🔢 mathematics

Optimal energy decay rates for Klein-Gordon equations with Kelvin-Voigt damping

This paper investigates the long-time behavior of a one-dimensional linear Klein-Gordon equation with Kelvin-Voigt damping, proving that solutions converge to zero and establishing an optimal polynomial energy decay rate despite the presence of multiple spectral points on the imaginary axis.

Original authors: Filippo Dell'Oro, Lassi Paunonen, David Seifert

Published 2026-03-30

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Filippo Dell'Oro, Lassi Paunonen, David Seifert

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

The Big Picture: A Vibrating String That Won't Stop (But Eventually Does)

Imagine you have a very long, infinite guitar string (representing the "Klein–Gordon equation"). You pluck it, and it starts vibrating. In the real world, air resistance and friction usually stop the string from vibrating eventually. This is called damping.

In this paper, the authors study a specific, tricky type of friction called Kelvin–Voigt damping. Think of this not just as air resistance, but as if the string itself is made of a thick, sticky material (like honey or silly putty). When the string bends, the material resists the change in shape, not just the movement.

The authors ask two main questions:

Will the string ever stop vibrating completely? (Does the energy go to zero?)
How fast will it stop? (Does it fade away quickly like a exponential decay, or slowly like a polynomial?)

The Surprise: The "Ghost" Frequencies

Usually, when you have a string on a finite length (like a guitar), the friction kills the vibration very fast. The energy drops off exponentially—like a light dimming instantly.

However, this paper studies a string that is infinite (the whole line). Here, things get weird. The authors discovered that this infinite string has two "ghost frequencies" (mathematically, points on the imaginary axis) that act like a stubborn echo.

The Analogy: Imagine a room with perfect acoustic echo. If you clap, the sound bounces back and forth. In a normal room, the sound dies out fast. In this "infinite" mathematical room, there are two specific pitches that the room refuses to absorb completely. They linger on the edge of silence.

Because of these two "ghost frequencies," the string cannot stop vibrating exponentially fast. The standard "quick kill" methods don't work here.

The Main Discovery: It Stops, But Slowly

Despite the stubborn echoes, the authors proved that the string does eventually stop vibrating. The energy does go to zero.

However, because of those two ghost frequencies, it doesn't stop quickly. It stops slowly, following a specific mathematical rule: $1/t^2$ .

The Analogy:
- Exponential Decay (Fast): Like a cup of hot coffee cooling down in a freezer. It gets cold very quickly.
- Polynomial Decay ( $1/t^2$ ) (Slow): Like a heavy boulder rolling down a hill covered in thick mud. It keeps moving for a long time, slowing down gradually. It never stops "suddenly"; it just gets slower and slower until it finally halts.

The paper proves that for a specific class of starting conditions (specific initial plucks), the energy of the vibration will fade away exactly at this $1/t^2$ rate. They also proved you can't do better than this; it is the optimal (best possible) speed for this specific setup.

The "Special Pluck" Requirement

The paper notes that this $1/t^2$ speed applies to a specific "class" of solutions.

The Analogy: Imagine you have a giant trampoline. If you jump randomly, the trampoline wobbles in a chaotic way. But if you jump in a very specific, coordinated rhythm (matching the "ghost frequencies"), the trampoline settles down in a predictable, slow pattern.
The authors identified exactly what those "special jumps" (initial data) look like. If your starting vibration fits this pattern, you get the $1/t^2$ decay. If you start with a "messy" vibration, it still stops, but the math to describe it is more complex.

Why Does This Matter?

It breaks the rules: In many physics problems, if you have friction, things stop fast. This paper shows that on an infinite domain with this specific type of friction, things don't stop fast. They stop slowly.
It's the best we can do: They didn't just find a speed; they proved you can't find a faster speed for this system. It's the "speed limit" of decay for this equation.
Real-world applications: While this is a math paper, the principles apply to engineering problems involving long structures (like bridges or pipelines) where materials have internal friction. Knowing that vibrations might linger longer than expected (slower decay) is crucial for safety and design.

Summary in One Sentence

The authors proved that an infinite vibrating string with "sticky" friction will eventually stop, but because of two stubborn mathematical "echoes," it will fade away slowly (at a rate of $1/t^2$ ) rather than vanishing instantly, and they found the exact conditions required to predict this slow fade.

1. Problem Statement

The paper investigates the long-time asymptotic behavior of solutions to the one-dimensional linear Klein–Gordon equation with Kelvin–Voigt damping posed on the whole real line ( $\mathbb{R}$ ). The equation is given by:
$u_{tt} - u_{xx} + mu - u_{txx} = 0, \quad x \in \mathbb{R}, t > 0$
with initial data $u(x,0) = u_0(x) \in H^1(\mathbb{R})$ and $u_t(x,0) = v_0(x) \in L^2(\mathbb{R})$ , where $m > 0$ is a fixed constant.

The energy of the system is defined as:
$E_{z_0}(t) = \frac{1}{2} \int_{\mathbb{R}} \left( m|u|^2 + |u_x|^2 + |u_t|^2 \right) dx$
It is known that the energy is non-increasing due to the dissipative nature of the Kelvin–Voigt term ( $-u_{txx}$ ). The central challenge addressed is determining whether the energy decays to zero and, if so, establishing the optimal rate of decay.

Contextual Contrast:

Bounded Domains: On bounded intervals with Dirichlet conditions, such systems are known to be exponentially stable (energy decays as $e^{-\gamma t}$ ) due to the Poincaré inequality.
Unbounded Domains: On $\mathbb{R}$ , the Poincaré inequality fails. The authors demonstrate that the system is not exponentially stable, nor even semi-uniformly stable, necessitating a different approach to quantify decay.

