pith. sign in

arxiv: 2605.17464 · v2 · pith:GHRRACOJnew · submitted 2026-05-17 · 🧮 math.OC

Fully Discrete High-Order DG Schemes for Waves: Dispersion and Observability

Yunzhang Li , Xiaoyang Wang , Enrique Zuazua This is my paper

Pith reviewed 2026-05-21 08:16 UTC · model grok-4.3

classification 🧮 math.OC
keywords discontinuous Galerkinwave equationnumerical dispersionobservabilityspectral filteringhigh-order methodsfully discrete schemes
0
0 comments X

The pith

High-order DG schemes for the wave equation create a trapping mechanism from vanishing group velocities that causes exponential blow-up in the observability constant.

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

This paper examines fully discrete P^k discontinuous Galerkin approximations to the one-dimensional wave equation and characterizes their coupled space-time numerical dispersion. The analysis shows that this dispersion forces group velocities of both physical and spurious modes to vanish at selected frequencies, creating a trapping effect. As a result the observability constant grows exponentially like exp(h^{-(1-ε)}). The authors introduce a spectral filtering strategy that restores uniform observability for any polynomial degree k, and note that higher-order schemes preserve a wider band of genuine physical frequencies, which reduces the filtering effort and shortens the required observation time.

Core claim

Characterizing the coupled space-time numerical dispersion relation in these fully discrete DG schemes reveals a trapping mechanism that drives group velocities to zero for both physical and spurious modes at certain frequencies; this produces an observability constant that blows up exponentially as exp(h^{-(1-ε)}), and spectral filtering overcomes the divergence while higher orders lower the associated costs by keeping more physical frequencies intact.

What carries the argument

The trapping mechanism identified through the coupled space-time numerical dispersion relation, which forces vanishing group velocities at selected frequencies for physical and spurious modes.

If this is right

  • The observability constant blows up exponentially of order exp(h^{-(1-ε)}) under the trapping mechanism.
  • Spectral filtering restores uniform observability for arbitrary polynomial degree k.
  • Higher-order methods preserve a larger genuine physical frequency band, thereby reducing filtering cost and observation time.

Where Pith is reading between the lines

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

  • Combining high-order DG with targeted spectral filtering may enable stable long-time wave simulations with lower overall computational cost.
  • The same dispersion-driven trapping could appear in other hyperbolic systems or higher-dimensional settings, suggesting the filtering approach has wider use.
  • The reduction in filtering cost with increasing k could make these schemes more practical for applications such as acoustic or seismic modeling.

Load-bearing premise

The analysis assumes that the coupled space-time numerical dispersion relation produces vanishing group velocities for both physical and spurious modes at selected frequencies, which is the root cause of the trapping mechanism and subsequent observability blow-up.

What would settle it

Numerically compute the observability constant for a sequence of decreasing mesh sizes h and verify whether its growth matches the predicted exp(h^{-(1-ε)}) rate, or extract group velocities directly from the dispersion relation and confirm they reach zero at the frequencies indicated by the analysis.

Figures

Figures reproduced from arXiv: 2605.17464 by Enrique Zuazua, Xiaoyang Wang, Yunzhang Li.

Figure 1
Figure 1. Figure 1: Dispersion relations for k = 0, 1, 2, with λ = 0.8, 0.3, 0.12, respectively [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: compares the dispersion relations slightly below and exactly at the strict stability boundaries, corroborating Theorem 3.3 [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Recovery of uniform observability via modal-frequency filtering. (a) Compar [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
read the original abstract

This paper investigates the spectral structure, numerical dispersion, and observability of fully discrete approximations of the one-dimensional wave equation by $P^k$ (local) discontinuous Galerkin methods. Characterizing the coupled space-time numerical dispersion reveals a trapping mechanism that forces the group velocities of both physical and spurious modes to vanish at selected frequencies. We then establish an exponential blow-up of order $\exp(h^{-(1-\varepsilon)})$ for the observability constant under this trapping mechanism. To overcome this divergence for arbitrary $k$, we propose a spectral filtering strategy to restore uniform observability. Theoretical analysis and numerical experiments indicate that higher-order methods may facilitate this recovery by preserving a larger genuine physical frequency band, thereby reducing filtering cost and observation time.

