Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

A physics-informed neural network reproduces the dispersive QED Dyson–Schwinger solution across momentum scales; the paper also argues spectral positivity should be a diagnostic, not a neural constraint.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-04 11:22 UTC pith:GRSQPOXG

load-bearing objection Useful neural-surrogate recipe, but the abstract claims a positivity result the body doesn't contain and the central 'reproduction' is trained in, so the paper needs major reframing before it is trustworthy. the 3 major comments →

arxiv 2510.24728 v2 pith:GRSQPOXG submitted 2025-10-06 hep-ph cs.LGhep-lathep-th

Spectral functions in Minkowski quantum electrodynamics from neural reconstruction

Rodrigo Carmo Terin This is my paper
classification hep-ph cs.LGhep-lathep-th
keywords Dyson–Schwinger equationsMinkowski spacetimephysics-informed neural networksspectral functionsLehmann representationquantum electrodynamicsdynamical mass generationrainbow truncation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Working in quenched rainbow QED with Landau gauge, the paper sets out to show that a Minkowski-space physics-informed neural network (M-PINN) can solve the Dyson–Schwinger equation for the fermion scalar dressing B(p^2) directly on the real momentum axis, in both spacelike and timelike regions. The network is trained with a residual loss that embeds the reduced rainbow integral equation, together with multiscale regularisation, Fourier features for the timelike branch, and monotonicity/smoothing penalties for the spacelike branch. Against a dispersive benchmark built from Lehmann spectral representations and subtracted dispersion relations, the M-PINN matches quantitatively from infrared to ultraviolet scales for couplings alpha = 0.1, 0.2, 0.4, 0.6, in both on-shell and momentum-subtraction renormalisation schemes. The abstract further claims that neural reconstructions with free output reproduce the zero crossing of the Fukuda–Kugo criterion above the critical coupling alpha_c = pi/3, while positivity-constrained ansätze fail there, leading to the methodological conclusion that spectral positivity is a diagnostic of the Lehmann representation, not a constraint to be imposed blindly. If these claims hold, residual-based neural solvers offer a compact, differentiable route to Minkowski-space propagators and ultimately to more realistic truncations.

Core claim

The central claim is that the M-PINN—a neural network B_theta trained by minimising the residual of the one-dimensional split rainbow equation R[B](p^2)=B(p^2) - (3 alpha/4 pi)[integral terms]—reproduces the B(p^2) computed by a spectral dispersive solver in quenched rainbow QED in Landau gauge. The training fixes the same truncation, gauge, and subtraction point as the benchmark, and the agreement holds across the full momentum domain for alpha in {0.1, 0.2, 0.4, 0.6}, including the timelike branch with its mild UV oscillations, for both on-shell and momentum-subtraction renormalisation. The paper treats M(p^2)=B(p^2)/A(p^2) with A(p^2)≈1, so the learned function is identified with the dyna

What carries the argument

The load-bearing objects are (i) the Lehmann spectral representation of the fermion propagator together with once-subtracted dispersion relations, which turn the DSE into coupled unitary integral equations for the spectral weights sigma_v and sigma_s, and (ii) the reduced rainbow residual R[B](p^2) that the M-PINN minimises. The dispersive solver discretises the spectral variable on a log grid, handles Cauchy principal values by an extrapolation scheme, and reconstructs B(p^2) by subtracted dispersion relations. The neural solver represents B(p^2) with a multilayer perceptron, using Fourier features on the timelike branch, a multiscale loss with IR/intermediate/UV windows, a perturbative UV

Load-bearing premise

The argument assumes A(p^2) ≈ 1 so that M(p^2) = B(p^2); if A deviates appreciably at the couplings studied, the network is solving a reduced equation, not the full rainbow DSE, and the benchmark agreement does not validate the dynamical mass.

