h-dichotomies via noncritical uniformity and expansiveness for evolution families

Basic Information

• Paper ID: 2511.03453
• Title: h-dichotomies via noncritical uniformity and expansiveness for evolution families
• Author: Davor Dragičević (University of Rijeka, Croatia)
• Classification: math.DS (Dynamical Systems)
• Submission Date: November 5, 2025 to arXiv
• Paper Link: https://arxiv.org/abs/2511.03453

Abstract

This paper generalizes the complete characterization of h-dichotomies for finite-dimensional ordinary differential equations by Elorreaga et al. in Math. Ann. 393 (2025) to arbitrary invertible evolution families acting on Banach spaces. The author employs a completely different and more concise approach from that of Elorreaga et al., based on the time rescaling technique introduced by Dragičević and Silva, characterizing h-dichotomies through h-expansiveness and h-noncritical uniformity concepts, extending classical results of Coppel and Palmer on exponential dichotomies.

Research Background and Motivation

1. Core Problem

The core problem this research addresses is: how to completely characterize the concept of h-dichotomy within the general framework of invertible evolution families on Banach spaces.

2. Problem Significance

• Exponential dichotomy (introduced by Perron in 1930) is a foundational concept in the qualitative theory of nonautonomous differential equations, serving as the nonautonomous counterpart of the classical concept of hyperbolicity
• It requires the phase space to decompose at each time into stable and unstable directions, exhibiting forward exponential contraction and backward exponential contraction (i.e., forward exponential expansion) respectively
• However, due to the flexibility of nonautonomous dynamical systems, many systems exhibit similar behavior but with contraction/expansion rates that are not necessarily exponential

3. Limitations of Existing Approaches

• Muldowney (1984) and Naulin-Pinto (1995) first systematically studied generalized dichotomies prescribed by general functions (growth rates)
• Elorreaga et al. (2025) obtained a complete characterization of h-dichotomies in the framework of finite-dimensional ordinary differential equations
• However, these results are limited to finite-dimensional spaces, and Elorreaga et al.'s method is based on the complex theory of completely ordered topological groups

4. Research Motivation of This Paper

• Generalize results to infinite-dimensional Banach spaces with general evolution families
• Employ the time rescaling method, providing a more concise and direct proof pathway than existing methods
• Establish equivalence relationships between h-dichotomy and h-expansiveness, h-noncritical uniformity

Core Contributions

1. Generalization of Elorreaga et al.'s main result: Extends the characterization of h-dichotomies for finite-dimensional ordinary differential equations to arbitrary invertible evolution families on Banach spaces (Theorem 2)
2. Development of time rescaling method: Systematically establishes the connection between h-dichotomies and exponential dichotomies, achieving equivalence between the two through constructing a new evolution family T^h (Proposition 2)
3. Obtaining new characterization of exponential dichotomies: As an intermediate result, provides a generalized version of Palmer's theorem for invertible evolution families (Theorem 1)
4. Equivalent characterization of new concepts: Proves that h-expansiveness and h-noncritical uniformity can be characterized through corresponding exponential versions (Propositions 3 and 4)
5. Providing more concise proof method: Avoids complex topological group theory, making the proof process more direct and understandable

Method Details

Task Definition

The core task of this paper is to establish the equivalence of the following three concepts:

Input:

• Invertible evolution family T = {T(t,s) : t ≥ s > a₀} ⊂ B(X), where X is a Banach space
• Growth rate function h : (a₀,∞) → (0,∞) (bijective increasing map)
• Interval [a₀*,∞), satisfying a₀* > a₀

Three concepts to be characterized:

1. h-dichotomy: Phase space decomposition with h-type contraction/expansion estimates
2. h-expansiveness: Vector norm can be controlled by evolution estimates at previous and subsequent times
3. h-noncritical uniformity: Vector norm is uniformly controlled by its evolution over small time windows

Model Architecture

1. Core Idea of Time Rescaling

For a given evolution family T and growth rate h, define a new evolution family: $T_h(t,s) := T(h^{-1}(e^t), h^{-1}(e^s)), \quad t,s \in \mathbb{R}, t \geq s$

Key observation: T's h-dichotomy is equivalent to T^h's exponential dichotomy.

