Pith. sign in

REVIEW 5 major objections 6 minor 43 references

Physics-informed token transformer methodology for nonlinear balance laws. I. Schwarzschild--Burgers fluid flows

T0 review · 5 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read A physics-informed token transformer that embeds an exact Rankine–Hugoniot shock-speed prior into its rollout reproduces Schwarzschild–Burgers shock and rarefaction solutions more accurately than a high-order finite-volume method, with the

desk verdict A genuinely hybrid architecture with an explicit Rankine-Hugoniot prior, but the headline accuracy claim largely measures self-consistency because the test data are generated by the same analytic Riemann solver the model hard-wires in. read the letter →

arxiv 2607.23143 v1 pith:FQL64SV3 submitted 2026-07-25 math.NA cs.NAphysics.flu-dyn

Philippe G. LeFloch , Shuyang Xiang This is my paper · ORCID
classification math.NAcs.NAphysics.flu-dyn MSC 35L6568T0765M08
keywords Schwarzschild–Burgersphysics-informedtokentransformerRankine–Hugoniotconditionbalancelawsshockcapturingneuraloperatorsteady-stateinvariantmassconservation
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 introduces a neural architecture for one-dimensional hyperbolic balance laws that combines symbolic equation tokenization with an exact computational shock-speed rule. The central claim is that the Rankine–Hugoniot jump condition should be built into the network's rollout as an analytical prior, rather than left to be learned from data. On the Schwarzschild–Burgers equation, this physics-informed token transformer achieves test-set field errors of 7.25e-4 versus 2.08e-3 for a high-order finite-volume method, with a 25-fold gain on moving shocks. Ablations show the Rankine–Hugoniot prior is the dominant component, while equation tokenization gives a small, seed-independent improvement. The paper also derives a mass-conservation displacement law for perturbed steady shocks and uses it as a diagnostic.

What carries the argument

The central object is the conserved steady-state invariant K = (1 - u^2) / g(r), where g(r)=1-2M/r. Along characteristics it is constant, so piecewise steady-state profiles u_{K,ς} = ς sqrt(1 - K g(r)) parameterize states by (K,ς). The shock speed is not learned: it is the analytic Rankine–Hugoniot quotient σ_RH = (F_R - F_L)/(u_R - u_L), evaluated at the current shock position from the left/right steady-state traces, integrated as an ODE in time. The architecture combines a Transformer that tokenizes the equation's symbolic form, a Fourier neural operator that encodes the initial field and geometry, and a bounded gated learned correction for the shock speed and field. A second identity, the

What would settle it

Run PITT on a two-shock collision requiring post-interaction re-decomposition of steady-state parameters, and compare the predicted field against a high-resolution finite-volume reference; if the post-collision field MAE jumps well above the single-front value (≈1e-4) or the Rankine–Hugoniot prior stops helping, the claim that the method extends beyond isolated Riemann problems is falsified. Alternatively, feed a compactly supported smooth bump (not piecewise steady) as initial data and observe whether the error stays O(|δu|) or grows.

Watch

Extended reading notes

Core claim

A token-transformer network that computes shock motion from the exact Rankine–Hugoniot jump condition — with steady-state traces fixed by the conserved invariant K=(1-u^2)/g(r) — reproduces analytic single-front solutions of the Schwarzschild–Burgers equation more accurately than a standard high-order finite-volume scheme, and learns only a bounded correction to the analytic prior. Test-set field MAE is 7.25e-4 versus 2.08e-3; moving shocks show 1.11e-4 versus 2.80e-3. Ablations show removing the Rankine–Hugoniot prior raises MAE by 117% and 143% at two of three seeds, while removing equation tokenization costs a consistent 7–10%. From the conservative form, the paper proves a mass-conservat

Load-bearing premise

The load-bearing premise is that the exact generalized Riemann structure of Proposition 2.3 completely describes the reference solutions used for training and testing; the network then reproduces the shock motion that its own analytical prior supplies, so the measured advantage is partly a measure of prior correctness, not of general learning.

