🔢 mathematics

Residual-Certified Fractional and Time-Delayed Duffing Dynamics: A Volterra--Memory Framework for Continuous and Discrete Past Dependence

This paper presents a residual-certified computational framework using Volterra formulations and independent error metrics to systematically compare and distinguish the dynamical behaviors of continuous fractional memory, discrete time-delay, and hybrid mechanisms in nonlinear Duffing oscillators, demonstrating that their observed differences are intrinsic rather than numerical artifacts.

Original authors: Alvaro Salas

Published 2026-07-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Alvaro Salas

Original paper licensed under CC BY 4.0 (https://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

The Ghost of the Past and the Clock of Tomorrow

Imagine you are trying to predict how a swing moves. In a perfect, boring world, you only need to know where the swing is right now and how hard you are pushing it. But the real world is messy. In physics, engineering, and even biology, things often remember what happened to them a moment ago. This "memory" comes in two main flavors. The first is like a heavy, slow-moving fog: the past influences the present continuously, with every single moment from the beginning of time leaving a tiny, fading mark on what happens right now. Scientists call this fractional calculus, and it's great for describing things that have a long, smooth history, like how a rubber band slowly snaps back or how heat spreads through a weird material.

The second flavor of memory is more like a strict rulebook with a specific delay. Imagine a game where you can only react to what your opponent did exactly one second ago. Nothing else matters; the past before that one-second mark is erased. This is time-delay dynamics, and it's crucial for things like traffic jams, where a car brakes because of the car in front of it, but only after a split-second reaction time. For decades, scientists have treated these two types of memory as totally different languages. One is a continuous, flowing river of history; the other is a series of discrete, ticking snapshots. The big question is: if you use these two different languages to describe the same chaotic system, do they tell the same story, or are they just sounding similar by accident?

The Duffing Swing and the Detective Work

Enter the Duffing oscillator, which is essentially a fancy, wobbly swing that doesn't just go back and forth in a simple line. It has a "nonlinear" spring, meaning the harder you push it, the weirder it gets to predict. It's a perfect playground for chaos. In this paper, researcher Alvaro Salas sets up a high-stakes showdown between the "foggy" fractional memory and the "snapshot" time-delay memory using this wobbly swing. But here's the twist: Salas doesn't just look at the pictures of the swings moving. He knows that in computer simulations, two different things can look almost identical just because the computer made the same tiny mistake in both. So, he brings in a detective tool called residual certification.

Think of residual certification like a lie detector test for math. After the computer calculates a path for the swing, Salas takes that path and shoves it back into the original equation. If the math is perfect, the equation should balance to zero. If there's a leftover number (a "residual"), it means the path doesn't actually fit the rules. By measuring exactly how much the path "fails" to fit the rules, Salas can prove whether the differences he sees are real physical effects or just computer glitches.

The Showdown: Fog vs. Snapshots

Salas ran three different simulations on his computer to see how the swing behaved.

The Pure Fog: A swing with only the continuous, fractional memory (no specific delays).
The Pure Snapshot: A swing with only the specific time-delay feedback (no continuous memory).
The Hybrid: A swing that has both the foggy memory and the specific delay at the same time.

He ran these simulations for a total time of 60 units, using a very small step size of 0.01 to ensure high precision. The results were fascinating. While all three swings looked somewhat similar to the naked eye, they weren't identical. The "Hybrid" swing actually ended up with a smaller swing amplitude (how high it went) than either of the pure versions.

To prove this wasn't a computer error, Salas checked the "lie detector" scores. For the pure fractional model, the biggest error (residual) was 1.14897 × 10⁻⁴. For the hybrid model, it was 9.05144 × 10⁻⁵. For the pure delayed model, the error was even smaller, staying below 1.38922 × 10⁻⁵. Because these errors were so tiny and consistent, Salas could confidently say: "These differences are real." The continuous memory and the discrete delay are not interchangeable. They produce distinct, measurable differences in how the system behaves over time.

The Time-Traveling Clock

To add another layer of mystery, Salas also tested a "nonlinear clock." Imagine if time didn't tick at a steady rate but sped up or slowed down depending on the situation. He used a Caputo–Katugampola derivative, which is like putting the fractional memory inside a clock that runs at a different speed (specifically with a parameter ρ = 1.20). Even though the swing and the forces acting on it were the same, changing the "clock" changed the result. This showed that even within the world of fractional memory, the way you measure time matters just as much as the memory itself.

The Takeaway

So, what's the big deal? The paper suggests that while fractional memory and time-delay feedback can create similar-looking effects (like slowing down a swing or changing its rhythm), they are fundamentally different mechanisms. You can't just swap one for the other and expect the exact same result. The "fog" of continuous history and the "snapshot" of a specific delay interact with the chaotic Duffing swing in unique ways.

Salas concludes that scientists should stop studying these two types of memory in isolation. Instead, we need to look at them together, especially in complex systems where both distributed history and specific delays might be happening at once. The paper doesn't claim to have solved the mystery of chaos, but it has built a very reliable, "residual-certified" framework that proves these two memory types are distinct players in the game of nonlinear dynamics. It's a reminder that in the universe of math, sometimes two things that look alike are actually wearing different masks.