2. Basic Definition System

Evolution Family (Definition 1):

• Satisfies T(t,t) = Id
• Semigroup property: T(t,s)T(s,r) = T(t,r)
• Continuity: t ↦ T(t,s)v is continuous

h-bounded growth (Definition 3): $\|T(t,s)\| \leq K\left(\frac{h(t)}{h(s)}\right)^\mu, \quad t \geq s$

h-dichotomy (Definition 5): There exist a projection family {P(t)} and constants D,λ > 0 such that:

• Projection invariance: P(t)T(t,s) = T(t,s)P(s)
• Stable direction estimate: $\|T(t,s)P(s)\| \leq D(h(t)/h(s))^{-λ}$, t ≥ s
• Unstable direction estimate: $\|T(t,s)(Id-P(s))\| \leq D(h(s)/h(t))^{-λ}$, t ≤ s

3. New Concept Definitions

h-expansiveness (Definition 6): There exist L,β > 0 such that $\|v\| \leq L\left[\left(\frac{h(t)}{h(a)}\right)^{-\beta}\|T(a,t)v\| + \left(\frac{h(b)}{h(t)}\right)^{-\beta}\|T(b,t)v\|\right]$ holds for all v ∈ X and a ≤ t ≤ b.

h-noncritical uniformity (Definition 7): There exist θ ∈ (0,1) and C > 0 such that $\|v\| \leq \theta \sup\{\|T(u,t)v\| : |\ln h(u) - \ln h(t)| \leq C\}$ holds for all v ∈ X and sufficiently large t.

Technical Innovation Points

1. Systematic Application of Time Rescaling

Innovation: Although the time rescaling idea appeared implicitly in previous work, this paper is the first to systematically handle the continuous time case.

Key Proposition 2: Establishes the following equivalence relationships

• T has h-bounded growth on [a₀*,∞) ⟺ T^h has bounded growth on [ln h(a₀*),∞)
• T has h-dichotomy on [a₀*,∞) ⟺ T^h has exponential dichotomy on [ln h(a₀*),∞)

Advantage: This allows all problems concerning h-dichotomies to be reduced to problems concerning exponential dichotomies, for which mature theory already exists.

2. Transitivity of Concept Equivalence

Proposition 3 (h-expansiveness equivalence): T is h-expansive ⟺ T^h is exponentially expansive

Proof strategy: Through variable substitution t → h^{-1}(e^t), utilizing the monotonicity of h^{-1}, transforming h-type estimates into exponential-type estimates.

Proposition 4 (h-noncritical uniformity equivalence): T is h-noncritically uniform ⟺ T^h is noncritically uniform

Technical details: The key is that the logarithmic transformation |ln h(u) - ln h(t)| = |ln(e^u) - ln(e^t)| = |u-t| preserves the structure of the time window.

3. Distinction from Elorreaga et al.'s Method

Experimental Setup

Note: This is a pure theoretical mathematics paper and does not include numerical experiments. The following analyzes the theoretical verification structure of the paper.

Theoretical Verification Framework

1. Counterexample Construction (Example 1)

Purpose: Proves that the assumption "S is complemented" in Theorem 1 cannot be omitted

Construction:

• Take a Hilbert space X and a hyperbolic operator A ∈ B(X) (from reference 5)
• A has an invariant closed subspace Y such that A₁ := A|_Y is not hyperbolic
• Define evolution family T(t,s) = A^{⌊t⌋-⌊s⌋}

Verification result:

• T has exponential dichotomy on [0,∞)
• T₁ = T|_Y is noncritically uniform but does not have exponential dichotomy
• Shows that noncritical uniformity cannot guarantee dichotomy unless S is complemented

2. Logical Chain of Theorem Proofs

Theorem 1 (Exponential dichotomy characterization)
    ↓ (Apply time rescaling)
Proposition 2 (h-dichotomy equivalent to exponential dichotomy)
    ↓ (Combine with)
Proposition 3 (h-expansiveness transformation) + Proposition 4 (h-noncritical uniformity transformation)
    ↓ (Derive)
Theorem 2 (h-dichotomy characterization)

Assumption Analysis

Assumptions of Theorem 2:

1. X is a reflexive Banach space
2. T has h-bounded growth and decay
3. There exists a closed subspace Z such that X = S ⊕ Z, where $S := \{v \in X : \sup_{t \geq a_0^*} \|T(t,a_0^*)v\| < +\infty\}$

Relaxations in Remark 8:

• If Z is finite-dimensional, the reflexivity assumption can be dropped
• If X is finite-dimensional, the existence of Z is automatically satisfied
• For evolution families generated by ODE, the h-bounded decay assumption can be dropped

Experimental Results

Main Theoretical Results

Theorem 1 (Complete Characterization of Exponential Dichotomy)

Statement: On reflexive Banach spaces, for invertible evolution families T with bounded growth and decay, the following are equivalent:

1. T has exponential dichotomy
2. T is exponentially expansive
3. T is noncritically uniform

Proof structure:

