Explicit and Stable Pseudospectral Time-Domain Method for the F\"oppl-von K\'arm\'an Equations

Stefan Bilbao; Victor Zheleznov

REVIEW 3 major objections 4 minor 16 references

Explicit and Stable Pseudospectral Time-Domain Method for the F\"oppl-von K\'arm\'an Equations

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A stable, explicit pseudospectral solver for the nonlinear Föppl–von Kármán plate equations

desk verdict New explicit SAV-pseudospectral method for FvK plates; the energy-stability core is sound and the stress-test's dealiasing concern doesn't survive the staggered-grid DCT. read the letter →

arxiv 2608.06139 v1 pith:SLH7YJCB submitted 2026-08-06 cs.SD physics.comp-ph

Victor Zheleznov , Stefan Bilbao This is my paper · ORCID

classification cs.SDphysics.comp-ph MSC 65M7074K2065T50

keywords nonlinearplatevibrationFöppl–vonKármánequationspseudospectralmethodmodalsynthesisscalarauxiliaryvariablediscretesineandcosinetransformsenergy-stabletimesteppingmusicalacoustics

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that the Föppl–von Kármán plate equations—the standard model for high-amplitude vibration of thin plates and for pitch glides and crashes in percussion instruments—can be simulated in the modal domain without paying the prohibitive cost of nonlinear mode coupling. The proposed pseudospectral method evaluates nonlinear products on a spatial grid while computing spatial derivatives exactly in the modal domain, using discrete sine and cosine transforms to enforce simply supported boundaries. The authors prove the discrete nonlinear potential is non-negative and use a scalar auxiliary variable technique to obtain an explicit, energy-stable time-stepping scheme. If correct, the method turns an $O(N^4)$ coupling-tensor evaluation into an $O(N^2 \log N)$ transform cost per step while retaining modal advantages such as frequency-range control. Numerical tests at audio rate show energy conservation to machine accuracy and controlled drift of the auxiliary variable.

What carries the argument

The load-bearing object is the discrete triple self-adjointness identity $\dot q^T f_1(q,\xi)=\xi^T f_2(q,\dot q)$, obtained from product-to-sum identities for sine and cosine transforms applied to grid functions bandlimited under $S_x\otimes S_y$ and $\hat{C}_x\otimes\hat{C}_y$. The product rules relate second-derivative operators like $D_{xx,c}H_c(\alpha\odot\beta)$ to combinations of first-order differentiation matrices acting on the factors, and the brick-wall filters $H_s=(S_x\otimes S_y)^T(S_x\otimes S_y)$ and $H_c=(C_x\otimes C_y)^T(C_x\otimes C_y)$ are what allow the transforms to be moved across products. The identity supplies the non-negative potential $V(q)=\frac14\|k^2\odot \xi(q)\|_2^2$ used by the scalar auxiliary variable technique; it is the mechanism that converts a stiff, implicitly coupled system into an explicit update with a Sherman-Morrison matrix inversion. The 3/2 rule sets the grid size $N_a=\lceil 3M_a/2\rceil+1$ so that aliasing products land in wavenumber thirds that are filtered out by the transforms.

What would settle it

Compute $\dot q^T f_1(q,\xi)-\xi^T f_2(q,\dot q)$ for a random bandlimited pair $(q,\dot q,\xi)$ with grid sizes set by the 3/2 rule; if the result exceeds roundoff, the potential $V(q)$ is not the true discrete potential and the stability proof collapses. A second check is long free vibration: any secular growth in the discrete energy $H^{n-1/2}$ beyond machine precision would contradict the claimed conservation.

Watch

Extended reading notes

Core claim

The central claim is that the nonlinearity of the Föppl–von Kármán equations can be written as $f_1(q,\xi)=-\nabla_q V(q)$ with $V(q)=\frac{1}{4}\|k^2\odot \xi(q)\|_2^2\ge 0$, where $q$ are sine modal coefficients of displacement, $\xi$ are cosine modal coefficients of the Airy function, and $k$ is the wavenumber vector. The proof rests on discrete product rules: under bandlimited grid functions and the 3/2-rule truncation with brick-wall filters, the identities $\dot q^T f_1(q,\xi)=\xi^T f_2(q,\dot q)$ and $f_2(q,\dot q)=\frac12 \frac{d}{dt}f_2(q,q)$ hold, so that $\xi^T f_2(q,\dot q)=-\dot V(q)$. With this potential, the paper introduces an auxiliary variable $\psi=\sqrt{2V(q)+\varepsilon}$ and rewrites the modal equations as an explicit system, adding a control term that drives the numerical $\psi$ toward the true $\sqrt{2V(q)+\varepsilon}$ and reduces drift. The scheme is shown to conserve a discrete energy to machine accuracy in an audio-rate simulation with 1027 modes. This would mean that nonlinear modal synthesis of plates no longer requires a fourth-order coupling tensor or an implicit iterative solve, making real-time nonlinear plate sound synthesis practical for larger mode counts.