Editorial extensions

If this is right

  • If the central claim holds, neural solvers for balance laws can localize moving discontinuities almost exactly by outsourcing shock motion to the analytic jump condition, eliminating the main source of error in learned shock capturing.
  • The consistent, seed-independent gain from equation tokenization implies that the symbolic structure of the PDE can be injected into the representation itself, not just the loss, with measurable benefit.
  • The mass-conservation displacement law provides a cheap, exact diagnostic: any solver — learned or numerical — that violates it on perturbed steady shocks is not conserving the weighted density u/g^2.
  • The method generalizes, by construction, to any one-dimensional balance law whose steady states and wave curves are known, including systems such as Schwarzschild–Euler flows treated in the companion paper.
  • In the flat spacetime limit M→0, the built-in Rankine–Hugoniot speed reduces exactly to (u_L+u_R)/2 for standard Burgers, so the method's accuracy carries over without retraining.

Reading between the lines

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

  • The 25-fold moving-shock gain may be largely an artifact of the analytic prior: a classical front-tracking or well-balanced finite-volume scheme using the same steady-state structure might achieve comparable accuracy with far less training data; the paper does not compare against such a physics-rich numerical baseline.
  • Because the Rankine–Hugoniot prior dominates, the method's generalization to truly unknown wave dynamics — post-collision, non-steady, or without an explicit invariant — is untested; a learned jump condition would likely lose most of the advantage.
  • The displacement-law residual could be converted into a physics loss term for training PITT on perturbed shocks, enforcing exact weighted-mass conservation during rollout rather than only as a post-hoc diagnostic.
  • If equation tokenization is the secondary contributor, the same PITT framework might benefit problems where the flux/source structure is only partially known: tokenization can flag model-form uncertainty while the Rankine–Hugoniot prior is withheld.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper introduces a Physics-Informed Token Transformer (PITT) method for scalar hyperbolic balance laws, demonstrated on the Schwarzschild--Burgers equation. The architecture combines symbolic equation tokenization, a Fourier neural operator encoder, and an explicit Rankine--Hugoniot (RH) prior: shock positions are advanced by the analytical jump ODE (2.14) evaluated via (3.10), while a bounded neural correction and a steady-state-based field reconstruction (3.13) represent smooth regions, rarefaction fans, and finite-resolution effects. The authors report that PITT outperforms a high-order finite-volume (RK3+MUSCL+HLL) baseline on moving shocks by a factor of 25, with an overall test field MAE of 7.25e-4 versus 2.08e-3, and that ablations show the RH prior is the dominant component, with equation tokenization giving a small consistent gain. The paper also studies multi-discontinuity initial data before wave interaction, the Burgers limit, long-time asymptotics, and a mass-conservation displacement law for perturbed steady shocks.

Significance. The analytic core of the paper is sound and useful: the steady-state invariant (Prop. 2.1), the generalized Riemann resolution (Prop. 2.3), and the shock ODE (2.14) are derived carefully and provide a clean testbed for hybrid physics-learning methods. The ablation study across three seeds is a genuine strength, as is the explicit admission of limitations in Section 8. If the numerical evaluation were independent of the analytical prior, the reported accuracy gains would be significant for operator learning of discontinuous solutions. As it stands, the headline comparison largely measures how accurately a neural rollout reproduces an algebraic/ODE prior that was used to generate the reference data, so the contribution is better described as a proof-of-concept hybrid architecture than as a validated general-purpose shock solver.

