Complexity function and entropy of induced maps on hyperspaces of continua

Here is an explanation of the paper "Complexity Function and Entropy of Induced Maps on Hyperspaces of Continua" using simple language, analogies, and metaphors.

The Big Picture: From One Actor to a Whole Crowd

Imagine you are watching a single actor on a stage. They move around, maybe they dance, maybe they sit still. In mathematics, we call this a dynamical system. We can measure how "chaotic" or "complex" their movement is. If they just walk in a perfect circle, it's simple. If they run around randomly, it's complex.

Now, imagine we don't just watch the actor. Instead, we watch every possible group of actors that could be on stage at the same time.

We watch the single actor.
We watch a pair of actors holding hands.
We watch a whole crowd moving together.
We watch a line of actors forming a shape.

This collection of all possible groups is called the Hyperspace. The paper asks a fascinating question: If the single actor is simple, is the whole crowd also simple? Or does the crowd become chaotic even when the individual is calm?

The authors, Jelena Katić, Darko Milinković, and Milan Perić, explore this relationship. They use two different "thermometers" to measure complexity:

Topological Entropy: Measures how fast things grow exponentially (like a virus spreading).
Polynomial Entropy: A finer tool for measuring growth that is slower, like a plant growing (linear or quadratic).

Key Concept 1: The "Wandering" Drifter

The paper introduces a specific type of movement called a wandering point.

Analogy: Imagine a person walking through a park. If they walk in a circle, they keep coming back to the same spot. That's "non-wandering." But if they walk in a straight line, never to return to a previous spot, they are "wandering."

The Big Discovery (Theorem 1 & 2):
The authors found a surprising rule for 3D spaces (like a sphere or a donut shape).

The Rule: If you have a 3D space and even one person is wandering (drifting away), then the complexity of watching all possible groups of people becomes infinite.
The Metaphor: Imagine a calm lake. If you drop one stone that creates ripples that never settle (a wandering point), the entire surface of the lake eventually becomes a chaotic mess of infinite waves. Even if the rest of the water is calm, that one drifter makes the whole system infinitely complex to predict.

Key Concept 2: The Star-Shaped Room

The paper also looks at 1D spaces, specifically a Star (a shape like a starfish with a center and several arms).

The Discovery (Theorem 3 & Corollary 4):
They looked at what happens if people are wandering on the arms of the star.

The Rule: The complexity of the crowd depends on how many arms have a drifter.
The Metaphor: Imagine a star-shaped room with $k$ hallways. If people are wandering down 3 of those hallways, the complexity of the "group dynamics" is exactly 3.
If no one is wandering, the complexity is 0 (it's perfectly predictable).
If people are wandering on 5 arms, the complexity is 5.

It's like a musical band: if you have 3 instruments playing a wandering melody, the complexity of the song is determined by those 3 instruments.

Key Concept 3: The "Group" vs. The "Individual"

One of the most interesting findings is about the difference between watching one person and watching a group.

The Scenario: Imagine a system where the individual actor is very simple (zero chaos).
The Result: The paper shows that you can create a system where:
- The complexity of watching 1 person is low.
- The complexity of watching groups of 2 people is higher.
- The complexity of watching groups of 3 people is even higher.
- And so on, forever.

The Metaphor: Think of a classroom.

If you watch one student sitting quietly, it's boring (low complexity).
If you watch two students talking, it's a little more complex.
If you watch three students forming a triangle, it's more complex again.
The paper proves you can design a classroom where the "group complexity" keeps rising the more students you watch, even if the individual students are just sitting there.

Why Does This Matter?

In the real world, we often study systems by looking at individuals (like a single stock price or a single neuron). But in reality, things happen in groups (a market crash, a neural network firing).

This paper gives mathematicians a new set of tools (Polynomial Entropy) to measure these group behaviors. It tells us:

Don't be fooled by simplicity: A system can look calm on the surface (one person walking), but if there's a "drifter" involved, the group dynamics can be infinitely wild.
Count the branches: In star-shaped systems, the number of "wandering" paths directly tells you the complexity of the group.
Groups have their own rules: The complexity of a group isn't just the sum of its parts; it can grow in a specific, predictable way based on how many parts are interacting.