Load-bearing premise

The load-bearing premise is that the product-to-sum identities for sine and cosine transforms remain exact for grid functions bandlimited under the 3/2-rule truncation and brick-wall filters; if that discrete identity fails, the non-negative potential and the explicit energy-stable time stepping no longer follow.

Editorial extensions

If this is right

Nonlinear plate synthesis can be computed with transforms per time step rather than an $O(N^4)$ coupling tensor, so larger mode counts become feasible on ordinary hardware.
The scheme is explicit and stable under the condition $\omega_{ij}<\pi f_s$, meaning all simulated modal frequencies must stay below the Nyquist limit.
Energy conservation holds to machine accuracy for free vibration, giving long-run reliability without iterative implicit solvers.
The auxiliary-variable drift can be suppressed by the added control term $\lambda_0$, preventing audible artefacts at high excitation amplitudes.
The method extends the scalar auxiliary variable approach, previously used for nonlinear strings, to the Föppl–von Kármán plate.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the product identities remain exact under more general boundary conditions, the same grid/modal hybrid could be applied to clamped or free plates, but the sine/cosine bases would need replacement and the potential proof would have to be reworked.
The $O(N^2\log N)$ estimate assumes fast sine/cosine transforms; a direct comparison against optimised finite-difference gong solvers at equal audio quality would test whether the modal advantages outweigh transform overhead.
The drift-control term depends on a parameter $\lambda_0$; an automatic schedule or controller for $\lambda_0$ could remove the tuning step and make the method more reliable for real-time use.
If the energy identity is exact only under the 3/2 rule, then the explicit stability guarantee may degrade for grids with $N_a\le M_a$; testing the scheme near this boundary would map the practical stability margin.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a genuine new method, and the central stability argument holds up. The stress-test worry about the 3/2 rule does not land once you account for the staggered DCT grid.

What's new: a pseudospectral SAV scheme for the Föppl–von Kármán plate that keeps derivatives exact in the modal domain, evaluates products on grid, and provably conserves a discrete energy. The non-negative potential V(q) is explicit, the triple self-adjointness identity is derived in detail, and the reported energy error is at roundoff. That's real, reproducible evidence—the code is shipped. The complexity reduction from O(N^4) coupling tensors to O(N^2 log N) per step is a meaningful practical gain for audio-rate modal synthesis.

Soft spots, all addressable: the drift-control term has a 0/0 singularity when velocity is zero, and the paper never says how it's regularized. The zero-constant-mode claim is asserted with 'it can be shown' rather than demonstrated. The discrete energy positivity bound is cited from [16], not proved here. And there's no wall-clock benchmark, so the complexity advantage is asymptotic only. These don't undermine the core method.

On the stress-test note: the M=2, N=4 counterexample is mistaken. On the staggered grid, cos(4πx) vanishes at the collocation points, so the product rule holds there. More generally, the DCT-II on half-integer points has a null at the Nyquist mode; aliasing from high product modes lands above the retained truncation. The 3/2 rule is sufficient for the filtered products, exactly as the paper claims. That said, the paper could have given this argument instead of just asserting the product rules.

Recommendation: yes, send to peer review. It's a compact conference paper with some hand-waving, but the method is new, the numerics are clean, and the open issues are fixable. A serious referee should ask for the drift-control regularization and a runtime comparison, but the work deserves referee time.

Referee Report

3 major / 4 minor

Summary. The paper proposes an explicit pseudospectral time-domain method for the Föppl–von Kármán plate equations with simply supported boundary conditions. Displacement and Airy function are expanded in sine and cosine series; products are evaluated on a staggered grid through discrete sine and cosine transforms, while spatial derivatives are computed exactly in the modal domain. The nonlinearity is rewritten as the gradient of a non-negative potential V(q)=1/4 ||k^2⊙ξ(q)||^2, using product rules for the discrete transforms, and a scalar auxiliary variable with a drift-control term yields an explicit time-stepping scheme with a claimed discrete energy balance. Numerical experiments at audio rate show relative energy error near machine precision and reduced auxiliary-variable drift, with spectrograms illustrating pitch glides and crash-like behavior. The central claim is an explicit, stable, and efficient solver for nonlinear modal synthesis.