major comments (5)
  1. [§5.1, Tables 5.1 and 5.4] The central numerical claim is compromised by circularity. Section 4.2 states that the exact Riemann solver is used to generate training and validation data, with shocks advanced by the ODE (2.14) and fan boundaries by (2.15). The PITT rollout then advances the same shock ODE via (3.10) and reconstructs the field from the same steady branches via (3.13). Thus the 25x moving-shock advantage in Table 5.1 largely measures how closely an explicit Euler step of an exact ODE matches a fourth-order integration of that same ODE, relative to a diffusive finite-volume scheme. It does not establish that the learned correction captures physics outside the built-in analytical prior. The paper should add tests on data not generated by this exact Riemann structure, e.g., post-collision solutions, solutions of the same equation obtained by an independent numerical solver, or initial data not of piecewis
  2. [§6.2] The reported quantitative results are internally inconsistent. Table 5.1 gives PITT-full overall MAE 7.25e-4, steady MAE 1.97e-3, and rarefaction MAE 9.14e-5. Tables 5.2 and 5.4 for seed 2026 give overall MAE 2.88e-4, steady MAE 8.28e-4, and rarefaction MAE 5.63e-5; seed 42 gives overall 3.12e-4. None of the three seeds in Table 5.4 reproduce Table 5.1's overall or per-type values. The authors must clarify which configuration Table 5.1 corresponds to, and correct the numbers. This inconsistency undermines confidence in the exact quantitative claims, including the 25x factor.
  3. [§6.1, Section 8] The long-time asymptotic study is not performed by PITT. Section 6.2 explicitly says all experiments use the finite-volume scheme of Subsection 4.3, and the figures show finite-volume results. Yet the abstract and Section 1 state that PITT is used to study long-time dynamics. Either the long-time experiments should be rerun with PITT and compared, or the claims should be restricted to what the finite-volume baseline demonstrates. This is a discrepancy between the paper's stated scope and its actual numerical support.
  4. [§7.1, Claim 7.1] The paper explicitly defers post-interaction benchmarks (Section 6.1: 'the more stringent regime is the post-collision evolution... requires a separate error study'; Section 8 repeats this). Given that the method is presented as relevant for 'complex shock-wave dynamics' and 'interacting discontinuities', the absence of any post-collision test is a load-bearing gap. The current multi-discontinuity results in Table 6.1 are confined to pre-collision times where the local generalized Riemann decomposition is exact, so they do not test the method's ability to handle wave interactions. A single post-collision test case, even with the current Glimm-type redecomposition, would substantially strengthen the validation.
  5. [§5.3] Claim 7.1 is used as a diagnostic for PITT predictions, but its proof relies on the assumption that the perturbed solution relaxes to a steady shock in the same κ0-family, an assumption the authors themselves flag as breaking down for large |Δm| (Remark 7.1). The table of residuals (Table 7.1) is therefore informative only under that branch assumption. This is not a fatal flaw, but the claim should be labeled as a conditional statement rather than a generally established law, and the residual analysis should emphasize that the good PITT residuals in the large-|Δm| bin are not evidence of predictive skill if the same-κ0 assumption fails.
minor comments (6)
  1. [§5.2] The reproducibility section says scripts 'will be archived with the submitted version'; for an arXiv submission, it would be better to include the code or a stable repository link now.
  2. [§7, notation] The ablation conclusions are stated as though removing the RH prior and the annealing schedule are primarily robustness issues. However, at seed 42 AblA performs nearly identically to PITT-full, so the claim that the RH prior is 'dominant' is only supported at two of three seeds. The text acknowledges this, but the abstract's stronger phrasing ('dominant role') should be softened or qualified.
  3. [§3.5] The notation κ0 = K0^2 in Section 7 is introduced after the paper has used K as the steady-state invariant. This is confusing; either use K consistently or explain the square explicitly in the main text.
  4. [§5.1] In (3.11), the neural correction Δσ_NN is computed from the spatial mean of the latent state. This choice removes all spatial information from the shock-speed correction; please justify or discuss why a local evaluation near the shock is not used.
  5. [§6.2] Figure 5.1's right panel is labeled 'steady state' while the test classification in Table 5.1 uses 'steady shock.' Please make the regime labels consistent.
  6. [§6.2] The long-time figures (Figures 6.2--6.6) are described verbally but no quantitative convergence data are given for the perturbed smooth steady state or the perturbed steady shock cases; the L1 distances are only given for two of the four general-data cases. Adding a small table of errors would improve reproducibility.

Circularity Check

1 steps flagged · score 6.0 of 10

Moving-shock validation is circular: test data and PITT rollout share the same Rankine–Hugoniot ODE and steady-state branches, so the headline 25x gain largely measures the built-in analytic prior, not learned shock discovery.

