2025-11-12T07:19:09.653600

Orders of Oscillation Motivated by Sarnak's Conjecture--Part II

Jiang
This work is a continuation of [13]. We study the linear disjointness between higher-order oscillating sequences and nonlinear dynamical systems. Specifically, we prove that any oscillating sequence of order $m=d+k-1$ and any simple polynomial skew product of degree $k$ on the $d$-Euclidean space are linearly disjoint. Additionally, we demonstrate that any oscillating sequence of order $d$ and any minimal mean attractable and minimal quasi-discrete spectrum dynamical system of order $d$ are linearly disjoint. Finally, we introduce multi-linearly disjoint sequences and construct examples of such sequences.
academic

Orders of Oscillation Motivated by Sarnak's Conjecture--Part II

Basic Information

  • Paper ID: 2201.08800
  • Title: Orders of Oscillation Motivated by Sarnak's Conjecture--Part II
  • Author: Yunping Jiang (Queens College & Graduate Center, City University of New York)
  • Classification: math.DS (Dynamical Systems), math.NT (Number Theory)
  • Publication Date: arXiv preprint, submitted January 2022, revised November 2025 (v2)
  • Paper Link: https://arxiv.org/abs/2201.08800v2

Abstract

This paper continues the author's previous work 13, investigating linear disjointness between higher-order oscillating sequences and nonlinear dynamical systems. The main results establish: (1) Any oscillating sequence of order m=d+k−1m=d+k-1 is linearly disjoint from simple polynomial skew products of degree kk on dd-dimensional Euclidean space; (2) Any oscillating sequence of order dd is linearly disjoint from dynamical systems that are minimally mean attracting (MMA) and minimally quasi-discrete spectrum (MQDS); (3) The concept of multiple linear disjointness is introduced with concrete examples constructed.

Research Background and Motivation

Core Problem

The core problem studied in this paper is a generalization and deepening of Sarnak's Conjecture. Sarnak's conjecture asserts that the Möbius sequence u=(μ(n))n∈Nu = (\mu(n))_{n\in\mathbb{N}} is linearly disjoint from any continuous dynamical system with zero topological entropy. This is an important problem at the intersection of number theory and dynamical systems.

Importance of the Problem

  1. Connecting Number Theory and Dynamical Systems: Sarnak's conjecture establishes profound connections between arithmetic functions in number theory (such as the Möbius function) and dynamical systems theory
  2. Theoretical Significance: Linear disjointness is a key concept for understanding the interaction between deterministic systems and "pseudorandom" sequences
  3. Generalization Value: Studying properties of general oscillating sequences helps understand which sequences possess properties similar to the Möbius sequence

Limitations of Existing Methods

  1. Previous Work: The author studied in 13 linear disjointness between oscillating sequences and affine toral maps, but only in the linear case
  2. Nonlinear Challenges: For nonlinear dynamical systems (such as polynomial skew products), higher-order oscillation properties are required
  3. Spectral Theory Limitations: Classical discrete spectrum system theory is insufficient for handling more general dynamical systems

Research Motivation

The author aims to:

  • Generalize from the linear case to nonlinear polynomial skew products
  • Introduce the concept of order for oscillating sequences to precisely characterize required oscillation properties
  • Explore which dynamical systems require what order of oscillating sequences to guarantee linear disjointness

Core Contributions

  1. Generalization to Nonlinear Systems: Proves that oscillating sequences of order m=d+k−1m=d+k-1 are linearly disjoint from simple polynomial skew products of degree kk on dd-dimensional space (Theorem 1)
  2. General Polynomial Skew Products: For general polynomial skew products, proves the existence of a finite order m=m(d,k,h1,…,hd)m=m(d,k,h_1,\ldots,h_d) such that oscillating sequences of this order are linearly disjoint from them (Theorem 2)
  3. Spectral Theory Generalization: Generalizes quasi-discrete spectrum (QDS) systems to minimally quasi-discrete spectrum (MQDS) systems, and proves that oscillating sequences of order dd are linearly disjoint from MMA and MQDS(d) systems (Theorem 3)
  4. Multiple Linear Disjointness: Introduces the new concept of multiple linear disjointness sequences and constructs concrete examples (Theorem 4), which relates to the Chowla conjecture
  5. Theoretical Framework Refinement: Generalizes the definition of linear disjointness to dynamical systems on Polish spaces, providing a framework for handling non-compact spaces