• (a)⇒(b): Directly derived from the projection decomposition of dichotomy to expansiveness estimate
• (b)⇒(c): By choosing an appropriate time window C such that 2Le^{-βC} < 1
• (c)⇒(a): The most complex direction, key steps:
  1. Prove exponential decay on stable subspace S(s) (equation 5.5)
  2. Use reflexivity to prove reverse exponential decay on unstable subspace Z (equation 5.6)
  3. Apply lemma from reference 10 to complete the dichotomy proof

Theorem 2 (Complete Characterization of h-Dichotomy)

Statement: Under the same assumptions, the following are equivalent:

1. T has h-dichotomy
2. T is h-expansive
3. T is h-noncritically uniform

Proof method: Complete reduction to Theorem 1 through the following correspondences:

• T's h-dichotomy ⟷ T^h's exponential dichotomy (Proposition 2(b))
• T's h-expansiveness ⟷ T^h's exponential expansiveness (Proposition 3)
• T's h-noncritical uniformity ⟷ T^h's noncritical uniformity (Proposition 4)
• Correspondence of stable subspaces: $S = \{v : \sup_{t \geq a_0^*} \|T(t,a_0^*)v\| < \infty\} = \{v : \sup_{t \geq \ln h(a_0^*)} \|T_h(t,\ln h(a_0^*))v\| < \infty\}$

Key Inequality Derivations

1. Derivation of Expansiveness Estimate (Theorem 1, (a)⇒(b))

Starting from the dichotomy condition, for any v ∈ X and t ∈ a,b: $v = P(t)v + (Id-P(t))v = T(t,a)P(a)T(a,t)v + T(t,b)(Id-P(b))T(b,t)v$

Applying dichotomy estimates: $\|v\| \leq De^{-λ(t-a)}\|T(a,t)v\| + De^{-λ(b-t)}\|T(b,t)v\|$

This is precisely the exponential version of the expansiveness definition (4.1).

2. Verification of Noncritical Uniformity (Theorem 1, (b)⇒(c))

From the expansiveness inequality, taking a = t-C, b = t+C: $\|v\| \leq L(e^{-βC}\|T(t-C,t)v\| + e^{-βC}\|T(t+C,t)v\|)$$\leq 2Le^{-βC} \sup\{\|T(u,t)v\| : |u-t| \leq C\}$

Choosing C sufficiently large so that θ := 2Le^{-βC} < 1 suffices.

3. Decay on Stable Subspace (Theorem 1, (c)⇒(a))

For v ∈ S(s), let ρ := sup_{t≥s} ∥T(t,s)v∥ < ∞. For t ≥ s+C, by noncritical uniformity: $\|T(t,s)v\| \leq \theta \sup_{|u-t| \leq C} \|T(u,t)T(t,s)v\| \leq \theta \sup_{u \geq s} \|T(u,s)v\| = θρ$

Therefore ρ = sup_{t∈s,s+C} ∥T(t,s)v∥ ≤ Ke^{μC}∥v∥, and further: $\|T(t,s)v\| \leq θρ \leq θKe^{μC}∥v∥ =: D∥v∥$

Iterative application yields exponential decay: ∥T(t,s)v∥ ≤ Be^{-α(t-s)}∥v∥.

Theoretical Findings

1. Universality of time rescaling: h-dichotomy theory can be completely understood through exponential dichotomy theory; the two are essentially equivalent
2. Geometric meaning of expansiveness: The expansiveness condition (4.1) indicates that the norm of a vector can be controlled by its "projections" in the past and future, which is the geometric essence of dichotomy
3. Dynamical meaning of noncritical uniformity: The condition θ < 1 in (4.2) indicates that the system cannot be "too close" to a critical state (where norm neither grows nor decays) within a local time window
4. Role of reflexivity: In proving (c)⇒(a), reflexivity ensures through weak compactness the reverse decay estimate on the unstable subspace Z

Related Work

1. Classical Theory of Exponential Dichotomy

• Perron (1930): First introduced the concept of exponential dichotomy
• Coppel (1978) 2: Systematically studied dichotomies in stability theory
• Palmer (2006) 13: Established the connection between exponential dichotomy and exponential expansiveness (prototype of Theorem 1 in this paper)
• Henry (1981) 9: Developed dichotomy theory in the geometric theory of semilinear parabolic equations

2. Generalized Dichotomies

• Muldowney (1984) 11: First systematically studied non-exponential generalized dichotomies
• Naulin-Pinto (1995) 12: Introduced (h,k)-dichotomy concept, where h and k control growth rates of stable and unstable directions respectively
• Dragičević et al. (2020) 4: Studied admissibility of evolution families and generalized dichotomies