  1. self definitional [Section 2.2 (after Prop. 2.3); Section 3.5, Eqs. (3.9)-(3.13); Section 4.2]
    "The shock trajectory given by the ODE (2.14) is precisely the quantity approximated by PITT: the analytic term σ(t)_RH in the rollout (3.10) evaluates the right-hand side of (2.14) at the current predicted shock position. ... We use the exact Riemann solver to generate training and validation data. ... Full trajectories are generated semi-analytically: the steady branches are evaluated from the closed formulas, while a fourth-order Runge-Kutta method with step ∆t is used for (2.14) in shock cases and for the rarefaction fan boundaries in rarefaction cases."

    The reference data are generated by integrating ODE (2.14) from the same steady-state branches used in PITT's rollout: (3.10) evaluates the RHS of (2.14) at the predicted shock position, and (3.13) reconstructs the field from the same u_L/u_R steady profiles that generate the data (4.1); rarefaction fan boundaries are advanced by the same characteristic ODE (2.15) used in data generation. Thus the 'predicted' moving-shock trajectory is not an independently learned quantity—it is a numerical solve of the very ODE that produced the reference, with only a small gated correction. The Table 5.1 moving-shock advantage (1.11e-4 vs 2.80e-3) therefore mostly measures exact-ODE integration against a diffusive finite-volume solver, not discovery of unknown shock physics. The circularity is partial: t

full rationale

The main reduction is in the evaluation protocol, not in the analytical theorems. Propositions 2.1 and 2.3 are proved self-contained in the paper, so the many self-citations to [19,20,22,23,24] are not themselves load-bearing circularity; they are supported by the included proofs. The genuine circular step is that the reference solutions and the PITT prior share the same exact generator: the generalized Riemann solver of Prop. 2.3, the steady-state invariant (2.5), the shock ODE (2.14), and the characteristic ODE (2.15). Consequently the headline moving-shock accuracy is largely a consistency check between an explicit Euler step of the analytic ODE and an RK4 integration of the same ODE, relative to a diffusive baseline. The paper is transparent about this—Section 3.5 states the RH speed is not learned, and Section 8 concedes that 'the training data use the exact generalized Riemann structure of a scalar balance law' and that validation is 'mainly focused on single-front dynamics and pre-interaction data'—but transparency does not remove the by-construction character of the moving-shock prediction. There is still non-circular content: equation tokenization gives a seed-independent but modest gain (7-10% in the ablation), the learned field residual acts on smooth and rarefaction regions that are not fully determined by the prior, and the finite-volume baseline is external. The long-time Section 6.2 results are produced by the finite-volume solver, not by PITT, so they do not rescue the PITT validation. The numerical inconsistency between Table 5.1 and Table 5.4 (steady MAE 1.97e-3 vs 8.28e-4 and inconsistent overall weighted values) is a correctness/reporting concern rather than a circularity concern, so it is not counted in the score. Overall this is a partial circularity: the central moving-shock claim reduces by construction, but the methodology retains independent learnable components, giving a score of 6 rather than 8.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

No new physical entities are postulated. The paper's central dependence is on hand-set ML hyperparameters and an evaluation protocol whose training data are generated by the same analytic Riemann solver that the architecture embeds.

free parameters (2)
  • learned correction bounds (epsilon, mu) and gate (lambda, delta) = epsilon=0.002; mu=0.05; lambda=500; delta=0.015
    Hand-chosen in Section 4.4. They bound the neural correction relative to the analytic RH and steady-state priors; Section 5.2 shows the architecture can be sensitive to training setup.
  • shock-width annealing schedule w = 0.01, then 0.005, then 0.003 by training epoch
    Hand-set in Section 4.4. Ablation PITT-AblC shows that removing the schedule causes an order-of-magnitude accuracy collapse at one of three seeds, so central performance depends on this choice.
assumptions (4)
  • standard math Weak solutions of scalar balance laws with regular source satisfy the Rankine-Hugoniot jump condition (2.6).
    Invoked in Sections 2.1 and 3.5 to build the analytic shock-speed prior; standard hyperbolic theory.
  • domain assumption The generalized Riemann solution of Prop. 2.3 is the physically admissible entropy solution.
    Used to generate all training and test targets in Section 4.2. The solution formulas come from prior work [22,23,24] and are not re-proven in full here.
  • ad hoc to paper For perturbed steady shocks, the solution relaxes to a steady shock in the same kappa0-family (Claim 7.1).
    Section 7.1-7.2. The paper itself flags branch switching when Delta m is too large in Remark 7.1; if this assumption fails, the displacement law is not valid.
  • domain assumption Piecewise steady-state profiles are dense in BV and the Glimm-type decomposition is valid pre-collision (Prop. 2.4).
    Used in Section 6.1 for general BV data. The paper only tests pre-interaction data and explicitly notes that post-interaction behavior requires a separate error study.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Physics-informed token transformer methodology for nonlinear balance laws. I. Schwarzschild--Burgers fluid flows." pith.science (2026). https://pith.science/paper/FQL64SV3