What would settle it

Solve the full coupled A(p^2)–B(p^2) rainbow system at alpha = 0.6 in the same scheme; if the vector dressing A deviates from 1 by more than a few percent at the scales plotted, then the reduced equation solved by the network is not the DSE being benchmarked, and agreement on B alone does not validate the dynamical mass M = B/A.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • For the four tested couplings and both renormalisation schemes, the M-PINN reproduces the dispersive B(p^2) from IR to UV, so residual-based neural solvers can handle nonlocal integral equations with realistic renormalisation directly on the real axis.
  • Because the neural solution is continuous and differentiable, it can serve as a compact surrogate for the spectral solver, easing extensions to dressed vertices, unquenched photons, and uncertainty-aware variants.
  • The abstract's claim that free-output networks capture the zero crossing above alpha_c = pi/3 while positivity-constrained ansätze fail implies that spectral positivity should be treated as a diagnostic of the Lehmann representation, not a constraint imposed a priori.
  • The same architecture, trained on lattice correlators instead of DSE residuals, could reconstruct spectral densities and bypass ill-posed analytic continuation, connecting DSE and lattice approaches.

Where Pith is reading between the lines

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

  • Because the body's comparison is between the free-output M-PINN and the dispersive solver, the abstract's stronger claim—that positivity-constrained ansätze fail above alpha_c—is not directly demonstrated by the figures; a constrained-versus-free run on the same benchmark would settle it.
  • The identification M(p^2)=B(p^2) rests on an unquantified A(p^2)≈1; computing the coupled A and B equations at the larger couplings would show whether the reduced residual is sufficient or whether the benchmark is only validating a surrogate of the scalar dressing.
  • The paper's framing suggests a broader methodological moral: in theories like QCD where loss of spectral positivity is a physical signal, imposing positivity as a hard neural constraint could suppress exactly the nonperturbative dynamics under study; one could test this by running the same M-PINN on a model with known positivity violation.
  • A differentiable, continuous surrogate for B(p^2) could make renormalization-group improvements and uncertainty quantification straightforward, since the UV template in the loss could be replaced by an RG-motivated tail and retrained without changing the solver, an extension the paper lists as a next step.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper presents a physics-informed neural network (M-PINN) for the quenched rainbow QED Dyson–Schwinger equation in Minkowski space, targeting the scalar fermion dressing B(p^2). The body develops a dispersive solver based on Lehmann representations and subtracted dispersion relations as a benchmark, then trains the network with a loss that combines the DSE residual with multi-scale regularization and, critically, direct data terms matching the dispersive benchmark. Figures 1–2 show agreement between the PINN and the dispersive solution for α = 0.1, 0.2, 0.4, 0.6 on timelike and spacelike momenta. The abstract additionally claims that free-output neural reconstructions reproduce a Fukuda–Kugo zero crossing above α_c = π/3 while positivity-constrained ansätze fail in the supercritical regime, and draws a conclusion about spectral positivity.

Significance. The methodological idea of combining a dispersive benchmark with a residual-based multi-scale PINN loss for Minkowski DSEs is useful and could be a step toward differentiable surrogates for propagator equations. The explicit comparison of a neural solver with a traditional spectral solver under a common truncation is a strength. However, the abstract's central physical claim about positivity-constrained ansätze is absent from the body, and the body's main agreement is largely trained into the network via data-loss terms, so the independent validation value is limited. If the claims were supported, the paper would be significant; in its current form, the primary advertised results are not established.

