pith. machine review for the scientific record. sign in

arxiv: 2604.19938 · v1 · submitted 2026-04-21 · 🧮 math.SP

Recognition: unknown

The Evans function as a lower bound on the spectral distance function

George Bayliss, Jared C. Bronski

Pith reviewed 2026-05-10 00:35 UTC · model grok-4.3

classification 🧮 math.SP
keywords Evans functionspectral distanceresolvent setboundary value problemsstability analysisnumerical spectrum computation
0
0 comments X

The pith

A normalized Evans function bounds the distance from any point to the spectrum of a boundary-value problem

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows how to normalize the Evans function so that its magnitude at a point in the resolvent set lower-bounds the distance from that point to the nearest point in the spectrum. Consequently, if the normalized Evans function is nonzero at λ*, an entire disk of radius |E(λ*)| centered at λ* lies in the resolvent set. This matters because the Evans function is already used to locate eigenvalues of operators arising in stability problems; the new normalization turns it into a tool that can also certify whole regions free of spectrum. The authors give explicit normalizations for standard boundary conditions on a compact interval and test the bound numerically on second- and fourth-order self-adjoint operators as well as a linearized modified Korteweg–de Vries equation.

Core claim

With a natural normalization on compact domains, the magnitude of the Evans function E(λ) at a point λ in the resolvent set is a lower bound on the distance from λ to the spectrum of the associated boundary-value problem; hence the disk of radius |E(λ)| centered at λ lies entirely in the resolvent set.

What carries the argument

The Evans function built from the matching condition of left- and right-traveling fundamental solutions that satisfy the boundary conditions of the ODE on a compact interval; its analytic properties under this normalization translate directly into a distance-to-spectrum bound.

If this is right

  • Non-vanishing of the normalized E at λ* certifies that the entire disk |λ − λ*| < |E(λ*)| contains no spectrum.
  • The same function now supplies both the location of eigenvalues (its zeros) and quantitative spectral-gap certificates (its magnitude).
  • The bound applies to standard boundary conditions on compact intervals and is demonstrated for second- and fourth-order self-adjoint operators and linearized mKdV.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The magnitude could be used as a cheap distance estimator to steer adaptive numerical searches away from already-cleared regions of the complex plane.
  • Similar distance bounds might be obtainable on unbounded domains once decay estimates at infinity are incorporated into the normalization.
  • The construction may extend to other analytic functions that encode spectral data, such as characteristic polynomials or Fredholm determinants.

Load-bearing premise

The normalization is chosen so that the analytic properties coming from the fundamental-solution construction on a finite interval directly yield the distance bound.

What would settle it

For the Dirichlet Laplacian on [0, π], evaluate the normalized Evans function at a point such as λ = 0.1 and check whether the distance to the nearest eigenvalue π² is at least |E(0.1)|.

Figures

Figures reproduced from arXiv: 2604.19938 by George Bayliss, Jared C. Bronski.

Figure 1
Figure 1. Figure 1: Evans function on [2, 5] [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Evans function on [2, 5] with weighted Evans function calculated at λ = 3. One can easily see that by computing the weighted Evans function, one finds an eigenvalue free region around the specified value of λ. We can expand this region by calculating the weighted Evans function at the bounds of this region: 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.6 0.4 0.2 0.0 0.2 0.4 0.6 E( ) [PITH_FULL_IMAGE:figures/full_fig_p013… view at source ↗
Figure 3
Figure 3. Figure 3: Evans function on [2, 5] with weighted Evans function calculated to two iterations at λ = 3 [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Evans function on [2.5, 17.5]. Now we calculate the weighted Evans function about λ = 11 to three iterations [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Evans function on [2.5, 17.5] with weighted Evans function calculated to three iterations at λ = 11. Since the proof of 2.4 assumes a second order operator it does not extend to this case. However, we believe that the theorem may still apply to higher order cases. In figure 6 we plot the weighted Evans function near the eigenvalue at ≈ 5. 4.00 4.25 4.50 4.75 5.00 5.25 5.50 5.75 6.00 0.0 0.2 0.4 0.6 0.8 1.0… view at source ↗
Figure 6
Figure 6. Figure 6: The weighted Evans function plotted at intervals of .1 against spectral distance. 4.3. The linearized mKdV equation. Now we use this method to study the stabil￾ity of traveling waves to the generalized KdV equation. For the mKdV equation, the [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Spectrum of L in [−0.6 − 2.5i, 0.6 + 2.5i]. linearized stability of the standing waves is governed by the spectrum of the third order non-self-adjoint operator Lu = −∂xxxu + ∂x(−3ϕ 2u) with ϕ an elliptic function of x. For our example, we choose the Jacobi elliptic function cn(x; m = 1 2 ) for the parameter value m = 1/2. We can calculate the spectrum via the Floquet-Fourier-Hill method[13] in figure 7. Si… view at source ↗
Figure 8
Figure 8. Figure 8: Spectrum of L in [−0.6 − 2.5i, 0.6 + 2.5i] with black dots representing spectrum of L(µ) and blue circle representing the weighted Evans function at λ = 0.1 + 0.5i. 0.4 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 0.5 0.6 |E( )| [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Black line represents spectral distance from λ = 0.1 + 0.5i for each value of µ and grey rectangles represent computations of the weighted Evans function for L(µ) at λ = 0.1 + 0.5i. 5. Conclusions In this paper we have considered the role of the normalization for the Evans function defined by an ODE eigenvalue problem on a compact interval. For a number of prob￾lems of the type that arise in practice there… view at source ↗
Figure 10
Figure 10. Figure 10: Spectrum of L in [−0.6 − 2.5i, 0.6 + 2.5i] with an eigenvalue free region centered at λ = 0.1 + 0.5i. is both theoretically simple and numerically easily computable that provides a natural geometric interpretation to the magnitude of the Evans function. In particular the mag￾nitude of the normalized Evans function guarantees a a ball in the spectral or parameter plane that is free of eigenvalues. While th… view at source ↗
read the original abstract