Technical Summary: Residual-Certified Fractional and Time-Delayed Duffing Dynamics

Problem Statement
Dynamical systems with memory are typically modeled using two distinct mathematical frameworks: fractional calculus, which represents the influence of the past through a continuous, distributed hereditary kernel, and time-delay dynamics, where the present state depends explicitly on discrete past values. While both approaches aim to model the phenomenon of past dependence, they are often studied in isolation. This paper addresses the need for a unified comparison of these mechanisms within a nonlinear context. Specifically, it investigates whether continuous fractional memory and discrete delayed feedback produce equivalent or distinct dynamical behaviors in Duffing-type oscillators, moving beyond visual trajectory comparison to a rigorous, residual-based verification.

Methodology
The study employs a unified framework comparing three specific models of the Duffing oscillator:

Pure Fractional Duffing: A system governed by Caputo fractional derivatives ( $0 < q < 1$ ) without delay.
Classical Delayed Duffing: A standard integer-order system ( $q=1$ ) incorporating linear delayed feedback ( $\kappa > 0$ ).
Hybrid Fractional-Delay Duffing: A system combining fractional derivatives ( $0 < q < 1$ ) with delayed feedback.

To ensure mathematical rigor, the fractional models are reformulated into Volterra integral equations, making the memory kernel explicit. The classical delayed model is treated via its differential form. The numerical solution utilizes:

A fractional Adams–Bashforth–Moulton predictor–corrector scheme for the Caputo systems.
A fourth-order explicit Runge–Kutta method for the classical delayed system.
Interpolation techniques to handle delayed states required at each time step.

A critical component of the methodology is residual certification. Rather than relying solely on the visual similarity of trajectories, the computed numerical solutions are substituted back into their respective governing equations (Volterra integrals for fractional systems; differential equations for delayed systems) to calculate the "defect" or residual. This quantifies how well the numerical solution satisfies the exact mathematical formulation. Additionally, a Caputo–Katugampola derivative formulation is included to test the effect of a nonlinear time scale (clock) on memory response.

Key Contributions
The paper makes five primary contributions:

Unified Framework: It establishes a single Duffing-type framework that simultaneously accommodates continuous fractional memory, discrete delayed feedback, and hybrid combinations.
Explicit Volterra Formulation: It derives Volterra integral representations for the fractional models, isolating the memory kernel for direct analysis.
Residual Certification Procedure: It introduces a rigorous method for certifying fractional and hybrid solutions by computing residuals directly from integral and differential formulations, ensuring that observed differences are not numerical artifacts.
Direct Comparison: It provides a quantitative comparison of late-time amplitude responses across the three models using identical parameters.
Clock Transformation Analysis: It demonstrates how a nonlinear time scaling (via the Caputo–Katugampola derivative) modifies fractional memory responses without altering the underlying nonlinear structure.

Results
Numerical experiments were conducted with a step size of $h = 0.01$ over a final time of $T = 60$ . The results indicate:

Distinct Dynamics: While fractional memory and delayed feedback produce comparable macroscopic effects (such as phase lag and amplitude modification), they are not identical. The hybrid model produces a late-time amplitude distinct from both pure limiting cases.
Amplitude Ordering: For the specific parameter set used ( $\delta=0.20, \gamma=0.30, \omega=1.00, \tau=0.80$ ), the late-time amplitudes followed the order: Hybrid ($1.19406$) < Pure Fractional ($1.22608$) < Pure Delayed ($1.24226$).
Residual Validation: The residuals confirmed high numerical consistency.
- Pure fractional Volterra residuals: Max $|R| \approx 1.149 \times 10^{-4}$ .
- Hybrid fractional-delay Volterra residuals: Max $|R| \approx 9.051 \times 10^{-5}$ .
- Classical delayed differential residuals: Max $|R| \approx 1.389 \times 10^{-5}$ .
  These low residual values confirm that the observed dynamical differences are inherent to the models, not discretization errors.
Clock Sensitivity: The Caputo–Katugampola comparison showed that altering the memory clock parameter ( $\rho$ ) significantly changes transient and late-time responses, even with the same forcing and nonlinearity.

Significance and Claims
The paper claims that its primary significance lies in providing a "practical bridge" between fractional and time-delayed descriptions of memory in nonlinear dynamics. By utilizing residual certification, the study demonstrates that the differences between continuous (distributed) and discrete (localized) memory mechanisms are real and measurable, rather than artifacts of visual interpretation or numerical error.

The authors assert that fractional and delayed formulations should be studied together rather than in isolation. The hybrid model is presented not as a simple average of the two mechanisms, but as a distinct dynamical system where distributed memory and discrete feedback interact through different channels of the nonlinear restoring force. The work suggests that this unified, residual-certified approach is essential for accurately modeling systems where both types of memory dependence may be present. Future work is noted to extend this comparison to bifurcation diagrams, resonance curves, and synchronization problems.

The Ghost of the Past and the Clock of Tomorrow

The Duffing Swing and the Detective Work

The Showdown: Fog vs. Snapshots

The Time-Traveling Clock

The Takeaway

More like this