major comments (3)
  1. [Abstract vs. Section IV] The abstract asserts: 'Neural reconstructions with free output reproduce this behavior, while positivity-constrained ansätze fail in the supercritical regime' and concludes that spectral positivity should not be imposed blindly. No positivity-constrained ansatz is defined, implemented, or tested anywhere in the body. The only related statement is Section IV: 'a systematic study of ... strong-coupling regimes near the loss of Lehmann positivity is still needed,' which explicitly defers such a study to future work. The Fukuda–Kugo zero crossing above α_c = π/3 is also not demonstrated in the text; it is only mentioned through reference [31]. The abstract therefore makes a central claim that the manuscript does not support.
  2. [§III, Eqs. (19)–(21)] The network is trained with L_data^TL/SL = ⟨(B_θ − B_trad)²⟩, L_mid_reg, and L_high_reg, which directly minimize the squared difference between the network output and the dispersive benchmark over the full momentum range. Consequently, the agreement shown in Figs. 1–2 is largely by construction: the network is fitted to the benchmark, so the 'reproduction' is not an independent check. To support the claim that the M-PINN 'reproduces the dispersive solution,' the authors should either train without the benchmark data terms (or on a held-out subset) and compare on untouched intervals, or report that the physical residual L_phys is small and show that the solution satisfies the DSE independently. Without this, the central validation claim is circular.
  3. [§III, Eq. (17) and Eq. (6)] The paper assumes A(p²) ≈ 1, so that M(p²) = B(p²), and the residual in Eq. (17) contains only B. However, the dispersive solver itself computes A(p²) from ρ_v via Eq. (15), and Eq. (6) defines the dynamical mass as M = B/A. The manuscript never quantifies how close A(p²) is to 1 for the couplings studied (α = 0.1–0.6). If A deviates from unity, the reduced equation solved by the PINN is not the full rainbow DSE, and the benchmark comparison on B alone does not validate the physical mass function. This assumption should be tested by evaluating A from the dispersive spectral solution and reporting it alongside B.
minor comments (6)
  1. [Title/Abstract] The title advertises 'spectral functions,' and the abstract mentions 'Fukuda–Kugo equation, the spectral unitary equations, and the modified unitary equations,' but the body does not present a separation of these equations or any spectral-density results. No figure shows ρ_s, ρ_v, or σ_s, σ_v. Consider either including such quantities or adjusting the title/abstract to match the actual content.
  2. [§III, Eq. (29)] The term 1_TL in Eq. (29) is not defined. If it is an indicator function for the timelike branch, it should be explicitly introduced; otherwise the reader cannot tell how B_UV is used in the spacelike loss L_tail.
  3. [§III, Eq. (27)–(28)] The loss weights (w_phys, w_data^TL/SL, w_mid, w_high, etc.) are listed symbolically, but their numerical values are never given. The statement 'the same weights are used for all couplings' is not reproducible without these values.
  4. [Abstract vs. Body] The abstract mentions a comparison of 'on-shell and momentum-subtraction schemes,' but the body appears to use a single renormalization condition (µ² = m²) throughout. No explicit comparison of two schemes is presented in the results.
  5. [§III, Eqs. (10)–(16)] The spectral solver equations are quoted without derivation or a pointer to the specific equations of Ref. [28]. Since these equations are the benchmark, the authors should state their provenance explicitly and note any approximations, ranges of validity, and the treatment of thresholds (e.g., s0) and UV cutoffs.
  6. [General] There are several presentation issues: 'Facuty' in the affiliation, missing superscripts in Eq. (17), incomplete reference [53] ('8 2025'), and minimal figure captions that do not identify line styles or list the couplings. These should be corrected.

Circularity Check

1 steps flagged

M-PINN's 'reproduction' of the dispersive benchmark is built into the loss (Eqs. 19-21, 27-28); the abstract's supercritical positivity-constrained claim is absent from the body, which defers that study to future work.

specific steps
  1. fitted input called prediction [Section III, Eqs. (19)-(21) and total losses (27)-(28)]
    "L^{TL/SL}_{data} = ⟨(B_θ − B_trad)²⟩; L^{mid}_{reg} = ⟨[log(1+|Bθ|)−log(1+|B trad|)]²⟩_{mid}; L^{high}_{reg} = ⟨[log(1+|Bθ|)−log(1+|B trad|)]²⟩_{high} ... and L_{TL}=... w^{TL}_{data} L^{TL}_{data} + ... L_{SL}=... w^{SL}_{data} L^{SL}_{data}+..."

    The benchmark comparison is not an external check: B_trad is inserted as the target in the squared and log-window losses that define the network's training objective. Any good optimizer will therefore return B_θ ≈ B_trad on the TL/SL grids and log windows, making Section IV's 'the M-PINN reproduces the dispersive solution' a property of the loss function. The residual Lphys tests only the reduced DSE, not the agreement with B_trad, so the central 'quantitative agreement' claim is at least partly by construction.

full rationale

The manuscript is self-contained against its chosen benchmark; there is no uniqueness theorem or load-bearing self-citation chain. However, the central quantitative claim reduces to a fit: L_data and L_reg minimize (B_θ−B_trad)² and log-differences, so the reported 'reproduction' is guaranteed by training, not independently predicted. The abstract's stronger claim that 'positivity-constrained ansätze fail in the supercritical regime' is not supported by any implementation: no constrained network, loss term, figure, or table appears, and Section IV explicitly says 'a systematic study of system dependence and of strong-coupling regimes near the loss of Lehmann positivity is still needed.' That is an evidence/reporting gap rather than a circular derivation, but it compounds the score because the advertised result is not derivable from the presented equations. The A(p²)≈1 reduction (M=B) is an unquantified assumption, but not circular. Overall: one central 'prediction' is fitted input; plus an unsupported abstract claim. Score 7.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