Summary in One Sentence

This paper proves that in certain mathematical worlds, the chaos of a "crowd" is determined by the number of "drifters" in the system, and that even a single wandering point can turn a simple 3D space into a system of infinite complexity when you start watching the groups.

Here is a detailed technical summary of the paper "Complexity Function and Entropy of Induced Maps on Hyperspaces of Continua" by Jelena Katić, Darko Milinković, and Milan Perić.

1. Problem Statement and Context

The paper investigates the relationship between the dynamics of a continuous map $f$ on a compact metric space $X$ and the dynamics of the induced map on the hyperspace of $X$ .

Induced Map: For a compact metric space $X$ , the hyperspace $2^X $consists of all non-empty closed subsets of$ X $. The map$ f: X \to X $induces a map$ 2^f: 2^X \to 2^X $defined by$ 2^f(A) = {f(x) \mid x \in A}$.
Hyperspace of Continua: If $X$ is connected, the paper focuses on $C(X)$ , the subspace of $2^X $consisting of all non-empty closed and connected subsets (subcontinua), and the restricted induced map$ C(f) = 2^f|_{C(X)}$.
The Core Question: How do dynamical invariants, specifically topological entropy ( $h$ $h$ ) and polynomial entropy ( $h_{pol}$ $h_{p o l}$ ), behave when passing from the base system $(X, f)$ $(X, f)$ to the induced system $(C(X), C(f))$ $(C (X), C (f))$ ?
- It is known that if $h(f) > 0$ , then $h(2^f) = \infty$ .
- However, the behavior of $h(C(f))$ is more nuanced. For example, if $h(f)=0$ , $h(C(f))$ can be 0, finite, or infinite depending on the space and the map.
- The authors aim to characterize $h_{pol}(C(f))$ for zero-entropy systems (where topological entropy is insufficient to distinguish complexity) and establish conditions under which $h(C(f))$ becomes infinite.

2. Methodology

The authors employ a combination of topological dynamics, combinatorial complexity, and symbolic dynamics.

Complexity Functions and Subshifts:
The central technical tool is the complexity function $p_\Sigma(m)$ $p_{Σ} (m)$ , which counts the number of distinct words of length $m$ $m$ in a subshift $\Sigma$ $Σ$ .
- They utilize a generalized result (Theorem 9) stating that for a $\sigma$ -invariant set $\Sigma$ (not necessarily closed), the topological and polynomial entropies of the shift map $\sigma$ can be computed directly via the complexity function:
  $h(\sigma; \Sigma) = \lim_{m \to \infty} \frac{\log p_\Sigma(m)}{m}, \quad h_{pol}(\sigma; \Sigma) = \lim_{m \to \infty} \frac{\log p_\Sigma(m)}{\log m}$
Symbolic Encoding of Hyperspaces:
To compute the entropy of $C(f)$ $C (f)$ , the authors construct specific subsets of the hyperspace (e.g., sets of stars or finite sets of points) and map them to symbolic spaces (subshifts).
- They define a map $\phi$ from a subset of the hyperspace to a sequence space $\{0,1\}^k$ .
- Crucially, they address the fact that this map $\phi$ is often not continuous. Therefore, standard factor arguments (which require continuous surjections) cannot be directly applied.
- Resolution: They prove that despite the discontinuity, the entropy of the hyperspace system is bounded below by the entropy of the symbolic system by using separated sets and the Hausdorff metric. They show that if sequences in the symbolic space are separated, the corresponding sets in the hyperspace are also separated by a fixed distance $\delta$ , provided the underlying points are wandering.
Lipschitz Factors:
They utilize Proposition 7, which states that if there is a Lipschitz continuous surjection between invariant sets, the entropy of the domain is greater than or equal to the entropy of the codomain. This is used to relate the entropy of $C(f)$ to simpler systems like symmetric products $F_n(X)$ .

