🔢 mathematics

Analytic patch trees: branch interface inheritance and fractal dimension fields

This paper generalizes analytic fractal curve trees to higher-dimensional surface patch trees by replacing branch points with interface manifolds that transmit analytical states, thereby establishing conditions for integrability, conformality, and self-similarity while introducing a smooth dimension field through the natural foliation of the trees into curve structures.

Original authors: Henk Mulder

Published 2026-06-05✓ Author reviewed ⓘ

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Henk Mulder

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 by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are building a fractal tree, but instead of drawing it with a pencil, you are "growing" it using a set of mathematical rules.

In a previous paper, the author showed how to grow lines (curves) that branch out infinitely. This new paper takes that idea and upgrades it to grow surfaces (patches), like leaves or sheets of paper, instead of just lines.

Here is the core idea broken down into simple concepts and analogies:

1. From "Branch Points" to "Branch Interfaces"

In a standard line-tree, branches split at a single point (like a Y-shape).
In this new "patch tree," branches split along a curve (a line).

The Analogy: Imagine a river delta. A single river doesn't just split into two tiny streams at one dot; it spreads out and splits into many channels along a wide front.
What it means: When a "parent" patch splits into "child" patches, it doesn't just pass a single coordinate. It passes an entire interface (a whole curve) carrying all the data (position, direction, speed) to the children. This interface is the most important part of the structure.

2. The "Seam" That Connects Everything

The paper introduces a concept called the Interface Evolution Operator. Think of this as a "seam" or a "hand-off" rule.

The Analogy: Imagine a relay race. In a normal race, a runner hands a baton to the next person. In this math world, the runner hands over a living, moving map of the track.
How it works: The "parent" patch grows to a certain depth. The edge where it ends is the "tip interface." This edge is handed to the "child" patches. The child patches then use that edge as their starting line to grow further.
The Twist: Sometimes, the "hand-off" is perfect and straight (the child looks exactly like the parent). Sometimes, the hand-off twists or stretches the edge (the child looks deformed). The paper studies how these edges change from generation to generation.

3. The "Smooth Dimension" Field

One of the most surprising findings is about dimension (how "rough" or "complex" the shape is).

The Analogy: Imagine a loaf of bread. If you slice it, each slice is a flat piece of bread. But in this math model, every single slice of the tree is actually a tiny, complex fractal line.
The Discovery: The author found that you can slice the entire 3D-looking tree into many 1D lines. Each line has its own "complexity score" (called Hausdorff dimension).
The Result: Instead of the whole tree having one single complexity score, the tree has a smooth field of complexity. Some parts of the tree are "rougher" than others, and this roughness changes smoothly across the surface, like a temperature map on a weather chart.

4. The "Perfect" Trees (Conformal Trees)

The paper identifies a special, "perfect" type of tree called a Conformal Patch Tree.

The Analogy: Think of a rubber sheet. If you stretch a rubber sheet evenly in all directions, circles stay circles, and angles stay 90 degrees. This is "conformal."
The Discovery: If the mathematical rules (generator fields) follow specific conditions (like the Cauchy-Riemann equations), the tree grows in a way that preserves angles perfectly.
Self-Similarity: Usually, to make a fractal look the same at every zoom level, you have to force it to shrink and rotate manually. Here, the author shows that if you use these "perfect" rules, the tree naturally becomes self-similar. The pattern repeats itself automatically because of how the "seams" (interfaces) interact with the growth rules.

5. Growing Beyond 2D

Finally, the paper explains that this isn't just for flat surfaces (2D).

The Analogy: Imagine a 3D block of cheese. If you cut it, you get 2D slices. If you have a 4D object, you cut it to get 3D "slices."
The General Rule: You can have "patches" of any size. If you have a 3D patch, the "seams" where it splits are 2D surfaces. If you have a 10D patch, the seams are 9D.
The Regimes: The paper notes that depending on how big the "patch" is compared to how many "branches" it has, the math behaves differently.
- If the patch is small and branches are many, it's mostly about the branching pattern (geometry).
- If the patch is huge and branches are few, it's mostly about transporting data through the patch (operational).

Summary

This paper replaces the idea of "branching at a point" with "branching along a curve." It shows that these surfaces are made of layers of fractal lines, creating a smooth map of complexity. It proves that if you follow specific "perfect" mathematical rules, these trees naturally grow in a self-repeating, angle-preserving way, and this whole system can be scaled up to any number of dimensions.

Technical Summary: Analytic Patch Trees

Problem Statement
This work addresses the extension of analytic fractal curve trees, previously defined in [1], from one-dimensional curves to two-dimensional surface patches and, subsequently, to arbitrary dimensions. In the curve framework, recursive geometry is generated by smooth field evolution rather than discrete geometric substitutions, with branching occurring at discrete points. The paper identifies a limitation in this model when applied to surfaces: discrete branch points are insufficient to transmit the full analytical state required for surface patches. The core problem is to define a recursive geometric structure where "branch points" are replaced by "branch interfaces" (curves or manifolds) that transmit the complete field state from parent patches to children, while establishing the conditions for integrability, well-posedness, and self-similarity in this new context.