3. Time Rescaling Method

• Dragičević-Sasu-Sasu (2022) 7: Developed time rescaling method for polynomial dichotomies in discrete nonautonomous systems
• Dragičević-Silva (2025) 6: Systematically developed time rescaling techniques for discrete time; this paper extends it to continuous time

4. Recent Progress

• Elorreaga-Peñe-Robledo (2025) 8: Complete characterization of h-dichotomies in finite-dimensional ODE framework using completely ordered topological group theory
• Peñe-Rivera Villagran (2025) 14: Studied uniform asymptotic h-stability

Relative Advantages of This Paper

Conclusions and Discussion

Main Conclusions

1. Core theorem: On reflexive Banach spaces, h-dichotomy, h-expansiveness, and h-noncritical uniformity for invertible evolution families are equivalent (Theorem 2)
2. Methodological contribution: The time rescaling technique provides a unified framework for studying generalized dichotomies, reducing complex h-type estimates to classical exponential-type estimates
3. Theoretical extension: Generalizes Palmer's characterization of exponential dichotomies to evolution families on infinite-dimensional Banach spaces (Theorem 1)
4. Necessity of conditions: Through counterexample (Example 1), proves that complementedness of the stable subspace S is a necessary condition for the theorem to hold

Limitations

1. Reflexivity Assumption

• Restriction: Requires Banach space X to be reflexive (or Z finite-dimensional)
• Reason: The proof requires using weak compactness to control the unstable direction
• Impact: Excludes certain important non-reflexive spaces (such as C0,1, L¹, etc.)

2. Complementedness Assumption

• Restriction: Requires the stable subspace S to have a closed complement Z
• Necessity: Example 1 shows this condition cannot be omitted
• Practical issue: Verifying this condition in concrete applications may be difficult

3. Bounded Growth and Decay Assumptions

• Restriction: Requires T to have both h-bounded growth and decay
• Relaxation: Remark 8(3) indicates this can be dropped for evolution families generated by ODE
• Open problem: Whether this condition can be further relaxed

4. Time Interval

• Restriction: Results apply only to half-infinite intervals [a₀*,∞)
• Practical application: Many problems concern behavior on finite time intervals
• Extension direction: Study similar characterizations on finite time intervals

Future Directions

Although the paper does not explicitly list future directions, the following can be inferred from the content:

1. Non-reflexive spaces: Develop proof techniques that do not rely on reflexivity
2. More general evolution families: Consider non-invertible or families without decay properties
3. Application research: Apply the theory to specific partial differential equations and dynamical systems
4. Unification of discrete and continuous: Establish a unified theoretical framework for discrete and continuous time
5. (h,k)-dichotomies: Generalize to cases where stable and unstable directions have different growth rates

In-Depth Evaluation

Strengths

1. Innovation of Method (★★★★★)

• Systematization of time rescaling: Although the idea existed, this is the first complete treatment for continuous time
• Simplified proofs: Avoids complex topological group theory, making proofs more transparent
• Unified framework: Establishes a bridge between h-type and exponential-type theories

2. Completeness of Theory (★★★★☆)

• Equivalence characterization: Provides complete equivalence relationships among three concepts
• Necessity verification: Proves necessity of assumption conditions through counterexamples
• Generalizability: From finite to infinite dimension, from ODE to general evolution families

3. Clarity of Writing (★★★★★)

• Clear structure: From preliminaries → time rescaling → new concepts → main theorems, with rigorous logic
• Precise definitions: All concepts have precise mathematical definitions
• Detailed proofs: Key steps are sufficiently explained

4. Elegance of Technique (★★★★★)

• Concise core idea: The transformation T_h(t,s) := T(h^{-1}(e^t), h^{-1}(e^s)) is extremely elegant
• Direct proof: The proof of Theorem 2 is almost a direct corollary of Theorem 1
• Strong generalizability: The method can be applied to other types of generalized dichotomies

Shortcomings

1. Restrictiveness of Assumption Conditions (★★★☆☆)

• Reflexivity: Limits the scope of applicability
• Complementedness: Difficult to verify in applications
• Room for improvement: Remark 8 provides partial relaxations, but still not general enough

2. Lack of Application Examples (★★☆☆☆)

• Pure theory: No concrete applications to differential equations or dynamical systems
• Usability: Readers find it difficult to judge the application value of the theory in practical problems
• Suggestion: Add 1-2 application examples from PDE or infinite-dimensional dynamical systems

3. Insufficient Comparison with Elorreaga et al.'s Work (★★★☆☆)

• Method comparison: Only briefly mentioned in introduction, lacking detailed comparison
• Advantage argument: Although claiming "more concise," no quantitative or detailed complexity analysis is provided
• Suggestion: Could add a section for detailed comparison of the two methods