Methodology Details

Task Definition

Linear Disjointness (Classical Definition): Let XX be a compact metric space, f:X→Xf: X \to X a continuous map, and c=(cn)n∈Nc = (c_n)_{n\in\mathbb{N}} a sequence of complex numbers. We say cc is linearly disjoint from ff if for all ϕ∈C(X)\phi \in C(X) and x∈Xx \in X, lim⁡N→∞1N∑n=1Ncnϕ(fnx)=0\lim_{N\to\infty} \frac{1}{N}\sum_{n=1}^N c_n \phi(f^n x) = 0

Order of Oscillating Sequences: We say cc is an oscillating sequence of order dd if for all real polynomials PP of degree at most dd, lim⁡N→∞1N∑n=1Ncne2πiP(n)=0\lim_{N\to\infty} \frac{1}{N}\sum_{n=1}^N c_n e^{2\pi i P(n)} = 0

Generalized Linear Disjointness: For a Polish space YY and quotient space X=Y/∼X = Y/\sim, projection π:Y→X\pi: Y \to X, and continuous map f:Y→Yf: Y \to Y, we say cc is linearly disjoint from ff if for all ϕ∈C(X)\phi \in C(X) and y∈Yy \in Y, lim⁡N→∞1N∑n=1Ncnϕ(π(fny))=0\lim_{N\to\infty} \frac{1}{N}\sum_{n=1}^N c_n \phi(\pi(f^n y)) = 0

Core Technical Framework

1. Structural Analysis of Polynomial Skew Products

Simple polynomial skew products are defined as:

