⚡ electrical engineering

Hill-Type Stability Analysis of Periodic Solutions of Fractional-Order Differential Equations

This paper proposes a Liouville-Weyl framework to enable the existence of periodic solutions in fractional-order differential equations and demonstrates that while an extended Floquet theory using Hill matrices can assess exponentially growing perturbations, it fundamentally fails to capture algebraically decaying solutions, a limitation that persists even in time-invariant systems.

Original authors: Paul-Erik Haacker, Remco I. Leine, Renu Chaudhary, Kai Diethelm, André Schmidt, Safoura Hashemishahraki

Published 2026-05-28

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Paul-Erik Haacker, Remco I. Leine, Renu Chaudhary, Kai Diethelm, André Schmidt, Safoura Hashemishahraki

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to predict the weather. In the "old days" of physics (classical integer-order math), if you had a system that repeated itself every day (like the seasons), you could use a specific set of rules to figure out if a small change—like a butterfly flapping its wings—would eventually blow the whole system apart or fade away harmlessly. These rules are called Floquet theory, and they rely on a mathematical tool called the Hill matrix to spot dangerous, growing storms.

However, many real-world systems (like rubber stretching, blood flowing, or materials with "memory") don't follow those old rules. They are better described by Fractional-Order Differential Equations (FODEs). These equations remember the past, which makes them very accurate but also very tricky.

Here is the core problem the paper tackles: The old rules don't work for these "memory" systems. Specifically, the standard math tools say that these systems can't have perfect, repeating cycles (periodic solutions) if you start them at a specific time (like $t=0$ ). It's like trying to make a perfect circle on a piece of paper that has a tear in the middle; the math breaks.

The Authors' Solution: Rewriting the Rules

To fix this, the authors propose a new framework. Instead of starting the clock at zero, they imagine the system has been running forever, from the distant past ( $t = -\infty$ ) up to now. They call this the Liouville-Weyl approach.

Think of it like this:

Old Way (Caputo): You start a race at the starting line. The runner has no memory of where they were before the gun went off.
New Way (Liouville-Weyl): The runner has been running forever. They have a "memory" of their entire history. This allows for perfect, repeating loops (periodic solutions) that the old way simply couldn't handle.

The Big Discovery: The One-Way Mirror

Once they established this new framework, they tried to apply the old "Storm Detection" rules (Floquet theory) to see if these repeating loops are stable or if they will explode.

They found a surprising limitation, which is the main point of the paper:

The "Explosion" Detector Works: If a system is unstable and the disturbance is going to grow exponentially (like a snowball rolling down a hill getting bigger and bigger), the new Hill matrix method can find it. It successfully identifies the "dangerous" growing solutions.
The "Fading" Detector Fails: If a system is stable and the disturbance is supposed to die down (like a ripple in a pond that eventually becomes flat), the new method cannot find it.

Why?
In the old "no-memory" world, things usually die down exponentially (like a ball rolling to a stop). In this new "memory" world, things die down much slower, like a heavy object sinking through thick honey (algebraic decay). The mathematical tool the authors built (the Hill matrix) is tuned to look for the fast, exponential growth. It is essentially "blind" to the slow, honey-like fading.

The Analogy of the "Infinite Hill"

The authors use a Hill Matrix, which is like an infinite ladder of numbers.

In the old world, climbing this ladder tells you exactly how fast a ball will roll down (decay) or up (grow).
In this new fractional world, the ladder still works perfectly if the ball is rolling up (growing unstable).
But if the ball is rolling down (decaying/stabilizing), the ladder has a gap. The ball falls through the cracks because the math of "memory" doesn't fit the shape of the ladder.

Summary of Findings

What they did: They created a new mathematical playground where fractional systems can have perfect repeating cycles.
What they tested: They tried to use a standard stability test (the Hill method) on these new systems.
What they found: The test is a great "alarm system" for things getting worse (instability). However, it is not a "calm-down system" for things getting better (stability). It cannot see the slow, algebraic decay that is characteristic of systems with memory.
The Conclusion: You can use this method to prove a system is unstable, but you cannot use it to prove a system is stable just because it didn't find a growing solution. The "fading" solutions are invisible to this specific tool.

The paper essentially says: "We built a better map for these memory-based systems, and we found a tool that works great for spotting disasters, but it's useless for spotting safety. We need new tools to find the safety."

Technical Summary: Hill-Type Stability Analysis of Periodic Solutions of Fractional-Order Differential Equations

Problem Statement
The paper addresses the stability analysis of periodic solutions in nonlinear fractional-order differential equations (FODEs). A fundamental limitation of classical Caputo-type FODEs is that they do not admit exactly periodic solutions; periodic functions under Caputo derivatives are only asymptotically periodic. Consequently, standard stability analysis tools, such as Floquet theory and the Hill matrix method, cannot be directly applied to the linearized perturbation dynamics of periodic solutions in Caputo systems. While Liouville-Weyl-type FODEs (defined with an infinite lower bound, $-\infty$ ) do admit exactly periodic solutions, a rigorous Floquet theory for these systems has not been established. Previous attempts to extend Floquet theory to fractional-order systems have relied on approximations (e.g., assuming multiplicative identities for Mittag-Leffler functions) or have failed to characterize stability boundaries and bifurcation types comprehensively.