@misc{pith2026260723143,
  author       = {Pith},
  title        = {Pith review of: Physics-informed token transformer methodology for nonlinear balance laws. I. Schwarzschild--Burgers fluid flows},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FQL64SV3}},
  note         = {Machine review of arXiv:2607.23143}
}
read the original abstract

We introduce a Physics-Informed Token Transformer (PITT) methodology for nonlinear hyperbolic balance laws in one space dimension, using piecewise steady-state profiles for the representation of approximate weak solutions. The method combines symbolic equation tokenization, a Fourier neural operator encoder, an explicit Rankine--Hugoniot law for shock motion, and a learned correction term. For clarity, we present it here for the relativistic Schwarzschild--Burgers equation, a scalar model for spherically symmetric fluid flows on a Schwarzschild background. For this model the steady-state invariant and the generalized Riemann solutions are explicit, and they can therefore be built into the neural evolution. In particular, the leading discontinuities are advanced by the analytical jump condition, while the learned part reconstructs smooth regions, rarefaction fans, geometric dependence, and finite-resolution effects. The method is designed to locate wave fronts accurately and to preserve the relevant steady states. We test our PITT method on moving shocks, stationary shocks, rarefaction waves, and compare it with a standard high-order finite-volume approximation. We also analyze the standard Burgers limit (when the Schwarzschild mass tends to zero). The Rankine--Hugoniot prior plays the dominant role in these tests, while equation tokenization gives a systematic additional gain. The method is relevant for problems involving geometric effects and/or complex shock-wave dynamics, and is used here to study the long-time dynamics of perturbations of steady-state solutions. In particular, we exhibit an asymptotic law of propagation for the shock location of perturbed steady-state flows.

Figures

Figures reproduced from arXiv: 2607.23143 by the authors.

Figure 5.1
Figure 5.1. Predicted velocity profiles at t ∈ {0.05, 0.10, 0.15, 0.20}s for three representative test cases (M = 1.0, r0 = 4.0). Solid lines: exact solution; dashed lines: PITT-Full. Left: moving shock (KL = 0.2, KR = 0.5, ςL = ςR = +1); center: steady shock (KL = KR = 0.3, ςL = −ςR = +1); right: steady state (KL = KR = 0.3, ςL = ςR = +1). In all cases the two curves are visually indistinguishable [PITH_FULL_IMAGE:figures/ful… view at source ↗
Figure 5.2
Figure 5.2. Benchmark summary for the test set. The figure compares the PITT prediction with the finite [PITH_FULL_IMAGE:figures/full_fig_p019_5_2.png] view at source ↗
Figure 6.1
Figure 6.1. PITT-General on general bounded-variation initial data ( [PITH_FULL_IMAGE:figures/full_fig_p023_6_1.png] view at source ↗
Figures from the paper (6 more)
Figure 6.2
Figure 6.2. Figure 6.2: Asymptotic behavior: perturbed smooth steady state ( [PITH_FULL_IMAGE:figures/full_fig_p024_6_2.png]
Figure 6.3
Figure 6.3. Figure 6.3: Asymptotic behavior: perturbed steady shock ( [PITH_FULL_IMAGE:figures/full_fig_p024_6_3.png]
Figure 6.4
Figure 6.4. Figure 6.4: Asymptotic behavior: general initial data with [PITH_FULL_IMAGE:figures/full_fig_p025_6_4.png]
Figure 6.5
Figure 6.5. Figure 6.5: Asymptotic behavior: general initial data with [PITH_FULL_IMAGE:figures/full_fig_p025_6_5.png]
Figure 6.6
Figure 6.6. Figure 6.6: Asymptotic behavior: general initial data with [PITH_FULL_IMAGE:figures/full_fig_p026_6_6.png]
Figure 7.1
Figure 7.1. Figure 7.1: Asymptotic shock displacement r1 − r0 vs. weighted perturbation mass ∆m = R δ(r)/g(r) 2 dr, at M = 1, r0 = 4.0, for K2 0 = 0.25 (in-distribution), 0.5, and 0.75. Each panel shows ∆m ∈ [−4, 2], well within the regime where the mass-conservation formula (7.4) admits a …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 2 linked inside Pith

