Complexity of measure-induced maps

Gabriel Lacerda · 2025

Read Complexity of measure-induced maps by Gabriel Lacerda (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-22944.txt.

Introduction

Gabriel Lacerda refines a classic claim about map chaos in a 2025 paper, as a solo author. The claim says a map and its lift to whole clouds of points share the same all-or-none chaos grade. Finer grades of growth can split them, though.

Lacerda shows the lift is strictly more complex when the base map has any true growth of a non-plain kind. When the still set of the map is finite, the lift growth stays at most in the tame, squared class. For shifts of vast spaces that bound is hit in full. Such fine splits let map scholars tell kin systems apart.

Catalog record

CREATORGabriel Lacerda
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-22944

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Complexity of measure-induced maps Gabriel Lacerda Abstract A classical result by Glasner and Weiss states that a map f f has zero topological entropy if and only if its measure-induced map f ∗ f_{*} , defined as the push-forward of a measure, does. In this work, we use generalized entropy to distinguish the complexity of these maps and prove that the measure-induced map is strictly more complex whenever the generalized entropy of the original map is nonzero and is not an abstract order of growth. Moreover, we show that the complexity of the measure-induced map is of at most polynomial type when the nonwandering set of f f is finite. For homeomorphisms of infinite spaces, this bound is attained, so measure-induced maps of Morse–Smale diffeomorphisms cannot be distinguished from one another. Mathematics Subject Classification (2020): 37B40, 37D15, 28A33. 1 Introduction In the field of dynamical systems, topological entropy is a classical tool used to evaluate the complexity of a continuous map f:M\to M on a compact metric space M . Denoted by h(f) , topological entropy measures the asymptotic exponential growth rate of the number of distinct orbits at small scales. However, for dynamical systems with sub-exponential growth of these distinct orbits, the topological entropy is zero. Positive topological entropy is often used to characterize systems that are chaotic or complex. However, there can be distinct maps with zero entropy where one is inherently more chaotic than the other. To illustrate, within the class of Morse–Smale dynamics—a classical example of zero-entropy maps—two distinct maps can have different phase diagrams [ 9 ] . This property implies that, depending on the phase space M , the behavior of their orbits can be completely different, and one can be strictly more complex than the other. To address this situation, Correa and Pujals introduced the concept of generalized entropy [ 3 ] , an approach that considers the growth rate of distinct orbits at all scales. By “growth rate,” we refer to the order of growth of the corresponding sequence, rather than a numerical asymptotic rate. Unlike topological entropy, generalized entropy is not a non-negative number; it is an equivalence class in a partially ordered set. More precisely, considering non-decreasing functions a:\mathbb{N}\to[1,\infty) , we define \mathbb{O} as the set of equivalence classes of such functions, where two functions are equivalent if they have the same growth rate. Hence, an element in \mathbb{O} is denoted by [a(n)] . The class corresponding to bounded growth is denoted by 0 , which represents the smallest possible growth. Furthermore, one can complete this partially ordered set to obtain the complete set of orders of growth \overline{\mathbb{O}} . This extended set contains abstract orders of growth, which are given by the suprema or infima of subsets of \mathbb{O} . Therefore, the generalized entropy of a map f:M\to M is denoted by o(f) and is an element of \overline{\mathbb{O}} . In this sense, generalized entropy provides a broader notion of complexity . This work is concerned with the generalized entropy of the measure-induced map f_{*}:\mathcal{P}(M)\to\mathcal{P}(M) , where \mathcal{P}(M) is the set of Borel probability measures on M and f_{*} is the push-forward operator. That is, for \mu\in\mathcal{P}(M) , f_{*}\mu=\mu\circ f^{-1} . A classical result regarding the complexity of the measure-induced map was established by Glasner and Weiss in [ 5 ] , where they