No new physical entities are introduced; the M-PINN is a numerical method, not new field content. The central comparison rests on the dispersive benchmark and a PINN trained to match it; no independent physics input beyond the rainbow DSE is supplied.

free parameters (3)
  • UV template coefficients β, γ, ω = not specified
    B_UV in Eq (29) anchors the ultraviolet branch; coefficients are not given and appear fitted/tuned to the benchmark or perturbative tail.
  • Loss weights {w_phys, w_data^TL/SL, w_mid, w_high, w_cont, w_ir, w_tail, w_mono, w_H1} = not specified
    Weights combine physical residual with benchmark-matching losses (Eqs 27–28); no values are given, and the paper only says they are the same for all α.
  • Network hyperparameters (5 layers, width 96, 32 Fourier pairs, 6000 Adam epochs) = 5/96/32/6000
    Architecture choices are stated but not justified; they are hand-selected and affect the benchmark fit.
axioms (4)
  • domain assumption The fermion propagator admits a Lehmann representation (Eq 3): S(p) = ∫ ds [ρ_s(s) + p̸ ρ_v(s)]/(p² − s + iϵ).
    Assumed for asymptotic states; the paper uses it to build the dispersive benchmark. The very point of the abstract's positivity diagnostic is to test this assumption, so using it as an input is a premise.
  • domain assumption The spectral unitary equations (10)–(16) from Ref [28] are the correct Minkowski-space DSE in rainbow-Landau truncation.
    Quoted without rederivation; the benchmark is taken as ground truth.
  • ad hoc to paper A(p²) ≈ 1, so M(p²) = B(p²) and the residual in Eq (17) is the full DSE.
    Not justified quantitatively; if A deviates, the solved equation is not the DSE.
  • ad hoc to paper The UV template B_UV(p²) = β − γ log(1+|p²|) + 1_TL asin(ω log(1+|p²|)) is the correct asymptotic form.
    The tail loss anchors the network to this form (Eq 29); the form is asserted, not derived.

pith-pipeline@v1.3.0-alltime-deepseek · 11050 in / 15215 out tokens · 119844 ms · 2026-08-04T11:22:48.051484+00:00 · methodology

0 comments
read the original abstract

We study neural reconstructions of quenched rainbow quantum electrodynamics (QED) Dyson--Schwinger benchmarks in Minkowski-related kinematics. Using the dispersive formulation as motivation, we separate the Euclidean Fukuda--Kugo equation, the spectral unitary equations, and the modified unitary equations. The Fukuda--Kugo benchmark is solved directly and shows the expected zero crossing above the critical region $\alpha_c=\pi/3$. Neural reconstructions with free output reproduce this behavior, while positivity-constrained ans\"atze fail in the supercritical regime. Thus, spectral positivity should be treated as a diagnostic of the Lehmann representation, not imposed blindly as a neural constraint.

Figures

Figures reproduced from arXiv: 2510.24728 by Rodrigo Carmo Terin.

Figure 1
Figure 1. Figure 1: Comparison among PINNs (left) and the traditional dispersive Dyson–Schwinger solver (right) for the [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparison between M-PINNs and the traditional dispersive Dyson–Schwinger solver for the real part of [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Solving Functional Renormalization Group Equations with Neural Networks

    hep-ph 2026-03 conditional novelty 6.0

    A neural network that learns fRG flows from the equation residual, with a large-N analytic baseline, matches finite-difference and discontinuous-Galerkin solvers for O(N) models.

Reference graph

Works this paper leans on

66 extracted references · 21 canonical work pages · cited by 1 Pith paper

  1. [1]

    Freeman J. Dyson. The s matrix in quantum electrodynamics.Phys. Rev., 75:1736–1755, 1949. doi: 10.1103/PhysRev.75.1736

    doi:10.1103/physrev.75.1736 1949

  2. [2]

    On the green’s functions of quantized fields.Proc

    Julian Schwinger. On the green’s functions of quantized fields.Proc. Natl. Acad. Sci. USA, 37:452–455, 1951. doi: 10.1073/pnas.37.7.452

    doi:10.1073/pnas.37.7.452 1951

  3. [3]