The Evans function is an analytic function that encodes information about the intersection of certain subspaces in ODE boundary value problems. As such it is a useful tool for computing the spectrum of boundary value problems arising in the stability of coherent structures. In typical applications one is interested in the roots of the Evans function, but the overall normalization is somewhat arbitrary. We present a natural normalization of the Evans function on compact domains such that the magnitude of the Evans function provides a lower bound on the distance to the nearest point in the spectrum. In other words the magnitude of the Evans function at a point in the resolvent set implies that a ball about the point in question lies in the resolvent set. Thus, when appropriately normalized, not only does the Evans function $E(\lambda)$ vanish if and only if $\lambda$ lies in the spectrum of the operator in question, but a non-zero value for the Evans function guarantees that a disk of radius $|E(\lambda^*)|$ about the point $\lambda^*$ lies in the resolvent set. We present some calculations for some common sets of boundary conditions on a compact interval, and present some numerical experiments for 2nd and 4th order self-adjoint operators and for a linearized modified Korteweg-De Vries equation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a natural normalization of the Evans function E(λ) for boundary-value problems arising from linear ODEs on compact intervals. With this normalization, |E(λ)| at any point λ in the resolvent set is asserted to lower-bound the distance from λ to the spectrum of the operator. Consequently, the open disk of radius |E(λ*)| centered at any resolvent point λ* is guaranteed to contain no spectrum. The authors illustrate the normalization with explicit calculations for standard boundary conditions and report numerical experiments on second- and fourth-order self-adjoint operators together with a linearized mKdV equation.

Significance. If the central claim were valid, the normalized Evans function would simultaneously locate eigenvalues via its zeros and certify spectral-free regions via its magnitude, providing a computable, rigorous a-posteriori bound useful for numerical stability analysis of coherent structures.

major comments (2)
  1. [Abstract] Abstract: The asserted normalization cannot satisfy the claimed bound. The Evans function constructed from fundamental solutions on a compact interval is entire and has infinitely many zeros (the discrete spectrum). No entire function with infinitely many zeros can obey |E(λ)| ≤ dist(λ, spec) for all λ in the resolvent set. This follows from the growth of entire functions: for any non-linear entire function the modulus grows faster than the distance to its zero set in the complex plane (e.g., sin(πz) violates the inequality at half-integers with large imaginary part). Multiplication by a nowhere-vanishing entire function (the only operation that preserves both analyticity and the zero set) leaves the contradiction intact. The claim is therefore false for all examples treated in the paper.
  2. [Abstract] Abstract (numerical experiments paragraph): The reported numerical experiments for the 2nd- and 4th-order self-adjoint operators and the mKdV linearization supply no tabulated values, error bars, comparison against independently computed spectra, or explicit checks that observed |E(λ)| respects the distance bound. Without these data the experiments cannot corroborate the central claim.
minor comments (2)
  1. The precise definition of the proposed normalization (including any λ-dependent scaling factors) is not stated explicitly enough to allow independent verification or reproduction.
  2. The manuscript would benefit from additional references to existing literature on Evans-function normalizations and on a-posteriori spectral bounds.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and for identifying a fundamental inconsistency in the central claim of the manuscript. We agree that the asserted bound cannot hold, and we will revise the paper to remove this claim while retaining the construction of the normalized Evans function. Our point-by-point responses to the major comments follow.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The asserted normalization cannot satisfy the claimed bound. The Evans function constructed from fundamental solutions on a compact interval is entire and has infinitely many zeros (the discrete spectrum). No entire function with infinitely many zeros can obey |E(λ)| ≤ dist(λ, spec) for all λ in the resolvent set. This follows from the growth of entire functions: for any non-linear entire function the modulus grows faster than the distance to its zero set in the complex plane (e.g., sin(πz) violates the inequality at half-integers with large imaginary part). Multiplication by a nowhere-vanishing entire function (the only operation that preserves both analyticity and the zero set) leaves the contradiction intact. The claim is therefore false for all examples treated in the paper.