Significance. If the algebraic identities are established rigorously, the method is a substantial practical advance: it reduces the nonlinearity cost from O(M^4) coupling-tensor evaluation to O(N^2 log N) transform cost while preserving modal-domain advantages such as frequency-range control. The paper ships reproducible source code, gives concrete numerical evidence of near-machine-precision energy conservation and of drift regulation, and makes falsifiable predictions about the effect of the control term on spectrograms. Its main weakness is that the load-bearing algebraic identities are asserted rather than proved, and the fully discrete energy balance is stated without derivation.

major comments (3)

[§3.5 (product rules) and §3.4 (3/2 rule)] The exactness of the product rules is the keystone of the SAV construction: it is used to show that qdot^T f1(q,ξ) = ξ^T f2(q,qdot), hence that V(q)=1/4||k^2⊙ξ(q)||^2 is the true potential. The manuscript only says 'Considering product-to-sum identities... the following product rules hold' and justifies the grid size by the 3/2 rule. This is not a complete proof for the specific type-II DCT/DST on the staggered grid (6). A formal proof should use the alias map of the half-integer DCT: a product frequency ℓ≤2M_a is visible in the retained band only if k≡±ℓ (mod 2N_a) for some k≤M_a. With N_a=ceil(3M_a/2)+1 this condition is indeed satisfied (the marginal value ℓ=N_a is invisible at the half-integer collocation points), so the claim is plausible, but the manuscript should state and prove it. Without this, the non-negativity of V and the discrete energy identity are not established for the implemented discretization.
[§3.5, last paragraph] The statement 'It can be shown that f2(q,q) has a zero constant mode so that (7b) is fully satisfied' is load-bearing because the operator k^4 is not invertible for the constant mode. The paper needs to provide the short argument (for example, by showing that the discrete projection of l2(u,u) onto the constant cosine mode vanishes, or by a product-to-sum/telescoping argument). As written, the definition of ξ(q) by elementwise division in (7b) is not fully justified.
[§4.2, energy balance] The fully discrete energy balance δt+ H^{n-1/2} = P^n_e - P^n_d is stated without derivation. Since the central claim is 'explicit and stable', the identity should be proved for the actual definitions of g^n including the drift-control term g_mod and for the interleaved time grids. In particular, the cancellation that removes the nonlinear term depends on a discrete product identity connecting μt+ψ^{n-1/2}, δt+ψ^{n-1/2}, and δt·q^n via (10b). Citing [16] for positivity of the kinetic-energy term is fine, but the energy balance itself should be derived or a precise pointer to the corresponding result in [7] should be given.

minor comments (4)

[§3.4] Typo: 'This is not be possible' should read 'This is not possible'.
[§4.1] The denominator ||qdot||_1 in g_mod is undefined when qdot=0; the paper should state the convention for that case (e.g., g_mod=0 when qdot=0).
[§5] The first case study initializes 'q00 = 5', but the modal indices in (5) start at 1; please clarify whether this is q11 or another convention.
[§4.2] The phrase 'yelding a sampling rate' contains a typo ('yielding').

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the central derivation is self-contained, with only minor self-citations that do not reduce the claims to their inputs.

full rationale

The central derivation chain is self-contained. Section 3.5 derives the discrete identity qdot^T f1(q,xi) = xi^T f2(q,qdot) from explicitly stated product rules for bandlimited grid functions, and the potential V(q) = 1/4 ||k^2 ⊙ xi(q)||_2^2 follows algebraically from (7b) and the symmetry f2(q,qdot) = f2(qdot,q); non-negativity of V is literal from its definition as a squared norm. Section 4's SAV update and discrete energy balance are derived inside the paper, with numerical energy conservation verified to machine accuracy (Fig. 1). The paper's reliance on [7], [9], and [16]—all with overlapping authorship—is as a toolbox: the SAV reformulation, the Sherman-Morrison update, and a general energy-nonnegativity bound. These are external published results with assumptions that do not include the plate-specific target claim, and none of them supplies fitted constants or defines the predicted behavior by construction. The only exposed assumption is exactness of the Section 3.5 product rules under the 3/2-rule grid; if that assumption fails, the discrete energy identity would not hold, which would be a correctness defect rather than circularity, because the identity is asserted and numerically tested rather than defined into existence by the potential. No step reduces a prediction to its own inputs, so there is no significant circularity; the small score reflects minor self-citations that are not load-bearing.