4. Computational Complexity Analysis (★★☆☆☆)

• Missing: No discussion of computational complexity for verifying h-expansiveness or h-noncritical uniformity
• Practicality: Unclear whether theoretical characterizations can be transformed into computable criteria

Impact Assessment

1. Contribution to the Field (★★★★☆)

• Theoretical deepening: Advances the theory of generalized dichotomies in nonautonomous dynamical systems
• Method innovation: Time rescaling method may become a standard tool for studying such problems
• Bridge role: Connects classical exponential theory with modern generalized theory

2. Practical Value (★★★☆☆)

• Theoretical tool: Provides theoretical foundation for studying non-exponential dynamics
• Application potential: Can be applied to slow-growth systems in biomathematics, control theory, etc.
• Limitation: Verification of assumption conditions may limit direct application

3. Reproducibility (★★★★★)

• Complete proofs: All theorems have detailed proofs
• Precise definitions: Concept definitions are clear and unambiguous
• Rigorous logic: Reasoning chains are complete and easy to verify

4. Citation Value (★★★★☆)

• Foundational: Provides theoretical foundation for subsequent research
• Methodological: Time rescaling technique deserves to be tried in other problems
• Survey value: Good organization of the development of generalized dichotomy theory

Applicable Scenarios

1. Theoretical Research

• Nonautonomous dynamical systems: Study systems with non-exponential growth rates
• Evolution equations: Parabolic or hyperbolic equations on infinite-dimensional spaces
• Random dynamical systems: Generalized dichotomies combined with random perturbations

2. Potential Application Fields

• Biomathematics: Population dynamics with polynomial or logarithmic growth
• Climate models: Non-exponential changes on long time scales
• Economic dynamics: Stability analysis of nonlinear growth models

3. Inapplicable Scenarios

• Finite time problems: Method targets half-infinite intervals
• Non-invertible systems: Requires invertible evolution families
• Strong nonlinearity: Theory applies only to linear or linearized systems

Technical Details Supplement

Role of Key Lemmas

The paper proof references Lemma 4.2 from reference 10, which is the key step in deriving complete dichotomy from separate estimates on stable and unstable directions (equations 5.5 and 5.6). The role of this lemma is:

Input:

• Subspace decomposition X = S(t) ⊕ Z(t)
• Exponential decay estimate on S(t)
• Reverse exponential decay estimate on Z(t)

Output: There exists a projection family {P(t)} such that T has exponential dichotomy

This technical lemma avoids lengthy standard arguments.

Geometric Intuition of Time Rescaling

The geometric meaning of the transformation T_h(t,s) = T(h^{-1}(e^t), h^{-1}(e^s)):

• Time axis reparametrization: Original time τ → new time t = ln h(τ)
• Growth rate normalization: The growth of h(τ) is transformed into exponential growth of e^t
• Linearization of estimates: (h(t)/h(s))^λ → e^{λ(ln h(t) - ln h(s))} = e^{λ(t-s)}

This transformation essentially "straightens out" nonlinear growth rates into linear (exponential) growth rates.

References

The paper cites 17 references, mainly including:

Classical Monographs

• 1 Chicone-Latushkin (1999): Systematic treatment of evolution semigroups
• 2 Coppel (1978): Dichotomies in stability theory (extensively referenced in this paper)
• 9 Henry (1981): Geometric theory of semilinear parabolic equations

Key Papers

• 8 Elorreaga-Peñe-Robledo (2025): Work directly generalized by this paper
• 13 Palmer (2006): Exponential dichotomy and expansiveness (prototype of Theorem 1)
• 10 Van Minh et al. (1998): Provides key technical lemmas

Author's Previous Work

• 6 Dragičević-Silva (2025): Time rescaling method for discrete time
• 7 Dragičević-Sasu-Sasu (2022): Polynomial dichotomies

Summary

This is a high-quality theoretical mathematics paper that, through clever time rescaling technique, generalizes the important characterization theorem of h-dichotomies in a more concise manner. The paper's main value lies in methodological innovation and theoretical generalization, providing a unified research framework for generalized dichotomy theory in nonautonomous dynamical systems. Although there is room for improvement in application examples and generality of assumption conditions, as foundational theoretical work, its contribution is significant. In particular, the systematization of the time rescaling method may become a standard tool for studying such problems, with considerable potential for academic impact.

Recommendation Index: ★★★★☆ (4.5/5)
Suitable Readers: Researchers in dynamical systems theory, functional analysis, and qualitative theory of differential equations