    Authors: We agree with the referee's analysis. The Evans function, being entire with infinitely many zeros, cannot satisfy |E(λ)| ≤ dist(λ, spec) everywhere in the resolvent set, as demonstrated by standard growth estimates for entire functions. This invalidates the central claim of the paper. We will revise the manuscript to eliminate the assertion that the magnitude provides a lower bound on the spectral distance for all resolvent points. The revised version will instead present the natural normalization and its role in identifying eigenvalues through zeros, without the distance-bound claim. revision: yes

  2. Referee: [Abstract] Abstract (numerical experiments paragraph): The reported numerical experiments for the 2nd- and 4th-order self-adjoint operators and the mKdV linearization supply no tabulated values, error bars, comparison against independently computed spectra, or explicit checks that observed |E(λ)| respects the distance bound. Without these data the experiments cannot corroborate the central claim.

    Authors: We acknowledge that the numerical experiments lack the necessary detail, such as tabulated values, error estimates, and direct comparisons to independently computed spectra, to support verification. Since the theoretical bound does not hold, these experiments cannot corroborate the original claim. In the revision we will remove or substantially rewrite the numerical section, focusing instead on explicit calculations for standard boundary conditions as described in the manuscript body. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The paper constructs a specific normalization of the Evans function from the fundamental solutions and boundary conditions on the compact interval. The lower-bound claim is asserted to follow directly from the resulting analytic properties of this normalized E(λ), without any reduction of the central statement to a fitted parameter, self-definition, or load-bearing self-citation. No equation or step is shown to be equivalent to its inputs by construction, and the numerical experiments on specific operators are presented as verification rather than as the source of the general bound.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The result rests on standard analytic properties of the Evans function for linear ODEs on compact intervals together with the existence of a natural normalization that preserves the zero set while controlling the magnitude; no new free parameters or invented entities are introduced.

axioms (2)
  • domain assumption The Evans function is analytic in the resolvent set and its zeros coincide exactly with the spectrum of the boundary-value problem.
    Invoked in the abstract as the foundational encoding property of the Evans function.
  • ad hoc to paper A natural normalization exists on compact domains that makes |E(λ)| a lower bound on the distance to the spectrum.
    This is the load-bearing construction presented by the authors; its validity is the central claim.

pith-pipeline@v0.9.0 · 5523 in / 1544 out tokens · 26005 ms · 2026-05-10T00:35:06.406990+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

32 extracted references

  1. [1]

    Numerical implementation of complex orthogonaliza- tion, parallel transport on Stiefel bundles, and analyticity.Physica D: Nonlinear Phenomena, 239(12):1038–1047, 2010

    Daniele Avitabile and Thomas J Bridges. Numerical implementation of complex orthogonaliza- tion, parallel transport on Stiefel bundles, and analyticity.Physica D: Nonlinear Phenomena, 239(12):1038–1047, 2010

    2010

  2. [2]

    Asymptotic stability of sharp fronts: Analysis and rigorous computation.Journal of Differential Equations, 444:113550, 2025

    Blake Barker, Jared C Bronski, Vera Mikyoung Hur, and Zhao Yang. Asymptotic stability of sharp fronts: Analysis and rigorous computation.Journal of Differential Equations, 444:113550, 2025

    2025

  3. [3]

    Blake Barker, Jeffrey Humpherys, Gregory Lyng, and Joshua Lytle. Evans function computation for the stability of travelling waves.Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 376(2117):20170184, 2018

    2018

  4. [4]

    Computing Evans functions numerically via boundary-value problems.Physica D: Nonlinear Phenomena, 367:1–10, 2018

    Blake Barker, Rose Nguyen, Bj¨ orn Sandstede, Nathaniel Ventura, and Colin Wahl. Computing Evans functions numerically via boundary-value problems.Physica D: Nonlinear Phenomena, 367:1–10, 2018