3. Key Contributions and Results

A. Infinite Topological Entropy on Manifolds (Theorem 1 & Corollary 2)

Result: If $M$ is a compact, connected topological manifold of dimension $\ge 2$ and $f: M \to M$ is a homeomorphism with at least one wandering point, then the topological entropy of the induced map on the hyperspace of continua is infinite: $h(C(f)) = \infty$ .
Significance: This is a surprising result because the existence of a wandering point in the base system does not necessarily imply infinite entropy for the base system itself (e.g., Morse-Smale systems). However, in the hyperspace of a manifold with dimension $\ge 2$ , the "collective" dynamics of continua become infinitely complex.
Corollary: This provides an alternative proof that for Morse-Smale diffeomorphisms on manifolds of dimension $\ge 2$ , $h(C(f)) = \infty$ .

B. Polynomial Entropy on Stars and Graphs (Theorem 3, Corollary 4, & Corollary 13)

For systems with zero topological entropy, the authors use polynomial entropy ( $h_{pol}$ ) as a finer invariant.

Theorem 3 (Stars): Let $X = S_k$ be a star graph with $k$ edges and $f$ a homeomorphism. If every edge contains a wandering point, then $h_{pol}(C(f)) = k$ .
Corollary 4/13 (General Graphs): For a homeomorphism $f$ $f$ on a graph $X$ $X$ , $h_{pol}(C(f))$ $h_{p o l} (C (f))$ is determined by the maximal number of edges meeting at a branch point that contain wandering points ( $k$ $k$ ):
- If $k \ge 2$ , $h_{pol}(C(f)) = k$ .
- If $k \le 1$ and there exists a closed path (cycle) where the dynamics are not conjugate to a circle rotation, $h_{pol}(C(f)) = 2$ .
- Otherwise, $h_{pol}(C(f)) = 0$ .
Significance: This provides a complete classification of polynomial entropy for induced maps on graphs, showing that the complexity scales linearly with the number of "active" (wandering) branches.

C. Strict Hierarchy of Entropy for Induced Maps (Theorem 5)

The authors answer an open question from previous literature (Arbieto and Bohorquez) regarding the hierarchy of entropies for higher-order induced maps $C^n(f)$ (maps on sets with at most $n$ connected components).

Result: There exist dynamical systems such that the topological entropy strictly increases with the order of the induced map:
$h(C(f)) < h(C^2(f)) < \dots < h(C^n(f)) < \dots < h(2^f) = \infty$
Polynomial Version: Similarly, for polynomial entropy, there exist systems where:
$h_{pol}(C(f)) < h_{pol}(C^2(f)) < \dots < h_{pol}(C^n(f)) < \dots < h_{pol}(2^f)$
Method: They construct examples using interval maps. For topological entropy, they use maps with positive entropy. For polynomial entropy, they use homeomorphisms of the interval with wandering points, showing that $h_{pol}(C^n(f)) = 2n$ .

4. Significance and Impact

Bridging Zero and Infinite Entropy: The paper clarifies the transition from zero to infinite entropy in induced systems. It shows that while zero topological entropy in the base system can lead to zero or finite polynomial entropy in the hyperspace, the mere presence of wandering points in higher-dimensional manifolds forces the hyperspace entropy to infinity.
Refining Complexity Measures: By utilizing polynomial entropy, the authors distinguish between systems that are indistinguishable by topological entropy (e.g., rotations vs. homeomorphisms with wandering points). The linear relationship found between the number of wandering branches and polynomial entropy ( $h_{pol} = k$ ) offers a precise quantitative measure of complexity.
Methodological Innovation: The technique of using complexity functions of non-closed invariant sets to bound the entropy of discontinuous symbolic encodings of hyperspaces is a robust tool that can likely be applied to other problems in continuum dynamics.
Resolution of Open Problems: The paper definitively answers the question of whether a strict hierarchy of entropies exists for $C^n(f)$ , confirming that the complexity of collective dynamics grows strictly with the number of allowed connected components.

In summary, this work significantly advances the understanding of how local dynamical properties (like wandering points) propagate to global collective behaviors in hyperspaces, providing exact formulas for polynomial entropy on graphs and establishing the inevitability of infinite topological entropy in higher-dimensional hyperspaces with wandering points.