x_1 + a \\ x_2 + h_2(x_1) \\ x_3 + b_{32}x_2 + h_3(x_1) \\ \vdots \\ x_d + b_{d(d-1)}x_{d-1} + \cdots + b_{d2}x_2 + h_d(x_1) \end{pmatrix}$$ where $h_i(x_1)$ are polynomials of degree at most $k$. **Key Observation**: After iteration, each coordinate $x_i^n = f^n(x)$ is a polynomial in $n$ with precisely computable degree: - $x_1^n$ is a polynomial of degree $\leq 1$ - $x_2^n$ is a polynomial of degree $\leq k+1$ - $x_i^n$ is a polynomial of degree $\leq k+i-1$ Therefore, for the evolution of the trigonometric polynomial $e^{2\pi i(k_1 x_1 + \cdots + k_d x_d)}$ along the orbit: $$e^{2\pi i(k_1 x_1^n + \cdots + k_d x_d^n)} = e^{2\pi i P_k(n)}$$ where $P_k(n)$ is a polynomial of degree at most $d+k-1$. #### 2. Trigonometric Polynomial Approximation Strategy **Proof Strategy**: 1. Use the Stone-Weierstrass theorem; trigonometric polynomials are dense in $C(T^d)$ 2. For any continuous function $\phi$, construct an approximating sequence of trigonometric polynomials $\{p_q\}$ 3. Decompose the sum into two parts: $$S_N\phi(x) = \underbrace{\frac{1}{N}\sum_{n=1}^N c_n(\phi(f^n x) - p_r(f^n x))}_{I} + \underbrace{\frac{1}{N}\sum_{n=1}^N c_n p_r(f^n x)}_{II}$$ **First Part Estimate (Error Term)**: Using Hölder's inequality, for dual exponents $\lambda, \lambda'$ (satisfying $1/\lambda + 1/\lambda' = 1$): $$|I| \leq \left(\frac{1}{N}\sum_{n=1}^N |c_n|^\lambda\right)^{1/\lambda} \left(\frac{1}{N}\sum_{n=1}^N |\phi(f^n x) - p_r(f^n x)|^{\lambda'}\right)^{1/\lambda'} \leq C^{1/\lambda} \cdot \frac{\epsilon}{2C^{1/\lambda}}$$ **Second Part Estimate (Main Term)**: Using the definition of oscillating sequences, for the trigonometric polynomial $$p_r(f^n x) = \sum_{m_{1r}\leq k_1 \leq s_{1r}} \cdots \sum_{m_{dr}\leq k_d \leq s_{dr}} a_{k,r} e^{2\pi i P_k(n)}$$ each term satisfies $$\left|\frac{1}{N}\sum_{n=1}^N c_n e^{2\pi i P_k(n)}\right| < \frac{\epsilon}{2L^d}$$ Therefore $|II| < \epsilon/2$. #### 3. Treatment of MMA and MQDS Systems **Minimally Mean Attracting (MMA)**: Every point in the system is mean attracted to some minimal set, i.e., $$\limsup_{N\to\infty} \frac{1}{N}\sum_{n=1}^N d(f^n x, f^n z) < \epsilon$$ **Minimally Quasi-Discrete Spectrum (MQDS(d))**: On minimal sets, the system has quasi-discrete spectrum structure of order $d$. This means there exists a hierarchical structure of unimodular eigenvectors: $$G_1 \subseteq G_2 \subseteq \cdots \subseteq G_{d+1}$$ where $G_n = \ker(\Lambda^n)$. **Proof Strategy**: 1. Use the MMA property to reduce the behavior of general points to points on minimal sets 2. On minimal sets, use the MQDS structure to decompose functions as linear combinations of quasi-eigenvectors 3. For quasi-eigenvectors $g \in G_n$ of order $n-1$, $$\Phi^n g(z) = e^{2\pi i P(n)}$$ where $P(n)$ is a polynomial of degree at most $d$ 4. Apply the definition of oscillating sequences to complete the proof ### Technical Innovations 1. **Precise Order Matching**: Proves that the required order of oscillating sequences is exactly $m=d+k-1$, which is tight (cannot be lowered) 2. **Inductive Construction**: Through induction, precisely computes the polynomial degree of each coordinate, avoiding crude estimates 3. **Unified Framework**: Incorporates both the affine case (Theorem A) and polynomial case (Theorem 1) into a unified framework 4. **Spectral Theory Generalization**: Generalizes classical discrete spectrum theory to quasi-discrete spectrum and introduces the order concept 5. **Polish Space Generalization**: Extends definitions to non-compact spaces, providing tools for studying more general dynamical systems ## Experimental Setup This is a pure mathematics theory paper without numerical experiments. Main results are obtained through rigorous mathematical proofs. ### Theoretical Verification Methods 1. **Constructive Examples**: Concrete completely oscillating sequences are constructed through Theorems B and 4 2. **Equivalence Proofs**: Proposition 1 proves the equivalence between oscillating sequences in the arithmetic sense and the standard definition 3. **Corollary Verification**: Multiple corollaries are derived from main theorems to verify theoretical consistency ### Mathematical Tools Used 1. **Harmonic Analysis**: Stone-Weierstrass theorem, trigonometric polynomial approximation 2. **Probability Theory**: Hölder's inequality, equidistribution theory 3. **Number Theory**: Weyl's criterion, Koksma's theorem 4. **Dynamical Systems Theory**: Topological entropy, quasi-discrete spectrum theory 5. **Functional Analysis**: Koopman operator theory ## Experimental Results ### Main Theoretical Results **Theorem 1 (Simple Polynomial Skew Products)**: - **Statement**: Any oscillating sequence of order $m=d+k-1$ is linearly disjoint from simple polynomial skew products of degree $k$ - **Significance**: Generalizes previous work from the linear case ($k=1$, corresponding to affine maps) to arbitrary degree polynomials - **Optimality**: The order $m=d+k-1$ is necessary and cannot be lowered **Theorem 2 (General Polynomial Skew Products)**: - **Statement**: For general polynomial skew products, there exists a finite order $m=m(d,k,h_1,\ldots,h_d)$ such that the conclusion holds - **Limitation**: The order depends on specific polynomial coefficients, less precise than Theorem 1 **Theorem 3 (MMA and MQDS Systems)**: - **Statement**: Oscillating sequences of order $d$ are linearly disjoint from MMA and MQDS(d) systems - **Application Range**: Covers a large class of zero