    2018

  5. [5]

    Brigham Young University, 2009

    Blake H Barker.Evans function computation. Brigham Young University, 2009

    2009

  6. [6]

    Validated spectral stability via conjugate points.SIAM Journal on Applied Dynamical Systems, 21(1):366–404, 2022

    Margaret Beck and Jonathan Jaquette. Validated spectral stability via conjugate points.SIAM Journal on Applied Dynamical Systems, 21(1):366–404, 2022

    2022

  7. [7]

    Electrical waves in a one-dimensional model of cardiac tissue.SIAM Journal on Applied Dynamical Systems, 7(4):1558–1581, 2008

    Margaret Beck, Christopher KRT Jones, David Schaeffer, and Martin Wechselberger. Electrical waves in a one-dimensional model of cardiac tissue.SIAM Journal on Applied Dynamical Systems, 7(4):1558–1581, 2008

    2008

  8. [8]

    Planar traveling waves in capillary fluids.Differential and integral equations, 26(3-4):433–478, 2013

    Sylvie Benzoni-Gavage. Planar traveling waves in capillary fluids.Differential and integral equations, 26(3-4):433–478, 2013

    2013

  9. [9]

    Transversality of homoclinic orbits, the Maslov index and the symplectic Evans function.Nonlinearity, 28(1):77, 2014

    Fr´ ed´ eric Chardard and Thomas J Bridges. Transversality of homoclinic orbits, the Maslov index and the symplectic Evans function.Nonlinearity, 28(1):77, 2014

    2014

  10. [10]

    Stability of shock wave structures in nonlinear elastic media.Mathematics and Mechanics of Solids, 24(11):3456–3471, 2019

    AP Chugainova, AT Il’ichev, and VA Shargatov. Stability of shock wave structures in nonlinear elastic media.Mathematics and Mechanics of Solids, 24(11):3456–3471, 2019

    2019

  11. [11]

    Evans functions for integral neural field equations with Heaviside firing rate function.SIAM Journal on Applied Dynamical Systems, 3(4):574–600, 2004

    Stephen Coombes and Markus R Owen. Evans functions for integral neural field equations with Heaviside firing rate function.SIAM Journal on Applied Dynamical Systems, 3(4):574–600, 2004

    2004

  12. [12]

    Fredholm determinants, Evans functions and Maslov indices for partial differential equations.Mathematische Annalen, 389(2):1963–2036, 2024

    Graham Cox, Yuri Latushkin, and Alim Sukhtayev. Fredholm determinants, Evans functions and Maslov indices for partial differential equations.Mathematische Annalen, 389(2):1963–2036, 2024

    1963

  13. [13]

    Computing spectra of linear operators using the floquet– fourier–hill method.Journal of Computational Physics, 219(1):296–321, 2006

    Bernard Deconinck and J Nathan Kutz. Computing spectra of linear operators using the floquet– fourier–hill method.Journal of Computational Physics, 219(1):296–321, 2006

    2006

  14. [14]

    Transverse instabilities of deep-water solitary waves.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 462(2071):2039–2061, 2006

    Bernard Deconinck, Dmitry E Pelinovsky, and John D Carter. Transverse instabilities of deep-water solitary waves.Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 462(2071):2039–2061, 2006

    2071

  15. [15]

    Nerve axon equations: 1 linear approximations.Indiana University Mathematics Journal, 21(9):877–885, 1972

    John W Evans. Nerve axon equations: 1 linear approximations.Indiana University Mathematics Journal, 21(9):877–885, 1972. THE EV ANS FUNCTION AND SPECTRAL DISTANCE 21

    1972

  16. [16]

    Nerve axon equations: Ii stability at rest.Indiana University Mathematics Journal, 22(1):75–90, 1972

    John W Evans. Nerve axon equations: Ii stability at rest.Indiana University Mathematics Journal, 22(1):75–90, 1972

    1972

  17. [17]

    Nerve axon equations: Iii stability of the nerve impulse.Indiana University Mathe- matics Journal, 22(6):577–593, 1972

    John W Evans. Nerve axon equations: Iii stability of the nerve impulse.Indiana University Mathe- matics Journal, 22(6):577–593, 1972

    1972

  18. [18]

    Nerve axon equations: Iv the stable and the unstable impulse.Indiana University Mathematics Journal, 24(12):1169–1190, 1975

    John W Evans. Nerve axon equations: Iv the stable and the unstable impulse.Indiana University Mathematics Journal, 24(12):1169–1190, 1975

    1975

  19. [19]

    PhD thesis, Boston University, 2010

    Matthew Denson Holzer.Renormalization group methods for singularly perturbed systems, normal forms and stability of traveling waves in a reaction-diffusion-mechanics system. PhD thesis, Boston University, 2010