Assumptions & free parameters 2 free parameters · 5 assumptions · 1 invented entities

The method introduces no new physical entities and no fitted constants; the only hand-chosen numbers are the SAV offset epsilon and the drift control gain lambda_0. The central results rest on the FvK model, orthogonality of the DCT and DST bases, exactness of the stated product rules in Section 3.5, the cited stability theorem [16], and the asserted zero-constant-mode property.

free parameters (2)

lambda_0 (drift control gain) = 10^3 s^-1 (with 0 used for comparison)
Chosen by hand to regulate drift of the auxiliary variable; the reported two-orders-of-magnitude drift reduction depends on this value, though the scheme's basic energy balance does not.
epsilon (offset in psi) = arbitrary positive constant (value not stated)
Introduced to keep sqrt(2V+epsilon) differentiable; the paper does not report the chosen value, though results should be insensitive to it.

assumptions (5)

domain assumption Föppl-von Kármán equations and simply supported boundary conditions (u=0, Delta u=0, partial_n phi=0, partial_n Delta phi=0) describe high-amplitude thin plate motion.
Adopted in Section 2 as the continuous model; standard in musical acoustics, cited to [1] and [12].
standard math Truncated DST-II and DCT-II rows are orthonormal (S_a S_a^T = I_{Ma}, C_a C_a^T = I_{Ma+1}) for Na > Ma, so modal and grid representations are norm-preserving.
Used in Sections 3.1 and 3.2 to define transforms and the 3/2 anti-aliasing rule.
standard math Product-to-sum rules in Section 3.5 hold exactly for bandlimited grid functions and justify the identity qdot^T f1 = xi^T f2(q,qdot).
The paper states these rules and uses them to prove V(q) >= 0; if they fail under truncation, the energy balance collapses.
domain assumption The discrete energy positivity bound leading to the stability condition omega < pi fs is accepted from reference [16].
Section 4.2 cites Bilbao-Ducceschi-Zama [16] for non-negativity of the discrete energy term; the proof is not reproduced.
ad hoc to paper f2(q,q) has a zero constant mode, so equation (7b) is invertible for xi by elementwise division.
Asserted in Section 3.5 as 'it can be shown' without proof; the inversion of the biharmonic operator depends on it.

invented entities (1)

Auxiliary variable psi = sqrt(2V(q) + epsilon)
purpose: Quadratizes the nonlinear potential so the time-stepping update is explicit and the discrete energy contains kappa^2 psi^2 / 2.
It is a numerical device carried over from SAV methods [7], not a physical quantity; it has no falsifiable external handle and its value is determined by the simulation state.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Explicit and Stable Pseudospectral Time-Domain Method for the F\"oppl-von K\'arm\'an Equations." pith.science (2026). https://pith.science/paper/SLH7YJCB

@misc{pith2026260806139,
  author       = {Pith},
  title        = {Pith review of: Explicit and Stable Pseudospectral Time-Domain Method for the F\"oppl-von K\'arm\'an Equations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SLH7YJCB}},
  note         = {Machine review of arXiv:2608.06139}
}

read the original abstract

Modal synthesis is a widely-used technique for simulation of musical instrument dynamics. In the linear case, a modal decomposition leads to an uncoupled system of damped and forced harmonic oscillators which can be efficiently solved by standard time-stepping methods. However, extensions to nonlinear problems are challenging due to the presence of products of modal expansions in the governing equations. In the case of the F\"oppl-von K\'arm\'an plate, the nonlinear coupling between the modes is described by a fourth-order tensor and is prohibitively expensive to evaluate in the modal domain. In this work, we propose a pseudospectral method in which the products are evaluated on a grid in the spatial domain while spatial derivatives are computed exactly in the modal domain. Discrete sine and cosine transforms between the modal and spatial domains are used to impose simply supported boundary conditions for the plate. Finally, we prove non-negativity of the nonlinear potential energy of the system and employ a scalar auxiliary variable technique for explicit and stable time integration in the modal domain. As a result, we reduce the computational cost of modal synthesis while preserving its advantages like a precise control over the simulated frequency range. Sound examples are presented.

Figures

Figures reproduced from arXiv: 2608.06139 by the authors.