  1. [1]

    and LeFloch P.G., Ergodicity and Hopf–Lax–Oleinik formula for fluid flows evolving around a black hole under a random forcing,Stoch PDE: Anal

    Bakhtin Y. and LeFloch P.G., Ergodicity and Hopf–Lax–Oleinik formula for fluid flows evolving around a black hole under a random forcing,Stoch PDE: Anal. Comp.6 (2018), 746–785

    2018

  2. [2]

    Bouchut F., Nonlinear stability of finite volume methods for hyperbolic conservation laws and well- balanced schemes for sources, Frontiers in Mathematics, Birkhäuser, Basel, 2004

    2004

  3. [3]

    34, Curran Associates, Inc., 2021, pp

    Cao S., Choose a transformer: Fourier or Galerkin, inAdvances in Neural Information Processing Systems, Vol. 34, Curran Associates, Inc., 2021, pp. 24924–24940

    2021

  4. [4]

    and Kerswell R.R., Godunov loss functions for modelling of hyperbolic conservation laws, Comput

    Cassia R.G. and Kerswell R.R., Godunov loss functions for modelling of hyperbolic conservation laws, Comput. Methods Appl. Mech. Engrg.437 (2025), article 117782

    2025

  5. [5]

    Ceylan T., LeFloch P.G., and Okutmustur B., A finite volume method for the relativistic Burgers equation on a FLRW background spacetime,Commun. Comput. Phys.23 (2018), 500–519

    2018

  6. [6]

    325, 5th ed., Springer Verlag, Berlin, 2026

    Dafermos C.M., Hyperbolic conservation laws in continuum physics, Grundlehren der Mathematischen Wissenschaften, Vol. 325, 5th ed., Springer Verlag, Berlin, 2026

    2026

  7. [7]

    De Ryck T., Mishra S., and Molinaro R., wPINNs: Weak physics-informed neural networks for approx- imating entropy solutions of hyperbolic conservation laws,SIAM J. Numer. Anal.62 (2024), 811–841

    2024

  8. [8]

    and LeFloch P.G., Convergence of the finite volume method on a Schwarzschild background, ESAIM Math

    Dong S. and LeFloch P.G., Convergence of the finite volume method on a Schwarzschild background, ESAIM Math. Model. Numer. Anal.53 (2019), 1459–1476

    2019

Show all 43 references
  1. [9]

    Methods Appl

    Ferrer-Sánchez A., Martín-Guerrero J.D., Ruiz de Austri-Bazan R., Torres-Forné A., and Font J.A., Gradient-annihilated PINNs for solving Riemann problems: Application to relativistic hydrodynamics, Comput. Methods Appl. Mech. Engrg.424 (2024), article 116906. 30

    2024

  2. [10]

    and LeFloch P.G., Formulation and convergence of the finite volume method for con- servation laws on spacetimes with boundary,Numer

    Giesselmann J. and LeFloch P.G., Formulation and convergence of the finite volume method for con- servation laws on spacetimes with boundary,Numer. Math.144 (2020), 751–785

    2020

  3. [11]

    Pure Appl

    Glimm J., Solutions in the large for nonlinear hyperbolic systems of equations,Comm. Pure Appl. Math. 18 (1965), 697–715

    1965

  4. [12]

