Table of Contents
Fetching ...

On the generalized dimensions of physical measures of chaotic flows

Théophile Caby, Michele Gianfelice

TL;DR

The paper establishes a general dimension-transformation principle for physical measures of chaotic flows that are constructed as suspensions over Poincaré maps: for all $q\neq 1$, $D_q(\mu)=D_q(\mu_R)+1$, and $d_{\mu}(x)=d_{\mu_R}(\xi)+1$, with $\xi=\pi_1\circ\Theta^{-1}(x)$; if $\mu_R$ is exact-dimensional, the same shift holds for $D_1$. This result is first proved in the bounded roof function case and then extended to general roofs via truncation arguments. The framework is then applied to two canonical systems: (i) a Rössler-like flow whose $D_q$ spectrum is computed explicitly in terms of the singularity parameter $\alpha$, producing a sharp formula for $D_q(\hat{\mu})$ that aligns with numerical estimates; (ii) Lorenz-like flows, for which the existence and exact-dimensionality of the base measure yield $D_1(\mu)=D_1(\mu_R)+1$ and a universal lower bound $D_q^{-}(\mu)\ge 2$, providing a concrete benchmark for the spectrum in singular-hyperbolic attractors.

Abstract

We prove that if $μ$ is the physical measure of a $C^2$ flow in $\mathbb{R}^d, d \geq 3,$ diffeomorphically conjugated to a suspension flow based on a Poincaré application $R$ with physical measure $μ_{R}$, then $D_{q}(μ)=D_{q}(μ_{R})+1$, where $D_{q}$ denotes the generalized dimension of order $q \neq1$. We also show that a similar result holds for the local dimensions of $μ$ and, under the additional hypothesis of exact-dimensionality of $μ_{R}$, that our result extends to the case $q=1$. We apply these results to estimate the $D_{q}$ spectrum associated with Rössler systems and turn our attention to Lorenz-like flows, proving the existence of their information dimension and giving a lower bound for their generalized dimensions.

On the generalized dimensions of physical measures of chaotic flows

TL;DR

The paper establishes a general dimension-transformation principle for physical measures of chaotic flows that are constructed as suspensions over Poincaré maps: for all , , and , with ; if is exact-dimensional, the same shift holds for . This result is first proved in the bounded roof function case and then extended to general roofs via truncation arguments. The framework is then applied to two canonical systems: (i) a Rössler-like flow whose spectrum is computed explicitly in terms of the singularity parameter , producing a sharp formula for that aligns with numerical estimates; (ii) Lorenz-like flows, for which the existence and exact-dimensionality of the base measure yield and a universal lower bound , providing a concrete benchmark for the spectrum in singular-hyperbolic attractors.

Abstract

We prove that if is the physical measure of a flow in diffeomorphically conjugated to a suspension flow based on a Poincaré application with physical measure , then , where denotes the generalized dimension of order . We also show that a similar result holds for the local dimensions of and, under the additional hypothesis of exact-dimensionality of , that our result extends to the case . We apply these results to estimate the spectrum associated with Rössler systems and turn our attention to Lorenz-like flows, proving the existence of their information dimension and giving a lower bound for their generalized dimensions.
Paper Structure (6 sections, 13 theorems, 155 equations, 4 figures)

This paper contains 6 sections, 13 theorems, 155 equations, 4 figures.

Key Result

Theorem 2

Suppose $\mu$ is the physical measure of a $C^{2}$ flow in $\mathbb{R}^d$ constructed as in the previous section. Then, provided the different dimensions associated with $\mu_{R}$ exist, we have

Figures (4)

  • Figure 1: Up: Attractor of the velocity field $\tilde{f}_c$. The box $B$ is represented in blue. Down: Attractor of the flow generated by $\bar{f}_c$. The Poincaré section $\Sigma$ is represented in blue. In both cases, we took with $a=b=0.1$ and $c=18$
  • Figure 2: Left: Graphical representation of the unimodal map $\mathbf{T}$, associated with the Rössler flow of parameters $a=b=0.1$, $c=18$ (the axes are inverted). Right: 1-D cross-section of the attractor with the Poincaré section $\Sigma$ for different discretization steps $h$.
  • Figure 3: Schwartzian derivative of $\mathbf{T}$ for the parameters $a=b=0.1$ and $c=18$.
  • Figure 4: Representation of the $D_q$ spectrum of the measure $\hat{\mu}$, as given by formula (\ref{['dqrr']}) for $\alpha=1/2$.

Theorems & Definitions (18)

  • Definition 1
  • Theorem 2
  • Proposition 3
  • Lemma 4
  • Lemma 5
  • Proposition 6
  • Theorem 7
  • Remark 8
  • Remark 9
  • Proposition 10
  • ...and 8 more