**Figure 1.** Relative energy error E n (left) and its enlarged section that reveals numerical quantisation (right). Horizontal lines indicate multiples of machine accuracy 2 −52 ≈ 2.2 × 10−16 . where the energy Hn−1/2 is defined as: Hn−1/2 = 1 2 ∥δt−q n ∥ 2 2 + 1 2 (q n ) T(ω˜ 2 ⊙ q n−1 ) + + κ 2 2 [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗

**Figure 3.** Spectrograms for displacement of the plate at increasing excitation amplitudes, as indicated. The third and fourth spectrograms are given at identical excitation amplitudes to demonstrate drift regulation of the numerical solver. approach [10] using an optimised C++ implementation. In addition, extension to other boundary conditions is needed to further evaluate suitability of pseudospectral methods to problems in … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[16]

Explicit exactly energy-conserving methods for Hamilto- nian systems,

S. Bilbao, M. Ducceschi, and F. Zama, “Explicit exactly energy-conserving methods for Hamilto- nian systems,”J. Comput. Phys., vol. 472, 2023, Art. no. 111697. 12th Convention of the European Acoustics Association Graz, Austria•8 th – 12 th September 2026•

work page 2023
[7]

Power- balanced drift regulation for scalar auxiliary variable methods: Application to real-time sim- ulation of nonlinear string vibrations,

T. Risse, T. H´ elie, and S. Bilbao, “Power- balanced drift regulation for scalar auxiliary variable methods: Application to real-time sim- ulation of nonlinear string vibrations,” inProc. 28th Int. Conf. Digital Audio Effects, Ancona, Italy, 2025, pp. 126–133

work page 2025
[1]

Bilbao,Numerical Sound Synthesis: Finite Difference Schemes and Simulation in Musical Acoustics

S. Bilbao,Numerical Sound Synthesis: Finite Difference Schemes and Simulation in Musical Acoustics. Chichester, UK: John Wiley & Sons, 2009

work page 2009
[2]

Modal approach for nonlinear vibrations of damped impacted plates: Application to sound synthesis of gongs and cymbals,

M. Ducceschi and C. Touz´ e, “Modal approach for nonlinear vibrations of damped impacted plates: Application to sound synthesis of gongs and cymbals,”J. Sound Vib., vol. 344, pp. 313– 331, 2015

work page 2015
[3]

J. P. Boyd,Chebyshev and Fourier Spectral Methods. New York: Dover Publications, 2000

work page 2000
[4]

Fast Differentiable Modal Simulation of Non-Linear Strings, Mem- branes, and Plates,

R. Diaz and M. Sandler, “Fast Differentiable Modal Simulation of Non-Linear Strings, Mem- branes, and Plates,” inProc. 28th Int. Conf. Dig- ital Audio Effects, Ancona, Italy, 2025, pp. 134– 141

work page 2025
[5]

Fornberg,A Practical Guide to Pseudospec- tral Methods

B. Fornberg,A Practical Guide to Pseudospec- tral Methods. Cambridge University Press, 1996

work page 1996
[6]

Pseudospectral Time-Domain (PSTD) Meth- ods for the Wave Equation: Realizing Bound- ary Conditions with Discrete Sine and Cosine Transforms,

E. S. Wise, J. Jaros, B. T. Cox, and B. E. Treeby, “Pseudospectral Time-Domain (PSTD) Meth- ods for the Wave Equation: Realizing Bound- ary Conditions with Discrete Sine and Cosine Transforms,”J. Theor. Comput. Acoust., vol. 29, no. 04, 2021, Art. no. 2050021

work page 2021

Show all 16 references

[8]

Solution of von- K´ arm´ an dynamic non-linear plate equations us- ing a pseudo-spectral method,

R. M. Kirby and Z. Yosibash, “Solution of von- K´ arm´ an dynamic non-linear plate equations us- ing a pseudo-spectral method,”Comput. Meth- ods Appl. Mech. Engrg., vol. 193, no. 6, pp. 575– 599, 2004

2004
[9]

Convergence analysis and relaxation techniques for modal scalar auxiliary variable methods applied to nonlinear transverse string vibration,

R. Russo, C. J. Webb, M. Ducceschi, and S. Bilbao, “Convergence analysis and relaxation techniques for modal scalar auxiliary variable methods applied to nonlinear transverse string vibration,”Proc. Meet. Acoust., vol. 56, no. 1, 2025, Art. no. 035007

2025
[10]

Real-Time Gong Synthesis,

S. Bilbao, C. Webb, Z. Wang, and M. Ducceschi, “Real-Time Gong Synthesis,” inProc. 26th Int. Conf. Digital Audio Effects, Copenhagen, Den- mark, 2023, pp. 21–28