2. Methodology

The authors employ a functional analytic approach based on $C_0$ -semigroup theory and spectral analysis.

A. Abstract Formulation

The PDE is reformulated as an abstract Cauchy problem $\dot{z}(t) = Az(t)$ on the Hilbert space $X = H^1(\mathbb{R}) \times L^2(\mathbb{R})$ .

The generator $A$ is defined as $A(u, v) = (v, (u+v)'' - mu)$.
The domain $D(A)$ requires $u+v \in H^2(\mathbb{R})$ .
The operator $A$ is shown to be closed, dissipative, and densely defined, generating a contraction $C_0$ -semigroup $(T(t))_{t \geq 0}$ .

B. Spectral Analysis

A critical step is characterizing the spectrum $\sigma(A)$ of the generator:

Imaginary Axis Spectrum: The authors prove that the intersection of the spectrum with the imaginary axis consists of exactly two points: $\{i\sqrt{m}, -i\sqrt{m}\}$ .
No Eigenvalues: Crucially, these points are not eigenvalues ( $\sigma_p(A) \cap i\mathbb{R} = \emptyset$ ); they are approximate eigenvalues.
Implication: Since the spectrum touches the imaginary axis at non-zero points, the semigroup cannot be exponentially stable. However, the absence of eigenvalues on the imaginary axis allows the application of the Arendt–Batty–Lyubich–Vũ theorem, which guarantees strong asymptotic stability (energy $\to 0$ ).

C. Resolvent Estimates

To quantify the decay rate, the authors derive precise bounds for the resolvent operator $R(is, A) = (is - A)^{-1}$ along the imaginary axis:

Near Spectral Points: As $s \to \pm\sqrt{m}$ , $\|R(is, A)\| = O(|s \mp \sqrt{m}|^{-1})$ .
At Infinity: As $|s| \to \infty$ , $\|R(is, A)\| = O(1)$ .

These estimates are derived using explicit Green's function representations and convolution estimates involving the parameter $\lambda_s = \frac{m-s^2}{1+is}$ .

D. Decay Rate Derivation

The authors combine the resolvent estimates with recent abstract results on decay rates (specifically from [4] and [30]).

They define operators $B_1 = (i\sqrt{m}-A)R(1, A)$ and $B_2 = (-i\sqrt{m}-A)R(1, A)$ .
They show that for initial data $z_0$ in the range of the product $B_1 B_2$ (which corresponds to the intersection of the ranges of $i\sqrt{m}-A$ and $-i\sqrt{m}-A$ ), the resolvent satisfies specific boundedness conditions near the spectral points.
This allows the application of a theorem linking resolvent growth to polynomial decay rates.

3. Key Contributions and Results

Main Theorem (Theorem 4.1)

Asymptotic Stability: For all initial data $z_0 \in X$ , the energy decays to zero:
$\lim_{t \to \infty} E_{z_0}(t) = 0$
Optimal Polynomial Decay: For a specific class of "smoother" initial data, specifically those in the intersection of the ranges:
$z_0 \in \text{Ran}(i\sqrt{m} - A) \cap \text{Ran}(-i\sqrt{m} - A)$
the energy decays at the optimal rate:
$E_{z_0}(t) = O(t^{-2}) \quad \text{as } t \to \infty$
Optimality: The rate $t^{-2}$ is sharp. If one assumes a decay rate $r(t) = o(t^{-2})$ , there exists an initial condition in the specified class for which the energy does not decay that fast.

Characterization of Initial Data

The paper provides a concrete characterization (Proposition 4.2) of the initial data $(u_0, v_0)$ that yield the $t^{-2}$ decay. These conditions involve the $L^2$ -integrability of specific iterated integrals of the linear combinations $i\sqrt{m}u_0 + v_0$ and $-i\sqrt{m}u_0 + v_0$ . Essentially, the initial data must possess sufficient regularity and decay properties to "cancel out" the singularities associated with the spectral points $\pm i\sqrt{m}$ .

Extension to Half-Line

The authors briefly discuss the case on the half-line ( $x > 0$ ) with Dirichlet boundary conditions ( $u(0,t)=0$ ). They demonstrate that their techniques adapt seamlessly, preserving the same spectral structure and optimal decay rates, with minor modifications to the Green's function to satisfy the boundary condition.

4. Significance

Resolution of Unbounded Domain Behavior: This work clarifies the long-standing gap in understanding Kelvin–Voigt damping on unbounded domains. It proves that while exponential stability is lost due to the lack of a spectral gap at zero, polynomial stability is achievable.
Optimality: Unlike previous works that might provide sub-optimal rates or rely on general Banach space criteria, this paper provides the optimal polynomial rate ( $t^{-2}$ ) for the Hilbert space setting.
Spectral Insight: The paper highlights the critical role of the non-eigenvalue nature of the imaginary spectral points. If these were eigenvalues, the system would not be asymptotically stable at all. The fact that they are merely in the continuous spectrum allows for decay, but the proximity of the spectrum to the imaginary axis dictates the polynomial rate.
Methodological Advancement: The combination of specific resolvent estimates near multiple spectral points with abstract decay theorems offers a robust framework for analyzing other PDEs with similar spectral configurations on unbounded domains.

In summary, the paper establishes that while the Kelvin–Voigt damped Klein–Gordon equation on $\mathbb{R}$ does not exhibit exponential decay, it is asymptotically stable with an optimal energy decay rate of $O(t^{-2})$ for sufficiently regular initial data.