showed that the topological entropy of f is zero if and only if the topological entropy of f_{*} is zero. Since generalized entropy serves as a tool to distinguish dynamical systems with zero topological entropy, a natural question arises: can it also capture the difference in complexity between the base map f and its measure-induced map f_{*} ? We answer this question in the affirmative by showing that the measure-induced map is significantly more complex. In fact, we provide a generalization of a well-known result by Bauer and Sigmund [ 1 ] , which states that if the topological entropy of f is positive, then the topological entropy of f_{*} is infinite. The strict growth of complexity, o(f_{*})>o(f) , does not occur for maps with the lowest ( o(f)=\textbf{0} ) and highest complexity ( o(f)=\sup\mathbb{E} , see § 2.2 ). More precisely, whenever o(f)\neq\textbf{0} , o(f_{*}) is an abstract order of growth; that is, it lies strictly in the completion \overline{\mathbb{O}}\setminus\mathbb{O} . Moreover, given a one-parameter family of orders of growth (e.g., \mathbb{A}=\{[a(n)^{t}]\in\mathbb{O}:t>0\} for a given [a(n)]\in\mathbb{O} ), we show that the complexity of f_{*} is at least the supremum of this family whenever o(f_{*}) is bounded below by a base element of \mathbb{A} . Our complete first result is the following: Theorem A . Let f:M\to M be a continuous map on a compact metric space. 1. o(f)=\emph{\textbf{0}} if, and only if, o(f_{*})=\emph{\textbf{0}} . Otherwise, o(f_{*})\in\overline{\mathbb{O}}\setminus\mathbb{O} . 2. For \mathbb{A}=\{[a(n)^{t}]\in\mathbb{O}:t>0\} , if o(f_{*})\geq[a(n)] , then o(f_{*})\geq\sup\mathbb{A} . 3. For \mathbb{B}=\{[b(n)^{t}]\in\mathbb{O}:t>0\} , if [b(n)]\geq o(f_{*}) , then \inf\mathbb{B}\geq o(f_{*}) . 4. In particular, if o(f)\geq[a(n)] , then o(f_{*})\geq\sup\mathbb{A} . Furthermore, if o(f)\in\mathbb{O} is such that o(f)>\emph{\textbf{0}} , then o(f_{*})>o(f) . The first statement of the preceding result is a generalized-entropy analogue of the main theorem in [ 5 ] , because 0 is the least element of the partially ordered set \overline{\mathbb{O}} . A consequence of the above theorem relates to the polynomial one-parameter family of orders of growth \mathbb{P}=\{[n^{t}]\in\mathbb{O}:t>0\} , illustrated in Figure 1 . Indeed, we can apply it in particular to homeomorphisms of the circle, where [ 3 ] showed that every circle homeomorphism that is not Lyapunov-stable has linear complexity [n] (e.g., Morse–Smale diffeomorphisms). Since Lyapunov-stable dynamics are tame, for homeomorphisms of the circle there are only two possibilities: either the complexity of the measure-induced map is 0 , or it is at least \sup\mathbb{P} , which we call super-polynomial . Moreover, we can use [ 3 , Theorem 3] to deduce the following result. Recall that the \alpha -limit set of a point x\in M is the set \alpha(x) of accumulation points of the backward orbit of x under f . Corollary 1 . If a homeomorphism f:M\to M has a point x\in M such that x\notin\alpha(x) , then o(f_{*})\geq\sup\mathbb{P} . In particular, if M is the circle S^{1} , then either o(f_{*})\geq\sup\mathbb{P} or o(f_{*})=\emph{\textbf{0}} . \inf\mathbb{P} [n^{t}] [n] [n^{2}] \sup\mathbb{P} Figure 1: The one-parameter family of polynomial orders of growth can be represented as a line in the space of orders of growth. Moreover, by taking the completion of this set, the infimum and supremum can be identified as the endpoints of the family. The regime of positive topological entropy is related to the exponential one-parameter family of orders of growth \mathbb{E}=\{[e^{tn}]\in\mathbb{O}:t>0\} . Specifically, there is a natural way to project the generalized entropy of a map f onto this family by considering the set I_{\mathbb{E}}(f)=\{t>0:o(f)\leq[e^{tn}]\}\subset\mathbb{R} and defining the projection map \pi_{\mathbb{E}}:\overline{\mathbb{O}}\to[0,\infty] by \pi_{\mathbb{E}}(o(f))=\inf I_{\mathbb{E}}(f) if this set is non-empty, and \pi_{\mathbb{E}}(o(f))=\infty otherwise. Justifying this projection, by Theorem 2 in [ 3 ] , we have \pi_{\mathbb{E}}(o(f))=h(f) . Consequently, one can extend the notion of topological entropy to the topological entropy with respect to the one-parameter family \mathbb{A} , given by the projection \pi_{\mathbb{A}} , and this quantity is denoted by h_{\mathbb{A}}(f) . Corollary 2 . If h_{\mathbb{A}}(f_{*})>0 ,

Read the complete text