    Harten A., Lax P.D., and van Leer B., On upstream differencing and Godunov-type schemes for hyper- bolic conservation laws,SIAM Rev.25 (1983), 35–61

    1983

  5. [13]

    Huang H., Yang V., and Liu Y., Neural networks with local converging inputs (NNLCI) for solving conservation laws, Part I: 1D problems,Commun. Comput. Phys.34 (2023), 290–317

    2023

  6. [14]

    and Kang M., Approximating numerical fluxes using Fourier neural operators for hyperbolic conservation laws,Commun

    Kim T. and Kang M., Approximating numerical fluxes using Fourier neural operators for hyperbolic conservation laws,Commun. Comput. Phys.37 (2025), 420–456

    2025

  7. [15]

    Kovachki N., Li Z., Liu B., Azizzadenesheli K., Bhattacharya K., Stuart A., and Anandkumar A., Neural operator: Learning maps between function spaces with applications to PDEs,J. Mach. Learn. Res.24 (2023), article 89, 1–97

    2023

  8. [16]

    Lax P.D., Hyperbolic systems of conservation laws. II,Comm. Pure Appl. Math.10 (1957), 537–566

    1957

  9. [17]

    Lax P.D., Hyperbolic systems of conservation laws and the mathematical theory of shock waves, SIAM, Philadelphia, 1973

    1973

  10. [18]

    LeFloch P.G., Hyperbolic systems of conservation laws: The theory of classical and nonclassical shock waves, Lectures in Mathematics ETH Zürich, Birkhäuser, Basel, 2002

    2002

  11. [19]

    Derivation and finite volume approximation,SIAM J

    LeFloch P.G., Makhlof H., and Okutmustur B., Relativistic Burgers equations on curved spacetimes. Derivation and finite volume approximation,SIAM J. Numer. Anal.50 (2012), 2136–2158

    2012

  12. [20]

    and Makhlof H., A geometry-preserving finite volume method for compressible fluids on Schwarzschild spacetime,Commun

    LeFloch P.G. and Makhlof H., A geometry-preserving finite volume method for compressible fluids on Schwarzschild spacetime,Commun. Comput. Phys.15 (2014), 827–852

    2014

  13. [21]

    LeFloch P.G., Parés C., and Pimentel-García E., A class of well-balanced algorithms for relativistic fluids on a Schwarzschild background,J. Sci. Comput.89 (2021), no. 1, article 3, 43 pp

    2021

  14. [22]

    and Xiang S., Weakly regular fluid flows with bounded variation on the domain of outer communication of a Schwarzschild black hole spacetime,J

    LeFloch P.G. and Xiang S., Weakly regular fluid flows with bounded variation on the domain of outer communication of a Schwarzschild black hole spacetime,J. Math. Pures Appl.106 (2016), 1038–1090

    2016

  15. [23]

    and Xiang S., A numerical study of the relativistic Burgers and Euler equations on a Schwarzschild black hole exterior,Commun

    LeFloch P.G. and Xiang S., A numerical study of the relativistic Burgers and Euler equations on a Schwarzschild black hole exterior,Commun. Appl. Math. Comput. Sci.13 (2018), 271–301

    2018

  16. [24]

    and Xiang S., Weakly regular fluid flows with bounded variation on the domain of outer communication of a Schwarzschild black hole spacetime

    LeFloch P.G. and Xiang S., Weakly regular fluid flows with bounded variation on the domain of outer communication of a Schwarzschild black hole spacetime. II,J. Math. Pures Appl.122 (2019), 272–317

    2019

  17. [25]

    and Xiang S., Physics-informed token transformer (PITT) methods for nonlinear balance laws

    LeFloch P.G. and Xiang S., Physics-informed token transformer (PITT) methods for nonlinear balance laws. II. Schwarzschild–Euler fluid flows, in preparation

  18. [26]

    LiZ., KovachkiN., AzizzadenesheliK., LiuB., BhattacharyaK., StuartA., andAnandkumarA., Fourier neural operator for parametric partial differential equations, inProceedings of the 9th International Conference on Learning Representations, OpenReview.net, 2021

    2021

  19. [27]