2023
[11]

Percussion Instrument Modelling In 3D: Sound Synthesis Through Time Domain Numerical Simulation,

A. Torin, “Percussion Instrument Modelling In 3D: Sound Synthesis Through Time Domain Numerical Simulation,” Ph.D. dissertation, Uni- versity of Edinburgh, 2015

2015
[12]

Boundary con- ditions and mode jumping in the buckling of a rectangular plate,

D. Schaeffer and M. Golubitsky, “Boundary con- ditions and mode jumping in the buckling of a rectangular plate,”Commun. Math. Phys., vol. 69, pp. 209–236, 1979

1979
[13]

The spectral method and numerical continuation algorithm for the von K´ arm´ an problem with postbuckling behaviour of solutions,

A. D. Muradova, “The spectral method and numerical continuation algorithm for the von K´ arm´ an problem with postbuckling behaviour of solutions,”Adv. Comput. Math, vol. 29, pp. 179– 206, 2008

2008
[14]

Geometrically non- linear flexural vibrations of plates: In-plane boundary conditions and some symmetry prop- erties,

O. Thomas and S. Bilbao, “Geometrically non- linear flexural vibrations of plates: In-plane boundary conditions and some symmetry prop- erties,”J. Sound Vib, vol. 315, no. 3, pp. 569– 590, 2008

2008
[15]

Britanak, P

V. Britanak, P. C. Yip, and K. R. Rao,Dis- crete Cosine and Sine Transforms. Oxford, UK: Elsevier Ltd., 2006

2006

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [16]

Explicit exactly energy-conserving methods for Hamilto- nian systems,

S. Bilbao, M. Ducceschi, and F. Zama, “Explicit exactly energy-conserving methods for Hamilto- nian systems,”J. Comput. Phys., vol. 472, 2023, Art. no. 111697. 12th Convention of the European Acoustics Association Graz, Austria•8 th – 12 th September 2026•

work page 2023

[2] [7]

Power- balanced drift regulation for scalar auxiliary variable methods: Application to real-time sim- ulation of nonlinear string vibrations,

T. Risse, T. H´ elie, and S. Bilbao, “Power- balanced drift regulation for scalar auxiliary variable methods: Application to real-time sim- ulation of nonlinear string vibrations,” inProc. 28th Int. Conf. Digital Audio Effects, Ancona, Italy, 2025, pp. 126–133

work page 2025

[3] [1]

Bilbao,Numerical Sound Synthesis: Finite Difference Schemes and Simulation in Musical Acoustics

S. Bilbao,Numerical Sound Synthesis: Finite Difference Schemes and Simulation in Musical Acoustics. Chichester, UK: John Wiley & Sons, 2009

work page 2009

[4] [2]

Modal approach for nonlinear vibrations of damped impacted plates: Application to sound synthesis of gongs and cymbals,

M. Ducceschi and C. Touz´ e, “Modal approach for nonlinear vibrations of damped impacted plates: Application to sound synthesis of gongs and cymbals,”J. Sound Vib., vol. 344, pp. 313– 331, 2015

work page 2015

[5] [3]

J. P. Boyd,Chebyshev and Fourier Spectral Methods. New York: Dover Publications, 2000

work page 2000

[6] [4]

Fast Differentiable Modal Simulation of Non-Linear Strings, Mem- branes, and Plates,

R. Diaz and M. Sandler, “Fast Differentiable Modal Simulation of Non-Linear Strings, Mem- branes, and Plates,” inProc. 28th Int. Conf. Dig- ital Audio Effects, Ancona, Italy, 2025, pp. 134– 141

work page 2025

[7] [5]

Fornberg,A Practical Guide to Pseudospec- tral Methods

B. Fornberg,A Practical Guide to Pseudospec- tral Methods. Cambridge University Press, 1996

work page 1996

[8] [6]

Pseudospectral Time-Domain (PSTD) Meth- ods for the Wave Equation: Realizing Bound- ary Conditions with Discrete Sine and Cosine Transforms,

E. S. Wise, J. Jaros, B. T. Cox, and B. E. Treeby, “Pseudospectral Time-Domain (PSTD) Meth- ods for the Wave Equation: Realizing Bound- ary Conditions with Discrete Sine and Cosine Transforms,”J. Theor. Comput. Acoust., vol. 29, no. 04, 2021, Art. no. 2050021

work page 2021