet's theorem can be infinitely improved for that matrix with respect to given weight vectors $a$ and $b$ . While the Hausdorff dimension of singular vectors has been extensively studied (notably by Cheung, Chevallier, Kadyrov, Kleinbock, Lindenstrauss, and Margulis), the packing dimension of these sets, particularly in weighted settings and restricted to fractals, remains less understood. The authors aim to provide the first known upper bounds for the packing dimension of:

Weighted singular matrices ($Sing(a, b)$).
Weighted $\omega$ -singular matrices ( $Sing(a, b, \omega)$ ).
The intersection of these sets with fractals defined by Iterated Function Systems (IFS).

Methodology
The authors employ methods from homogeneous dynamics, specifically utilizing Dani's correspondence, which relates singular matrices to divergent trajectories in the space of unimodular lattices $X_{m+n} = SL_{m+n}(\mathbb{R}) / SL_{m+n}(\mathbb{Z})$ .

The core technical framework involves three main components:

Dani's Correspondence and Divergence: The authors translate the Diophantine properties of matrices into dynamical properties of orbits under diagonal flows $g_t$ . A matrix is singular if its corresponding orbit diverges. They also consider "singular on average" trajectories and trajectories that escape to infinity at a specific rate.
Height Functions and the Contraction Hypothesis: To bound the dimension of the set of divergent trajectories intersected with a fractal $K$ , the authors construct a family of height functions $\{f_\tau\}$ on the space of lattices. These functions are tailored to satisfy a "contraction hypothesis" with respect to the fractal measure $\mu$ .
- The construction relies on establishing an expansion property for representations of the group $SL_{m+n}(\mathbb{R})$ acting on exterior powers of the vector space.
- They utilize integral inequalities (inspired by work on the Oppenheim conjecture) to prove that the fractal measure satisfies the contraction hypothesis. This involves proving that for a specific sequence of exponents $\eta_l$ , the integral of the height function under the flow decays exponentially.
Iterative Dimension Bounds: The authors apply the contraction hypothesis iteratively. By analyzing the measure of sets where the height function remains large (indicating divergence), they derive upper bounds on the packing dimension. This process requires handling unequal weights ( $a_i \neq a_j$ and $b_i \neq b_j$ ), which introduces technical difficulties regarding uneven expansion under conjugation. The authors address this by proving the contraction hypothesis not only for the original measure but also for measures arising from small perturbations.

Key Contributions and Results

General Upper Bounds: The paper establishes upper bounds for the packing dimension of $Sing(a, b) \cap K$ and $Sing(a, b, \omega) \cap K$ , where $K$ is a product of limit sets of IFSs satisfying the open set condition.
- The bound is given by:
  $\dim_P(Sing(a, b) \cap K) \le \dim_P(K) - \frac{1}{a_1 + b_1} \left( \min_{1 \le l \le d-1} \eta_l w_l \right)$
  where $w_l$ are critical exponents derived from the weights, and $\eta_l$ are constants depending on the fractal geometry (specifically related to critical exponents of the measure).
Explicit Constants: The authors provide explicit formulas for the constants $\eta_l$ in three specific cases:
1. The full matrix space $K = M_{m \times n}([0, 1])$ .
2. The case where $n=1$ (row vectors).
3. The case where $m=1$ (column vectors).
  In these cases, the constants are expressed in terms of the dimensions of the fractal components.
Extension to Fractals: The results generalize previous findings on singular vectors on fractals. While prior work (e.g., Khalil) provided upper bounds for Hausdorff dimension on self-similar fractals in $\mathbb{R}^n$ satisfying the open set condition, this paper extends the scope to:
- Weighted singular matrices.
- Packing dimension (which is generally larger than or equal to Hausdorff dimension).
- Products of self-similar fractals in $\mathbb{R}$ with equal contraction ratios.
- New examples, such as the product of a middle-third Cantor set with a middle-fifth Cantor set.
Critical Exponents: The paper introduces and analyzes "critical exponents" $\zeta_l(\mu)$ associated with the fractal measure. These exponents encode information about the measure's behavior near certain hypersurfaces and are crucial for determining the magnitude of the dimension reduction.

Significance and Claims
The authors claim to provide the first known upper bounds for the packing dimension of weighted singular and weighted $\omega$ -singular matrices, both in the general setting and when intersected with fractals.

Novelty in Unweighted Settings: Even in the unweighted case, the results are new for matrices, expanding the known scope of Problem 6 posed by Bugeaud, Cheung, and Chevalier regarding singular pairs in the middle-third Cantor set.
Generalization: The results enlarge the class of fractals for which dimension bounds are known, moving beyond simple self-similar sets in $\mathbb{R}^n$ to products of fractals in $\mathbb{R}$ .
Methodological Contribution: The paper highlights two key contributions to the methodology:
1. The construction of height functions satisfying the contraction hypothesis for fractal measures, which involves proving expansion properties for representations.
2. The handling of unequal weights in the iterative contraction process, which required careful analysis of measures under perturbations to manage the technical difficulties of uneven expansion.

The paper concludes that these bounds are derived from a more general theorem concerning the packing dimension of points on the space of unimodular lattices whose orbits under diagonal flows "escape on average" or diverge. The results are presented as upper bounds, acknowledging that the packing dimension is an upper bound for the Hausdorff dimension, thereby providing new constraints on the size of these Diophantine sets.