    On the green’s functions of quantized fields ii.Proc

    Julian Schwinger. On the green’s functions of quantized fields ii.Proc. Natl. Acad. Sci. USA, 37:455–459, 1951. doi: 10.1073/pnas.37.7.455

    doi:10.1073/pnas.37.7.455 1951

  4. [4]

    Roberts and Anthony G

    Craig D. Roberts and Anthony G. Williams. Dyson–schwinger equations and their application to hadronic physics.Prog. Part. Nucl. Phys., 33:477–575, 1994. doi:10.1016/0146-6410(94)90049-3

    doi:10.1016/0146-6410(94)90049-3 1994

  5. [5]

    The infrared behavior of qcd green’s functions.Phys

    Reinhard Alkofer and Lorenz von Smekal. The infrared behavior of qcd green’s functions.Phys. Rept., 353:281–465, 2001. doi:10.1016/S0370-1573(01)00010-2

    doi:10.1016/s0370-1573(01)00010-2 2001

  6. [6]

    Roberts and Sebastian M

    Craig D. Roberts and Sebastian M. Schmidt. Dyson–schwinger equations: Density, temperature and continuum strong qcd.Prog. Part. Nucl. Phys., 45:S1–S103, 2000. doi:10.1016/S0146-6410(00)90011-5

    doi:10.1016/s0146-6410(00)90011-5 2000

  7. [7]

    Christian S. Fischer. Qcd at finite temperature and chemical potential from dyson–schwinger equations.Prog. Part. Nucl. Phys., 105:1–60, 2019. doi:10.1016/j.ppnp.2019.01.002

    doi:10.1016/j.ppnp.2019.01.002 2019

  8. [8]

    Atkinson, P

    D. Atkinson, P. W. Johnson, and P. Maris. Dynamical chiral symmetry breaking with an infrared dominant vertex.Phys. Rev. D, 42:602–609, 1990. doi:10.1103/PhysRevD.42.602

    work page doi:10.1103/physrevd.42.602 1990

  9. [9]

    Maris, C

    P. Maris, C. D. Roberts, and P. C. Tandy. Pion mass and decay constant.Phys. Lett. B, 420:267–273, 1998. doi: 10.1016/S0370-2693(97)01535-9

    doi:10.1016/s0370-2693(97)01535-9 1998

  10. [10]

    Fischer, Reinhard Alkofer, Tobias Dahm, and Peter Maris

    Christian S. Fischer, Reinhard Alkofer, Tobias Dahm, and Peter Maris. Dynamical chiral symmetry breaking in unquenched qed3.Phys. Rev. D, 70:073007, 2004. doi:10.1103/PhysRevD.70.073007. 8

    work page doi:10.1103/physrevd.70.073007 2004

  11. [11]

    Rojas, A

    E. Rojas, A. Ayala, A. Bashir, and A. Raya. Dynamical mass generation in qed with magnetic fields.Phys. Rev. D, 77: 093004, 2008. doi:10.1103/PhysRevD.77.093004

    work page doi:10.1103/physrevd.77.093004 2008

  12. [12]

    Pennington, Anthony G

    Ayse Kızılersu, Tom Sizer, Michael R. Pennington, Anthony G. Williams, and Richard Williams. Reconstruct- ing the fermion–photon vertex in strongly coupled gauge theories.Phys. Rev. D, 91(6):065015, 2015. doi: 10.1103/PhysRevD.91.065015

    work page doi:10.1103/physrevd.91.065015 2015

  13. [13]

    J. S. Ball and T.-W. Chiu. Analytic properties of the vertex function in gauge theories. i.Phys. Rev. D, 22:2542–2549,

  14. [14]

    Curtis, Michael R

    David C. Curtis, Michael R. Pennington, and D. A. Walsh. On the gauge dependence of dynamical fermion masses.Phys. Lett. B, 249:528–532, 1990. doi:10.1016/0370-2693(90)91005-3

    work page doi:10.1016/0370-2693(90)91005-3 1990

  15. [15]

    Curtis and Michael R

    David C. Curtis and Michael R. Pennington. Nonperturbative truncation of the schwinger–dyson equations.Phys. Rev. D, 44:536–539, 1991. doi:10.1103/PhysRevD.44.536

    work page doi:10.1103/physrevd.44.536 1991

  16. [16]

    Cornwall

    John M. Cornwall. Dynamical mass generation in continuum qcd.Phys. Rev. D, 26:1453–1478, 1982. doi: 10.1103/PhysRevD.26.1453

    doi:10.1103/physrevd.26.1453 1982

  17. [17]