topological entropy systems, including those with quasi-discrete spectrum **Theorem 4 (Multiple Linear Disjointness)**: - **Statement**: For $g \in C_+^2((1,\infty))$, for almost all parameters $(\alpha, \beta)$, the sequence $$c = (e^{2\pi i(\alpha\beta^n g(\beta))})_{n\in\mathbb{N}}$$ is multiply linearly disjoint - **Significance**: Provides examples of completely oscillating sequences other than the Möbius sequence ### Theoretical Corollaries **Corollary 1**: Applies Theorem 1 to iterations of affine toral maps, handling the case where eigenvalues are roots of unity but not all equal to 1 **Corollary 2**: Generalizes Corollary 1 to the general polynomial case **Corollary A**: Directly from Theorem A, any oscillating sequence of order $d$ is linearly disjoint from affine toral maps with zero topological entropy ### Key Lemmas **Proposition 1 (Equivalence)**: - Oscillating sequences in the standard sense are equivalent to those in the arithmetic sense - Proof uses orthogonality of roots of unity and polynomial extension techniques **Lemma 1**: - All integer matrices with eigenvalue 1 can be transformed to lower triangular form - This provides a standard form for handling affine skew products ### Theoretical Findings 1. **Necessity of Order**: Different types of dynamical systems require different orders of oscillating sequences, reflecting system complexity 2. **Nonlinear Effects**: The degree of polynomial nonlinear terms directly affects the required order of oscillating sequences 3. **Spectral Theory Limitations**: Remark 4 points out that general zero topological entropy toral maps are not necessarily MQDS(d), indicating need for finer classification 4. **Role of Entropy**: The completely oscillating sequences constructed in Theorem B have positive entropy (as dynamical systems themselves), yet remain linearly disjoint from zero entropy systems ## Related Work ### Number Theory Foundations 1. **Davenport [7]**: Proved that the Möbius sequence is an oscillating sequence (order 1) 2. **Hua [12]**: Proved that the Möbius sequence is a completely oscillating sequence 3. **Chowla [5]**: Proposed conjectures about multiple correlations of the Möbius sequence ### Dynamical Systems Theory 1. **Fomin [8]**: Introduced pure point spectrum dynamical systems and MLS concept 2. **Abramov [2], Hahn & Parry [11]**: Developed quasi-discrete spectrum theory 3. **Sinai [19]**: Characterized topological entropy of affine toral maps 4. **Kronecker [14]**: Proved properties of matrices with roots of unity as eigenvalues ### Author's Previous Work 1. **Jiang [9]**: Defined oscillating sequences, studied linear disjointness with MLS systems 2. **Fan & Jiang [9]**: Introduced MMA and MMLS concepts 3. **Jiang [13]**: Defined order of oscillating sequences, proved linear disjointness with affine skew products (Theorem A) 4. **Akiyama & Jiang [3]**: Constructed examples of completely oscillating sequences (Theorem B) ### Related Progress 1. **el Abdalaoui [1]**: Used Gowers norms to study oscillating sequences 2. **Byszewski et al. [4]**: Studied Gowers norms of automatic sequences 3. **Haase & Moriakov [10]**: Systematically studied quasi-discrete spectrum systems 4. **Dai & Jiang [6]**: Defined distance entropy on non-compact spaces ### Positioning of This Work This paper advances the field in the following ways: 1. **From Linear to Nonlinear**: First systematic treatment of polynomial skew products 2. **Precise Order Theory**: Establishes exact relationships between oscillation order and system complexity 3. **Spectral Theory Generalization**: Connects quasi-discrete spectrum theory to linear disjointness 4. **Multiple Linear Disjointness**: Introduces new concept, connecting to Chowla conjecture ## Conclusions and Discussion ### Main Conclusions 1. **Successful Nonlinear Generalization**: Proves that oscillating sequence theory effectively handles polynomial nonlinear systems 2. **Refined Order Theory**: Establishes clear order hierarchy: - Order 1: Linearly disjoint from linear rotations - Order $d$: Linearly disjoint from $d$-dimensional affine skew products - Order $d+k-1$: Linearly disjoint from $d$-dimensional degree $k$ polynomial skew products 3. **Spectral Theory Connection**: Through the MQDS concept, connects spectral theory to oscillating sequence theory 4. **Constructive Results**: Provides numerous completely oscillating sequence examples beyond the Möbius sequence ### Limitations 1. **General Polynomial Case**: In Theorem 2, the order $m(d,k,h_1,\ldots,h_d)$ depends on specific coefficients without a uniform bound 2. **Non-polynomial Case**: For general continuous toral maps (Problem 1), remains open 3. **Necessity of MMA Condition**: Unclear whether the MMA condition can be weakened 4. **Universality of MQDS**: Remark 4 shows general zero entropy systems are not necessarily MQDS, limiting Theorem 3's applicability 5. **Computational Complexity**: Computing the minimum required oscillation order for specific systems may be difficult ### Future Directions **Problem 1**: Are completely oscillating sequences linearly disjoint from all continuous toral maps with zero topological entropy? **Problem 2**: Are completely oscillating sequences constructed in Theorem B (with positive entropy) linearly disjoint from all zero entropy systems? **Potential Research Directions**: 1. Generalize to more general nonlinear systems (rational functions, transcendental functions) 2. Study general