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

1 major / 1 minor

Summary. The paper analyzes the spectral structure, numerical dispersion, and observability properties of fully discrete P^k discontinuous Galerkin approximations to the one-dimensional wave equation. It characterizes the coupled space-time dispersion relation, identifies a trapping mechanism that forces vanishing group velocities for both physical and spurious modes at selected frequencies, derives an exponential blow-up of the observability constant of order exp(h^{-(1-ε)}), and proposes a spectral filtering strategy to restore uniform observability. Higher-order methods are suggested to reduce filtering cost by preserving a larger physical frequency band. The claims are supported by theoretical analysis and numerical experiments.

Significance. If the central results hold, the work provides a precise quantification of observability degradation in fully discrete high-order DG schemes for waves, which is relevant to numerical control and inverse problems. The exponential blow-up rate and the proposed filtering recovery constitute a concrete contribution, particularly if the analysis is uniform in the polynomial degree k. The observation that higher-order schemes may lower the filtering overhead is practically useful and could guide method selection in applications.

major comments (1)
  1. [Dispersion relation and trapping mechanism] The exponential blow-up claim of order exp(h^{-(1-ε)}) for the observability constant rests on the exact vanishing of group velocities in the coupled space-time dispersion relation at selected frequencies for arbitrary k. The analysis must explicitly demonstrate that the relevant roots of this dispersion relation are independent of the mesh size h (or shift only in a controlled way) uniformly across the discrete spectrum; without such a uniformity statement the rate and the necessity of the spectral filter become conditional. Please supply the algebraic or asymptotic characterization of these roots in the section on spectral structure.
minor comments (1)
  1. [Abstract and main results] Ensure that all statements about the order of the blow-up and the filtering cost reduction are cross-referenced to the precise theorem or proposition that establishes them.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the thorough review and constructive comments, which help strengthen the presentation of our results on the spectral properties and observability of fully discrete DG schemes. We address the major comment below and will incorporate the requested clarifications in the revised manuscript.

read point-by-point responses

  1. Referee: [Dispersion relation and trapping mechanism] The exponential blow-up claim of order exp(h^{-(1-ε)}) for the observability constant rests on the exact vanishing of group velocities in the coupled space-time dispersion relation at selected frequencies for arbitrary k. The analysis must explicitly demonstrate that the relevant roots of this dispersion relation are independent of the mesh size h (or shift only in a controlled way) uniformly across the discrete spectrum; without such a uniformity statement the rate and the necessity of the spectral filter become conditional. Please supply the algebraic or asymptotic characterization of these roots in the section on spectral structure.

    Authors: We agree that an explicit uniformity statement strengthens the argument. In Section 3 we derive the coupled space-time dispersion relation by substituting a plane-wave ansatz into the fully discrete DG scheme, yielding a determinant condition on a matrix pencil whose entries depend on the scaled variables Ω = ω Δt and Ξ = ξ h together with the CFL ratio and the reference-element DG matrices. After this nondimensionalization the equation is manifestly independent of h; the critical frequencies at which the group velocity vanishes are therefore fixed roots of an h-independent algebraic (or transcendental) equation in the scaled spectrum. These roots are uniform across the discrete spectrum because the periodicity of the dispersion relation in Ξ is preserved for any fixed k. We will add a short subsection (or expanded paragraph) in the spectral-structure section that states this scaling explicitly, supplies the algebraic form of the characteristic equation for general k, and includes a brief asymptotic description of the root locations for large k. This addition will make the h-independence and the resulting exponential blow-up rate fully rigorous and will also support the claim that higher-order schemes reduce filtering cost. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained spectral analysis

full rationale

The paper derives the trapping mechanism and exponential observability blow-up directly from characterizing the coupled space-time numerical dispersion relation of the fully discrete P^k DG scheme, which reveals vanishing group velocities for physical and spurious modes at selected frequencies. This leads to the exp(h^{-(1-ε)}) blow-up estimate and the subsequent proposal of spectral filtering. No steps reduce by construction to fitted inputs, self-definitions, or load-bearing self-citations; the analysis rests on the discrete operator's spectral properties without renaming known results or smuggling ansatzes via prior work. The derivation chain is independent and self-contained against the stated assumptions on the dispersion relation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The analysis rests on standard properties of DG spatial discretization and time-stepping schemes for the wave equation. No free parameters or invented entities are introduced in the abstract; the trapping is derived from the dispersion relation itself.

