Hausdorff Dimension — Mathematics Knowledge System

Overview

The Hausdorff dimension is a measure of the 'size' or 'roughness' of a set in a metric space that generalises the familiar notion of integer dimension. While smooth manifolds have integer dimensions (curves: 1, surfaces: 2, …), fractals such as the Cantor set and the Koch snowflake have non-integer Hausdorff dimension. Defined via the Hausdorff measure — a family of outer measures Hˢ — the Hausdorff dimension dim_H(E) is the critical exponent at which Hˢ(E) transitions from ∞ to 0.

Intuition

Cover a set E with balls of radius at most δ and ask: how does the 'cost' (measured as the sum of rˢ for each ball of radius r) scale as δ → 0? For small s, even coarse covers suffice; for large s, the cost explodes. The Hausdorff dimension is the threshold s* where the cost switches from ∞ to 0. For a smooth curve in ℝ², s* = 1 (you need 1/δ ≈ δ^{-1} balls, cost ~ δ^{-1} · δ^s = δ^{s-1} → 0 iff s > 1). For the Cantor set, s* = log 2 / log 3 ≈ 0.63.

Formal Definition

Definition

For s ≥ 0 and δ > 0, define the s-dimensional Hausdorff δ-premeasure of E ⊆ ℝⁿ as the infimum over all δ-covers.

\mathcal{H}^s_\delta(E) = \inf\left\{\sum_i (\mathrm{diam}\, U_i)^s : E \subseteq \bigcup_i U_i,\; \mathrm{diam}\, U_i \leq \delta\right\}

s-dimensional Hausdorff δ-premeasure

\mathcal{H}^s(E) = \lim_{\delta \to 0} \mathcal{H}^s_\delta(E)

s-dimensional Hausdorff measure

\dim_H(E) = \inf\{s \geq 0 : \mathcal{H}^s(E) = 0\} = \sup\{s \geq 0 : \mathcal{H}^s(E) = \infty\}

Hausdorff dimension

\mathcal{H}^s(E) = \begin{cases} \infty & s < \dim_H(E) \\ 0 & s > \dim_H(E) \end{cases}

Jump at the critical dimension

Notation

Notation	Meaning
$\mathcal{H}^s(E)$	s-dimensional Hausdorff measure of E
$\dim_H(E)$	Hausdorff dimension of E
$\mathrm{diam}(U)$	Diameter of a set U
$\dim_B(E)$	Box-counting (Minkowski) dimension — often easier to compute but may differ from dim_H

Properties

Monotonicity

E \subseteq F \implies \dim_H(E) \leq \dim_H(F)

Countable stability

\dim_H\!\left(\bigcup_{n=1}^\infty E_n\right) = \sup_n \dim_H(E_n)

Cantor set dimension

\dim_H(C) = \frac{\ln 2}{\ln 3}

Self-similar sets

\text{If } E = \bigcup_{i=1}^N f_i(E) \text{ with } |f_i(x)-f_i(y)| = r_i|x-y|,\text{ then } \sum_{i=1}^N r_i^s = 1 \text{ determines } \dim_H(E)

Worked Examples

1
At stage n, the Cantor set is covered by 2ⁿ intervals of length (1/3)ⁿ.
$\mathcal{H}^s_{(1/3)^n}(C) \leq 2^n \cdot \left(\frac{1}{3^n}\right)^s = \left(\frac{2}{3^s}\right)^n$
2
For s = log 2/log 3, we get 2/3^s = 2/2 = 1, so the sum stays bounded.
$s = \frac{\ln 2}{\ln 3} \implies 2^n \cdot 3^{-ns} = 1$
3
One can show Hˢ(C) > 0 for this s by the mass distribution principle.
$\mathcal{H}^{\ln 2/\ln 3}(C) \in (0, \infty)$

✓ Answer

dim_H(C) = log 2 / log 3 ≈ 0.631.

Practice Problems

Mediumfree response

Show that dim_H({x}) = 0 for any single point x in ℝⁿ, and dim_H(ℝⁿ) = n.

Hardfree response

Use the Moran equation to find dim_H of the Koch snowflake curve.

Common Mistakes

Common Mistake

Assuming Hausdorff dimension always equals the box-counting dimension

For many 'nice' sets (self-similar fractals satisfying the open set condition) dim_H = dim_B. But they can differ: a countable dense set has dim_H = 0 but dim_B = the ambient dimension.

Common Mistake

Thinking H^{dim_H(E)}(E) is always finite and positive

At the critical dimension s = dim_H(E), the Hausdorff measure Hˢ(E) can be 0, finite positive, or ∞. The dimension is simply the threshold between ∞ and 0.

Quiz

If Hˢ(E) = ∞ and H^t(E) = 0 where t > s, the Hausdorff dimension of E is:

stA value in the interval (s, t)Not determined by this information alone

The Hausdorff dimension of the middle-thirds Cantor set is:

01/2log 2 / log 3 ≈ 0.6311

Historical Background

Felix Hausdorff introduced the measure and dimension bearing his name in his 1918 paper 'Dimension und äußeres Maß'. He computed dim_H of the Cantor middle-thirds set. Besicovitch and Taylor clarified the theory in the 1950s. With Mandelbrot's popularisation of fractals in the 1970s–80s, Hausdorff dimension became a central concept in dynamical systems, complex analysis, and mathematical physics.

1918
Hausdorff defines s-dimensional measure and Hausdorff dimension
Felix Hausdorff
1928
Besicovitch studies irregular sets and refines the theory
Abram Besicovitch
1967
Mandelbrot uses fractal dimension to study coastline lengths
Benoit Mandelbrot
1982
Mandelbrot's 'The Fractal Geometry of Nature' popularises the concept
Benoit Mandelbrot

Summary

The Hausdorff dimension dim_H(E) is the critical exponent at which Hˢ(E) transitions from ∞ (s < dim_H) to 0 (s > dim_H).
It generalises integer dimension to fractals: smooth d-manifolds have dim_H = d; the Cantor set has dim_H = log2/log3.
Countable stability: dim_H(∪Eₙ) = sup dim_H(Eₙ).
Self-similar fractals satisfying the open set condition have dim_H determined by the Moran equation Σ rᵢˢ = 1.
The mass distribution principle gives lower bounds on dim_H by constructing appropriate measures.

References

BookFalconer, K. — Fractal Geometry: Mathematical Foundations and Applications, 3rd ed. (2014), Wiley
BookMattila, P. — Geometry of Sets and Measures in Euclidean Spaces (1995), Cambridge University Press
WebsiteWikipedia — Hausdorff dimension
WebsiteMathWorld — Hausdorff Dimension