Tube formulas for self-similar fractals

Tube formulas (by which we mean an explicit formula for the volume of an $\epsilon$-neighbourhood of a subset of a suitable metric space) have been used in many situations to study properties of the subset. For smooth submanifolds of Euclidean space, this includes Weyl’s celebrated results on spectral asymptotics, and the subsequent relation between curvature and spectrum. Additionally, a tube formula contains information about the dimension and measurability of rough sets. In convex geometry, the tube formula of a convex subset of Euclidean space allows for the definition of certain curvature measures. These measures describe the curvature of sets which are not too irregular to support derivatives. In this survey paper, we describe some recent advances in the development of tube formulas for self-similar fractals, and their applications and connections to the other topics mentioned here.

💡 Research Summary

The paper surveys recent progress on “tube formulas” for self‑similar fractals, extending the classical notion of the volume of an ε‑neighbourhood (or tube) around a set to highly irregular, non‑smooth objects. It begins by recalling the well‑known Weyl tube formula for smooth submanifolds of ℝⁿ, where the volume V(ε) admits an asymptotic expansion in powers of ε whose coefficients are curvature integrals. In convex geometry the same idea yields curvature measures for convex bodies. Both contexts illustrate how the tube expansion encodes geometric information (dimension, curvature) and analytic information (spectral asymptotics).

The authors then introduce the machinery needed to treat fractal sets: the theory of complex dimensions, the fractal zeta function ζ_F(s)=∑_j r_j^s (with r_j the similarity ratios of the iterated function system), and Mellin transform techniques. The poles of ζ_F(s) form the “complex dimension spectrum” 𝔇, a discrete set of complex numbers whose real parts cluster around the Hausdorff (or Minkowski) dimension D of the fractal, while the imaginary parts reflect the inherent log‑periodicity of the self‑similar construction.

A central result is the explicit tube formula
 V_F(ε)=∑_{ω∈𝔇} c_ω ε^{N−ω}+R(ε),
where N is the ambient Euclidean dimension, c_ω are residues of a meromorphic continuation of ζ_F(s) multiplied by a simple factor, and R(ε) is an error term of order o(ε^{N−σ}) with σ= sup{Re ω : ω∈𝔇, ω≠D}. When the fractal is Minkowski measurable (i.e., its ε‑neighbourhood volume behaves like a single power of ε up to a constant), the only contributing complex dimension is the real one D, and the oscillatory terms disappear. In contrast, non‑measurable fractals exhibit infinitely many non‑real complex dimensions, producing a series of oscillatory “fractal waves” in the volume expansion.

The paper works out several canonical examples in detail. For the Koch curve (similarity ratio 1/3, four copies) the complex dimensions are ω_k=log 4/log 3+2π i k /log 3 (k∈ℤ). The tube expansion therefore contains an infinite sum of terms ε^{2−ω_k} whose coefficients can be written explicitly. Numerical experiments confirm that as ε→0 the oscillations become increasingly pronounced. Similar calculations are presented for the Sierpiński gasket (ratio 1/2, three copies) and the Sierpiński carpet (ratio 1/3, eight copies), each illustrating how the spacing of the imaginary parts of the complex dimensions is dictated by the logarithm of the scaling factor. A non‑self‑similar but “approximately self‑similar” set is also examined to illustrate the limits of the theory.

Beyond pure geometry, the authors explore two major applications. First, they connect the tube formula to spectral asymptotics of the Laplacian on fractal domains. The counting function N(λ) for eigenvalues λ satisfies a “fractal Weyl law” of the form
 N(λ)=C λ^{N/2}+∑_{ω∈𝔇{D}} c_ω λ^{(N−ω)/2}+o(λ^{(N−σ)/2}),
showing that the same complex dimensions that govern the volume of ε‑neighbourhoods also generate fine oscillations in the eigenvalue distribution. This provides a precise quantitative link between geometry and quantum (or wave) phenomena on fractal drums.

Second, the paper proposes a new notion of curvature for fractals. In the classical setting curvature measures arise as coefficients of the tube expansion; analogously, the residues c_ω associated with the real dimension D can be interpreted as a “fractal mean curvature,” while residues attached to non‑real dimensions give rise to “fractal oscillatory curvature” terms that capture the roughness of the set. These quantities extend the curvature measure theory of convex bodies to a much broader class of sets that lack classical differentiable structure.

The survey concludes with a discussion of open problems. Computing the complex dimension spectrum for general (especially random or non‑self‑similar) fractals remains challenging; numerical methods are still being refined. Extending the tube formula to multifractal measures, to higher‑order curvature invariants, and to dynamical systems where the underlying set evolves in time are promising directions. Finally, the authors highlight potential interdisciplinary applications—in material science, image analysis, and signal processing—where an explicit understanding of ε‑neighbourhood volumes could improve texture classification, surface roughness quantification, and the design of fractal antennas. The paper thus positions tube formulas as a unifying bridge linking geometric measure theory, spectral analysis, and applied mathematics in the study of self‑similar fractals.