Methodology
The authors propose a framework based on Liouville-Weyl-type FODEs to study periodic solutions. The core methodology involves:

System Formulation: Defining a class of nonlinear Liouville-Weyl differential equations with an infinite lower bound ( $-\infty$ ). The initial value problem is formulated with an initialization function $x_0(t)$ defined on $(-\infty, t_0]$ , belonging to a specific function space $X(-\infty, t_0]$ that ensures the existence of the forcing term derived from the history of the system.
Linearization: The stability of a periodic solution $x_p(t)$ is analyzed by linearizing the system around it, resulting in a linear time-periodic fractional-order differential equation (LTP-FODE) with a periodic Jacobian matrix $J(t)$ .
Liouville-Weyl Derivative Properties: Utilizing the property that the Liouville-Weyl derivative of an exponential function $e^{st}$ is $s^\alpha e^{st}$ (for $\text{Re}(s) \ge 0$ ), the authors attempt to extend the Floquet ansatz $y(t) = e^{\lambda t} p(t)$ to fractional-order systems.
Fractional Hill Matrix: By substituting the Floquet ansatz into the LTP-FODE and utilizing Fourier series expansions for the periodic coefficients, the authors derive an infinite-dimensional nonlinear eigenvalue problem. This leads to the definition of the "fractional Hill matrix," $H^{(\alpha)}_\infty(\lambda)$ , where the diagonal blocks involve terms of the form $(\lambda + ik\omega)^\alpha I - J_k$ .
Analysis of Linear Time-Invariant (LTI) Subsystems: Before tackling the periodic case, the authors analyze LTI-FODEs with infinite memory. They compare solutions of the Liouville-Weyl system with those of the standard Caputo system (zero lower bound) to understand the behavior of decaying solutions.

Key Contributions and Results

Framework for Periodic Solutions: The paper establishes a rigorous initial value problem formulation for Liouville-Weyl FODEs that admits exactly periodic solutions, overcoming the limitation of Caputo derivatives.
Extension of Floquet Theory: The authors derive a fractional Hill matrix approach for LTP-FODEs. They prove that if a solution exists in the form $y(t) = e^{\lambda t} p(t)$ with $\text{Re}(\lambda) \ge 0$ , it must satisfy the nonlinear eigenvalue problem $H^{(\alpha)}_\infty(\lambda)p_\infty = 0$ .
Limitation on Decaying Solutions: A central finding is that the proposed fractional Hill method (and the exponential ansatz generally) can only assess exponentially growing solutions (instability). It fails to capture algebraically decaying solutions.
- In the LTI case, the authors show that while diverging solutions correspond to eigenvalues with positive real parts, converging solutions (which decay algebraically, $O(t^{-\alpha})$ ) do not correspond to eigenvalues with negative real parts in the characteristic equation derived from the exponential ansatz.
- This limitation is traced to the fact that solutions to Liouville-Weyl FODEs with stable parameters typically exhibit algebraic decay rather than exponential decay. The exponential ansatz is invalid for these decaying modes because they do not belong to the required function space for the Liouville-Weyl derivative to be defined in the standard way.
Numerical Validation: The authors demonstrate the method numerically on scalar and 2D systems (including a fractional Mathieu-type system). They show that the truncated fractional Hill matrix successfully identifies eigenvalues with positive real parts corresponding to unstable trajectories, confirming the method's utility for detecting instability.
Properties of the Fractional Hill Matrix: The paper proves that the eigenvalues of the truncated fractional Hill matrix come in complex conjugate pairs and are located within specific Gershgorin-like regions in the complex plane.

Significance and Claims
The paper claims that the primary merit of this work is the insight that exponentially decaying Floquet solutions do not exist for LTP-FODEs, while exponentially growing solutions can be observed and assessed via a Hill-type approach.

The authors state that their work provides a novel path to accessing solutions of LTP fractional differential equations where classical Floquet theory fails. They emphasize that while the method is limited to detecting instability (diverging solutions), it offers a rigorous alternative to transition curve analysis, which only identifies stability boundaries without characterizing the type of bifurcation or the nature of the unstable modes.

The paper concludes modestly, noting that the absence of eigenvalues with non-negative real parts in the fractional Hill problem is a strong indicator of stability but requires further investigation to be considered conclusive. Furthermore, the authors identify the construction of a monodromy operator for LTP-FODEs as an open challenge, suggesting that the spectrum of such an operator would be necessary to fully characterize stability, including the algebraically decaying modes that the current Hill method misses.

The Authors' Solution: Rewriting the Rules

The Big Discovery: The One-Way Mirror

The Analogy of the "Infinite Hill"

Summary of Findings

More like this