axioms (1)
  • domain assumption The coupled space-time dispersion relation for the fully discrete scheme admits frequencies where group velocities of physical and spurious modes vanish.
    This is the key premise invoked to establish the trapping mechanism and the subsequent exponential blow-up of the observability constant.

pith-pipeline@v0.9.0 · 5651 in / 1383 out tokens · 32556 ms · 2026-05-21T08:16:28.820791+00:00 · methodology

Review history (2 revisions) →

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    1995 , PAGES =

    Kato, Tosio , TITLE =. 1995 , PAGES =

    work page 1995

  2. [2]

    2014 , PAGES =

    Marica, Aurora and Zuazua, Enrique , TITLE =. 2014 , PAGES =

    work page 2014

  3. [3]

    SIAM Rev

    Zuazua, Enrique , TITLE =. SIAM Rev. , FJOURNAL =. 2005 , NUMBER =

    work page 2005

  4. [4]

    M2AN Math

    Infante, Juan Antonio and Zuazua, Enrique , TITLE =. M2AN Math. Model. Numer. Anal. , FJOURNAL =. 1999 , NUMBER =

    work page 1999

  5. [5]

    Micu, Sorin , TITLE =. Numer. Math. , FJOURNAL =. 2002 , NUMBER =

    work page 2002

  6. [6]

    , TITLE =

    Zuazua, E. , TITLE =. J. Math. Pures Appl. (9) , FJOURNAL =. 1991 , NUMBER =

    work page 1991

  7. [7]

    2011 , PAGES =

    Bahouri, Hajer and Chemin, Jean-Yves and Danchin, Rapha\"el , TITLE =. 2011 , PAGES =

    work page 2011

  8. [8]

    Systems Control Lett

    Negreanu, Mihaela and Zuazua, Enrique , TITLE =. Systems Control Lett. , FJOURNAL =. 2003 , NUMBER =

    work page 2003

  9. [9]

    Negreanu, Mihaela and Zuazua, Enrique , TITLE =. SIAM J. Numer. Anal. , FJOURNAL =. 2006 , NUMBER =

    work page 2006

  10. [10]

    Cockburn, Bernardo and Shu, Chi-Wang , TITLE =. SIAM J. Numer. Anal. , FJOURNAL =. 1998 , NUMBER =

    work page 1998

  11. [11]

    Marica, Aurora and Zuazua, Enrique , TITLE =. SIAM J. Numer. Anal. , FJOURNAL =. 2012 , NUMBER =

    work page 2012

  12. [12]

    Control of partial differential equations , SERIES =

    Ervedoza, Sylvain and Zuazua, Enrique , TITLE =. Control of partial differential equations , SERIES =. 2012 , ISBN =

    work page 2012

  13. [13]

    Uniform boundary observability for a fourth order finite-differences semi-discretization of the 1-d wave equation , author=

    work page

  14. [14]

    Castro, Carlos and Micu, Sorin , TITLE =. Numer. Math. , FJOURNAL =. 2006 , NUMBER =

    work page 2006

  15. [15]

    2013 , PAGES =

    Ervedoza, Sylvain and Zuazua, Enrique , TITLE =. 2013 , PAGES =

    work page 2013

  16. [16]

    Ervedoza, Sylvain and Marica, Aurora and Zuazua, Enrique , TITLE =. IMA J. Numer. Anal. , FJOURNAL =. 2016 , NUMBER =

    work page 2016

  17. [17]

    ESAIM Control Optim

    Ervedoza, Sylvain , TITLE =. ESAIM Control Optim. Calc. Var. , FJOURNAL =. 2010 , NUMBER =

    work page 2010

  18. [18]