    Marciano and Heinz Pagels

    William J. Marciano and Heinz Pagels. Quantum chromodynamics: A review.Phys. Rept., 36:137–276, 1978. doi: 10.1016/0370-1573(78)90208-9

    doi:10.1016/0370-1573(78)90208-9 1978

  18. [18]

    Pennington

    Adnan Bashir, Ayse Kızılersu, and Michael R. Pennington. Gauge independent solutions to the fermion propagator dyson–schwinger equation.Phys. Rev. D, 57:1242–1248, 1998. doi:10.1103/PhysRevD.57.1242

    work page doi:10.1103/physrevd.57.1242 1998

  19. [19]

    Victor M. B. Guzm´ an and Adnan Bashir. Constraining the transverse structure of the fermion–photon vertex in qed.Phys. Rev. D, 107(7):073008, 2023. doi:10.1103/PhysRevD.107.073008

    work page doi:10.1103/physrevd.107.073008 2023

  20. [20]

    A. C. Aguilar, A. A. Natale, and P. S. Rodrigues da Silva. Relating a gluon mass scale to an infrared fixed point in pure gauge qcd.Phys. Rev. Lett., 90:152001, 2003. doi:10.1103/PhysRevLett.90.152001

    work page doi:10.1103/physrevlett.90.152001 2003

  21. [21]

    Markus Q. Huber. Nonperturbative properties of yang–mills theories.Phys. Rept., 879:1–92, 2020. doi: 10.1016/j.physrep.2020.04.004

    doi:10.1016/j.physrep.2020.04.004 2020

  22. [22]

    Maris and P

    P. Maris and P. C. Tandy. Bethe–salpeter study of vector meson masses and decay constants.Phys. Rev. C, 60:055214,

  23. [23]

    Gernot Eichmann, Helios Sanchis-Alepuz, Richard Williams, Reinhard Alkofer, and Christian S. Fischer. Baryons as relativistic three-quark bound states.Prog. Part. Nucl. Phys., 91:1–100, 2016. doi:10.1016/j.ppnp.2016.07.001

    doi:10.1016/j.ppnp.2016.07.001 2016

  24. [24]

    Roberts, and Sebastian M

    Minghui Ding, Craig D. Roberts, and Sebastian M. Schmidt. Emergence of hadron mass and structure.Particles, 6(1): 57–120, 2023. doi:10.3390/particles6010005

    doi:10.3390/particles6010005 2023

  25. [25]

    Gimeno-Segovia and F

    M. Gimeno-Segovia and F. J. Llanes-Estrada. From euclidean to minkowski space with the cauchy–riemann equations. European Physical Journal C, 56:557–569, 2008. doi:10.1140/epjc/s10052-008-0676-5. URLhttps://doi.org/10.1140/ epjc/s10052-008-0676-5

    work page doi:10.1140/epjc/s10052-008-0676-5 2008

  26. [26]

    Huber, Wolfgang J

    Markus Q. Huber, Wolfgang J. Kern, and Reinhard Alkofer. Analytic structure of three-point functions from contour deformations.Phys. Rev. D, 107:074026, Apr 2023. doi:10.1103/PhysRevD.107.074026. URLhttps://link.aps.org/ doi/10.1103/PhysRevD.107.074026

    doi:10.1103/physrevd.107.074026 2023

  27. [27]

    Gernot Eichmann, Pedro Duarte, M. T. Pe˜ na, and Alfred Stadler. Scattering amplitudes and contour deformations.Phys. Rev. D, 100:094001, Nov 2019. doi:10.1103/PhysRevD.100.094001. URLhttps://link.aps.org/doi/10.1103/PhysRevD. 100.094001

    work page doi:10.1103/physrevd.100.094001 2019

  28. [28]

    Minkowski solution of dyson–schwinger equations in momentum subtraction scheme.J

    Vladim ´ ırˇSauli. Minkowski solution of dyson–schwinger equations in momentum subtraction scheme.J. High Energy Phys., 2003(02):001, 2003. doi:10.1088/1126-6708/2003/02/001

    work page doi:10.1088/1126-6708/2003/02/001 2003

  29. [29]