    Liu L., Liu S., Xie H., Xiong F., Yu T., Xiao M., Liu L., and Yong H., Discontinuity computing using physics-informed neural networks,J. Sci. Comput.98 (2024), no. 1, article 22, 38 pp

    2024

  20. [28]

    Liu L., Li T., Gelb A., and Lee Y., Entropy-stable conservative flux-form neural networks,J. Sci. Comput.107 (2026), no. 1, article 3

    2026

  21. [29]

    Liu L., Zhang L., and Gelb A., Neural entropy-stable conservative flux-form neural networks for learning hyperbolic conservation laws,J. Comput. Phys.553 (2026), article 114719. 31

    2026

  22. [30]

    Liu Y., Miao S., and Wei C., On classification of dynamics for dust fluid under spherical symmetry in Schwarzschild spacetime, Preprint arXiv:2606.28375, 2026

    2026 arXiv

  23. [31]

    and Novruzi A., Non-diffusive neural network method for hyperbolic conservation laws,J

    Lorin E. and Novruzi A., Non-diffusive neural network method for hyperbolic conservation laws,J. Comput. Phys.513 (2024), article 113161

    2024

  24. [32]

    Learn.: Sci

    Lorsung C., Li Z., and Barati Farimani A., Physics informed token transformer for solving partial differential equations,Mach. Learn.: Sci. Technol.5 (2024), article 015032

    2024

  25. [33]

    and Hutter F., Decoupled weight decay regularization, inProceedings of the 7th Interna- tional Conference on Learning Representations, OpenReview.net, 2019

    Loshchilov I. and Hutter F., Decoupled weight decay regularization, inProceedings of the 7th Interna- tional Conference on Learning Representations, OpenReview.net, 2019

    2019

  26. [34]

    Lu L., Jin P., Pang G., Zhang Z., and Karniadakis G.E., Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators,Nat. Mach. Intell.3 (2021), 218–229

    2021

  27. [35]

    Methods Appl

    Mao Z., Jagtap A.D., and Karniadakis G.E., Physics-informed neural networks for high-speed flows, Comput. Methods Appl. Mech. Engrg.360 (2020), article 112789

    2020

  28. [36]

    Oubarka I., Kissami I., Boubekeur M., Benkhaldoun F., Madrane A., and Saadi Z., Weak and entropy physics-informed neural networks for conservation laws, Preprint arXiv:2603.24819, 2026

    2026

  29. [37]

    Patsatzis D.G., di Bernardo M., Russo L., and Siettos C., GoRINNs: Godunov–Riemann informed neural networks for learning hyperbolic conservation laws,J. Comput. Phys.534 (2025), article 114002

    2025

  30. [38]

    Methods Appl

    Peyvan A., Oommen V., Jagtap A.D., and Karniadakis G.E., RiemannONets: Interpretable neural operators for Riemann problems,Comput. Methods Appl. Mech. Engrg.426 (2024), article 116996

    2024

  31. [39]

    Pronk J., Porth O., and Davelaar J., Fill the gaps: Continuous in time interpolation of fluid dynamical simulations, Preprint arXiv:2511.08239, 2025

    2025

  32. [40]

    Raissi M., Perdikaris P., and Karniadakis G.E., Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J. Comput. Phys.378 (2019), 686–707

    2019

  33. [41]

    vanLeerB., Towardstheultimateconservativedifferencescheme.V.Asecond-ordersequeltoGodunov’s method,J. Comput. Phys.32 (1979), 101–136

    1979

  34. [42]

    30, Curran Associates, Inc., 2017, pp

    Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A.N., Kaiser Ł., and Polosukhin I., Attention is all you need, inAdvances in Neural Information Processing Systems, Vol. 30, Curran Associates, Inc., 2017, pp. 5998–6008

    2017

  35. [43]

    A Machine learning terminology This appendix recalls a few machine learning-related terms used in the paper

    Wang C., Luo H., Wang K., Zhu G., and Luo M., Solving Euler equations with multiple discontinuities via Separation-Transfer Physics-Informed Neural Networks, Preprint arXiv:2505.20361, 2025. A Machine learning terminology This appendix recalls a few machine learning-related te...

    2025 arXiv

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.