    Castro, Carlos and Micu, Sorin and M\"unch, Arnaud , TITLE =. IMA J. Numer. Anal. , FJOURNAL =. 2008 , NUMBER =

    work page 2008

  19. [19]

    M2AN Math

    M\"unch, Arnaud , TITLE =. M2AN Math. Model. Numer. Anal. , FJOURNAL =. 2005 , NUMBER =

    work page 2005

  20. [20]

    Cockburn, Bernardo and Shu, Chi-Wang , TITLE =. J. Comput. Phys. , FJOURNAL =. 1998 , NUMBER =

    work page 1998

  21. [21]

    Cockburn, Bernardo and Shu, Chi-Wang , TITLE =. Math. Comp. , FJOURNAL =. 1989 , NUMBER =

    work page 1989

  22. [22]

    , TITLE =

    Zuazua, E. , TITLE =. J. Math. Pures Appl. (9) , FJOURNAL =. 1999 , NUMBER =

    work page 1999

  23. [23]

    Discrete Contin

    Zhang, Xu and Zheng, Chuang and Zuazua, Enrique , TITLE =. Discrete Contin. Dyn. Syst. , FJOURNAL =. 2009 , NUMBER =

    work page 2009

  24. [24]

    Uniform boundary observability with

    Gagnon, Ludovick and Urquiza, Jos\'e. Uniform boundary observability with. Evol. Equ. Control Theory , FJOURNAL =. 2021 , NUMBER =

    work page 2021

  25. [25]

    and Kinton, W

    Glowinski, R. and Kinton, W. and Wheeler, M. F. , TITLE =. Internat. J. Numer. Methods Engrg. , FJOURNAL =. 1989 , NUMBER =

    work page 1989

  26. [26]

    Glowinski, Roland and Li, Chin Hsien and Lions, Jacques-Louis , TITLE =. Japan J. Appl. Math. , FJOURNAL =. 1990 , NUMBER =

    work page 1990

  27. [27]

    and Lions, J.-L

    Glowinski, R. and Lions, J.-L. , TITLE =. Acta numerica, 1994 , SERIES =. 1994 , ISBN =

    work page 1994

  28. [28]

    Ervedoza, Sylvain and Zheng, Chuang and Zuazua, Enrique , TITLE =. J. Funct. Anal. , FJOURNAL =. 2008 , NUMBER =

    work page 2008

  29. [29]

    , TITLE =

    Lions, J.-L. , TITLE =. 1988 , PAGES =

    work page 1988

  30. [30]

    and Monk, P

    Ainsworth, M. and Monk, P. and Muniz, W. , TITLE =. J. Sci. Comput. , FJOURNAL =. 2006 , NUMBER =

    work page 2006

  31. [31]

    Cockburn, Bernardo and Hou, Suchung and Shu, Chi-Wang , TITLE =. Math. Comp. , FJOURNAL =. 1990 , NUMBER =

    work page 1990

  32. [32]

    Cockburn, Bernardo and Lin, San Yih and Shu, Chi-Wang , TITLE =. J. Comput. Phys. , FJOURNAL =. 1989 , NUMBER =

    work page 1989

  33. [33]

    RAIRO Mod\'el

    Cockburn, Bernardo and Shu, Chi-Wang , TITLE =. RAIRO Mod\'el. Math. Anal. Num\'er. , FJOURNAL =. 1991 , NUMBER =

    work page 1991

  34. [34]

    Biccari, Umberto and Zuazua, Enrique , TITLE =. Found. Comput. Math. , FJOURNAL =. 2025 , NUMBER =

    work page 2025

  35. [35]

    [2024] 2024 , PAGES =

    Zuazua, Enrique , TITLE =. [2024] 2024 , PAGES =

    work page 2024

  36. [36]

    , TITLE =

    Glowinski, R. , TITLE =. J. Comput. Phys. , FJOURNAL =. 1992 , NUMBER =

    work page 1992

  37. [37]

    and Zuazua, Enrique , TITLE =

    Ignat, Liviu I. and Zuazua, Enrique , TITLE =. J. Eur. Math. Soc. (JEMS) , FJOURNAL =. 2009 , NUMBER =

    work page 2009