continuous maps on higher-dimensional tori 3. Explore applications of multiple linear disjointness 4. Develop algorithms to determine minimum required oscillation order for given systems 5. Study relationships between oscillating sequences and positive entropy systems ## In-Depth Evaluation ### Strengths 1. **Theoretical Depth**: - Establishes precise quantitative relationships between oscillation sequence order and dynamical system complexity - Proof techniques are innovative, particularly the inductive construction of polynomial degrees - Unifies previous work into a systematic theoretical framework 2. **Mathematical Rigor**: - All theorems have complete rigorous proofs - Logical relationships between lemmas and corollaries are clear - Proposition 1's equivalence proof eliminates definitional ambiguities 3. **Originality**: - First systematic study of linear disjointness for nonlinear polynomial skew products - Introduces MQDS concept, generalizing classical spectral theory - Multiple linear disjointness is a novel concept 4. **Writing Quality**: - Clear structure, progressing from simple to complex - Motivation well explained, connection to Sarnak's conjecture explicit - Technical details complete, highly reproducible 5. **Theoretical Significance**: - Deepens understanding of Sarnak's conjecture - Provides new tools for cross-disciplinary research between number theory and dynamical systems - Constructed examples (Theorem 4) have independent value ### Weaknesses 1. **Limited Application Scope**: - Theorem 1 applies only to "simple" polynomial skew products (special form) - Theorem 2's order estimate lacks precision - MQDS condition is strong, excluding many zero entropy systems 2. **Many Open Problems**: - Problem 1 (general continuous toral maps) remains unsolved - Answer to Problem 2 (positive entropy oscillating sequences) unclear - Remark 4 indicates theoretical framework limitations 3. **Lack of Numerical Verification**: - While a pure mathematics paper, some numerical experiments might aid understanding - For example, computing oscillation orders of specific examples 4. **Technical Complexity**: - Proofs involve multiple mathematical branches, potentially unfriendly to non-specialists - Some technical details (e.g., Hölder inequality applications) could be more detailed 5. **Distance from Sarnak's Conjecture**: - Sarnak's conjecture involves all zero entropy systems, while this work treats specific categories - Special properties of Möbius sequences (e.g., connections to automorphic forms) underutilized ### Impact 1. **Contribution to the Field**: - Provides new perspectives and tools for Sarnak's conjecture research - Oscillation order theory may become standard in the field - Connects number theory, dynamical systems, harmonic analysis and other fields 2. **Practical Value**: - While theoretical, provides methods for determining properties of specific systems - Constructed completely oscillating sequences useful for testing other conjectures - Technical methods (e.g., trigonometric polynomial approximation) have universal applicability 3. **Reproducibility**: - All proofs are constructive, verifiable in principle - Theorem 4 construction generates concrete sequence examples - Clear theoretical framework facilitates subsequent research 4. **Directions for Future Research**: - Motivates multiple open problems (Problems 1, 2) - Multiple linear disjointness concept may spawn new research directions - MQDS theory deserves further development ### Applicable Scenarios 1. **Theoretical Research**: - Studying Sarnak's conjecture and variants - Developing oscillating sequence theory - Exploring number theory and dynamical systems connections 2. **Specific System Analysis**: - Determining linear disjointness of polynomial skew products - Analyzing iteration behavior of affine toral maps - Studying properties of quasi-discrete spectrum systems 3. **Constructing Counterexamples**: - Using Theorem 4 to construct sequences with specific properties - Testing boundary cases of other conjectures 4. **Educational Use**: - Case study for cross-disciplinary courses in number theory and dynamical systems - Demonstrating harmonic analysis applications in dynamical systems ## Key References 1. **Sarnak, P.** (2009, 2012): Original literature proposing Sarnak's conjecture 2. **Jiang, Y.** (2019): Foundational work for this paper (Theorem A) 3. **Fan, A. & Jiang, Y.** (2018): Source of MMA and MMLS concepts 4. **Davenport, H.** (1937) & **Hua, L.** (1966): Classical results on Möbius sequence oscillation 5. **Kuipers, L. & Niederreiter, H.** (1974): Standard reference for equidistribution theory 6. **Haase, M. & Moriakov, N.** (2015): Modern treatment of quasi-discrete spectrum systems --- **Overall Assessment**: This is a high-quality theoretical mathematics paper achieving substantial progress in Sarnak's conjecture research. By introducing order theory for oscillating sequences, the author establishes precise quantitative relationships between number-theoretic sequences and dynamical system complexity. Main results (Theorems 1-4) are technically innovative with rigorous complete proofs. While distance remains from fully resolving Sarnak's conjecture, this work provides solid foundations for future research in this direction and raises multiple worthwhile questions. For scholars working in number theory, dynamical systems, or related interdisciplinary fields, this is essential reading.