    2010

  20. [20]

    An efficient shooting algorithm for Evans function calcu- lations in large systems.Physica D: Nonlinear Phenomena, 220(2):116–126, 2006

    Jeffrey Humpherys and Kevin Zumbrun. An efficient shooting algorithm for Evans function calcu- lations in large systems.Physica D: Nonlinear Phenomena, 220(2):116–126, 2006

    2006

  21. [21]

    Modulational instability in the Whitham equation for water waves.Studies in Applied Mathematics, 134(1):120–143, 2015

    Vera Mikyoung Hur and Mathew A Johnson. Modulational instability in the Whitham equation for water waves.Studies in Applied Mathematics, 134(1):120–143, 2015

    2015

  22. [22]

    Unstable Stokes waves.Archive for Rational Mechanics and Analysis, 247(4):62, 2023

    Vera Mikyoung Hur and Zhao Yang. Unstable Stokes waves.Archive for Rational Mechanics and Analysis, 247(4):62, 2023

    2023

  23. [23]

    Instability of solitary waves on Euler’s elastica.Zeitschrift f¨ ur angewandte Math- ematik und Physik ZAMP, 57(4):547–566, 2006

    Andrej T Il’ichev. Instability of solitary waves on Euler’s elastica.Zeitschrift f¨ ur angewandte Math- ematik und Physik ZAMP, 57(4):547–566, 2006

    2006

  24. [24]

    The Evans function for nonlocal equations

    Todd Kapitula, Nathan Kutz, and Bj¨ orn Sandstede. The Evans function for nonlocal equations. Indiana University mathematics journal, pages 1095–1126, 2004

    2004

  25. [25]

    Instability of local deformations of an elastic rod: numerical evaluation of the Evans function.SIAM Journal on Applied Mathematics, 71(5):1653– 1672, 2011

    St´ ephane Lafortune, Joceline Lega, and S Madrid. Instability of local deformations of an elastic rod: numerical evaluation of the Evans function.SIAM Journal on Applied Mathematics, 71(5):1653– 1672, 2011

    2011

  26. [26]

    Computing stability of mul- tidimensional traveling waves.SIAM Journal on Applied Dynamical Systems, 8(1):480–507, 2009

    Veerle Ledoux, Simon JA Malham, Jitse Niesen, and Vera Th¨ ummler. Computing stability of mul- tidimensional traveling waves.SIAM Journal on Applied Dynamical Systems, 8(1):480–507, 2009

    2009

  27. [27]

    Traveling waves in a three-variable reaction-diffusion-mechanics model of cardiac tissue.Communications on Pure and Applied Analysis, 24(5):892–925, 2025

    Ji Li and Qing Yu. Traveling waves in a three-variable reaction-diffusion-mechanics model of cardiac tissue.Communications on Pure and Applied Analysis, 24(5):892–925, 2025

    2025

  28. [28]

    YA Li. The eigenvalue problem for solitary waves of the Green–Naghdi equations.Mathematical and Numerical Aspects of Wave Propagation (Santiago de Compostela, 2000), SIAM, Philadelphia, PA, pages 339–343, 2000

    2000

  29. [29]

    Evaluating the Evans function: order reduction in numerical methods.Mathematics of computation, 77(261):159–179, 2008

    Simon Malham and Jitse Niesen. Evaluating the Evans function: order reduction in numerical methods.Mathematics of computation, 77(261):159–179, 2008

    2008

  30. [30]

    Stability of periodic solutions of conservation laws with vis- cosity: Analysis of the Evans function.Archive for Rational Mechanics and Analysis, 166:99–166, 2003

    Myunghyun Oh and Kevin Zumbrun. Stability of periodic solutions of conservation laws with vis- cosity: Analysis of the Evans function.Archive for Rational Mechanics and Analysis, 166:99–166, 2003

    2003

  31. [31]

    Stability of travelling waves

    Bj¨ orn Sandstede. Stability of travelling waves. InHandbook of dynamical systems, volume 2, pages 983–1055. Elsevier, 2002

    2002

  32. [32]

    The stability of homoclinic pulses: a generalisation of Evans’s method.Physics Letters A, 163(1):57–62, 1992

    Jonathan Swinton. The stability of homoclinic pulses: a generalisation of Evans’s method.Physics Letters A, 163(1):57–62, 1992. University of Illinois, Urbana, IL 61801, USA Email address:bayliss2@illinois.edu Department of Mathematics, University of Illinois, Urbana, IL 61801, USA Email address:bronski@illinois.edu

    1992