Methodology
The paper employs a framework of analytic generator fields and differential geometry to construct surface patch trees.

Generator System and Integrability: A surface patch is defined by a generator domain $D$ and a state vector $X$ evolving according to analytic generator fields $V_1$ and $V_2$ . The existence of a unique analytic solution is governed by the Frobenius integrability condition (Equation 2), which ensures path-independence of the integration in the generator domain.
Geometric Realisation: The abstract state $X$ is projected into an embedding space via a map $\Pi$ , creating a realised geometric patch $\gamma$ .
Recursive Construction: A patch tree is formed by recursively branching the tip interface of a parent patch into $N$ child patches. Crucially, the child patches inherit the parent's state at the tip interface via an analytic transport map $T^{(k)}$ .
Interface Evolution Operator: The paper introduces an operator $E$ that maps a base interface $\Gamma_0$ to a tip interface $\Gamma_1$ through the integration of the generator system. This operator, combined with transport maps, governs the recursive geometry.
Foliation Analysis: The patch tree is analyzed as a foliation of one-dimensional curve trees (slices at constant $s_1$ ). The Hausdorff dimension of each slice is calculated using Moran's equation, leading to the concept of a "smooth dimension field" across the tree.
Classification and Extension: The framework is classified based on the behavior of the interface evolution operator (preserving vs. modifying) and the properties of the generator fields (conformality via Cauchy–Riemann equations). The theory is generalized to $n$ -dimensions, where patches of dimension $d_p$ are connected by interfaces of dimension $d_p - 1$ .

Key Contributions

Foundational Structure: The paper establishes the analytic well-posedness of surface patch trees via the Frobenius theorem. It demonstrates that every such tree is foliated by a smooth one-parameter family of fractal curve trees, each possessing its own Hausdorff dimension, thereby creating a continuous "dimension field" across the patch tree.
Interface Theory: Branch interfaces are elevated to "first-class geometric objects." The introduction of the interface evolution operator ( $E$ ) provides a mechanism to classify patch trees based on how inherited interfaces transform under recursion (e.g., interface-preserving, interface-modifying, conformal, and self-similar).
Conformality and Intrinsic Self-Similarity: The paper defines conformal patch trees where generator fields satisfy the Cauchy–Riemann equations. It proves that a specific subclass—canonical self-similar conformal trees—emerges where self-similarity is not externally imposed but arises intrinsically from the interaction between the conformal structure and the interface evolution.
Higher-Dimensional Generalization: The framework is extended to arbitrary dimensions $n$ $n$ . The paper distinguishes between three analytical regimes based on the relative scaling of patch dimension ( $d_p$ $d_{p}$ ) and branching dimension ( $d_b$ $d_{b}$ ):
- Combined Regime ( $d_p \approx d_b$ ): Focus on recursive geometry and attractor structure.
- Geometric Regime ( $d_p \ll d_b$ ): Recursive organization dominates; focus on topology and branching dynamics.
- Operational Regime ( $d_p \gg d_b$ ): Patch geometry dominates; focus on transport, spectral theory, and diffusion.

Results

Theorem 2.1: Proves that the system of generator fields admits a unique analytic solution if and only if the Frobenius compatibility condition is met.
Theorem 2.3: Demonstrates that under contractive conditions, the patch tree is foliated by smooth fractal curve trees. It derives the formula for the slice Hausdorff dimension $d_{slice}(c) = \frac{\log N}{\log(1/r(c))}$ , showing that the dimension varies smoothly across the tree.
Theorem 4.1: Establishes that generator fields satisfying the Cauchy–Riemann equations yield locally conformal realisations, preserving angles and orthogonality of foliations.
Theorem 4.2: Proves that constant-gradient conformal fields generate a canonical self-similar tree where the interface evolution operator acts as a similarity transformation. The contraction factor is determined by the imaginary part of the complex gradient.
Classification: The paper categorizes patch trees into interface-preserving (translation), interface-modifying (deformation), conformal, and canonical self-similar conformal classes.

Significance
The paper claims that the transition from curves to surfaces fundamentally changes the nature of recursive geometry: branch points unfold into interface curves that mediate both geometric and non-geometric data. The interface evolution operator is identified as the central organizing principle for classifying these structures.

A significant theoretical implication noted is the potential for spectral analysis. Unlike classical fractal analysis which attempts to define operators directly on irregular limit sets, conformal patch trees are assembled from smooth recursive manifolds where standard differential geometry and PDE machinery are available. The fractal geometry emerges only as a recursive limit of these smooth structures. The authors suggest this framework offers a natural setting to approach difficult analytical questions on fractals indirectly through their smooth recursive approximations, though they note that recovering known spectral results or yielding new ones remains an open question.

The work concludes that the interplay between patch generators and inherited interfaces determines the recursive topology, with self-similarity emerging as a specific dependency of this coupling rather than an external constraint.