    New insight in dynamical symmetry breaking via imaginary part analysis.Nucl

    Andreas Blumhofer and J¨ urgen Manus. New insight in dynamical symmetry breaking via imaginary part analysis.Nucl. Phys. B, 515:522–548, 1998. doi:10.1016/S0550-3213(98)00017-1

    work page doi:10.1016/s0550-3213(98)00017-1 1998

  30. [30]

    David Atkinson and J. C. R. Bloch. Qcd in the infrared with exact angular integrations.Mod. Phys. Lett. A, 13:1055–1064,

  31. [31]

    Schwinger–dyson equation for massless vector theory and the absence of a fermion pole.Nuclear Physics B, 117(1):250–264, 1976

    Reijiro Fukuda and Taichiro Kugo. Schwinger–dyson equation for massless vector theory and the absence of a fermion pole.Nuclear Physics B, 117(1):250–264, 1976. ISSN 0550-3213. doi:10.1016/0550-3213(76)90572-1. URLhttps://www. sciencedirect.com/science/article/pii/0550321376905721

    arXiv 1976

  32. [32]

    G.V. Efimov. Non-local quantum theory of the scalar field.Communications in Mathematical Physics, 5:42–56, 1967. doi:10.1007/BF01646357. URLhttps://doi.org/10.1007/BF01646357

    doi:10.1007/bf01646357 1967

  33. [33]

    Cornwall and Joannis Papavassiliou

    John M. Cornwall and Joannis Papavassiliou. Gauge invariant three-gluon vertex in qcd.Phys. Rev. D, 40:3474–3483,

  34. [34]

    Gauge-independent off-shell fermion self-energies at two loops: the cases of qed and qcd.Phys

    Daniele Binosi and Joannis Papavassiliou. Gauge-independent off-shell fermion self-energies at two loops: the cases of qed and qcd.Phys. Rev. D, 65:085003, 2002. doi:10.1103/PhysRevD.65.085003

    work page doi:10.1103/physrevd.65.085003 2002

  35. [35]

    Application of the background field method to the electroweak standard model.Nucl

    Ansgar Denner, Stefan Dittmaier, and Gerhard Weiglein. Application of the background field method to the electroweak standard model.Nucl. Phys. B, 440:95–128, 1995. doi:10.1016/0550-3213(95)00037-S

    work page doi:10.1016/0550-3213(95)00037-s 1995

  36. [36]

    Looking at qed with dyson–schwinger equations: Basic equations, ward–takahashi identities and the two-photon–two-fermion irreducible vertex.Few Body Syst., 64(3):67, 2023

    Orlando Oliveira, Helena Lessa Macedo, and Rodrigo Carmo Terin. Looking at qed with dyson–schwinger equations: Basic equations, ward–takahashi identities and the two-photon–two-fermion irreducible vertex.Few Body Syst., 64(3):67, 2023. doi:10.1007/s00601-023-01846-5

    work page doi:10.1007/s00601-023-01846-5 2023

  37. [37]

    Lagaris, Aristidis Likas, and Dimitrios I

    Isaac E. Lagaris, Aristidis Likas, and Dimitrios I. Fotiadis. Artificial neural networks for solving ordinary and partial differential equations.IEEE Trans. Neural Netw., 9(5):987–1000, 1998. doi:10.1109/72.712178

    doi:10.1109/72.712178 1998

  38. [38]

    Karniadakis

    Maziar Raissi, Paris Perdikaris, and George E. Karniadakis. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations.J. Comput. Phys., 378:686–707,

  39. [39]

    Karniadakis, Ioannis G

    George E. Karniadakis, Ioannis G. Kevrekidis, Lu Lu, Paris Perdikaris, Sifan Wang, and Liu Yang. Physics-informed machine learning.Nat. Rev. Phys., 3:422–440, 2021. doi:10.1038/s42254-021-00314-5

    doi:10.1038/s42254-021-00314-5 2021

  40. [40]

    Karniadakis

    Maziar Raissi, Paris Perdikaris, and George E. Karniadakis. Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations.Science, 367(6481):1026–1030, 2020. doi:10.1126/science.aaw4741

    doi:10.1126/science.aaw4741 2020

  41. [41]

    Georgios Kissas, Yibo Yang, Ehsan H. H. Kim, William R. H. Magland, Paris Perdikaris, and Michael A. G. Fern´ andez. Ma- chine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non-invasive 4d flow mri data using physics-informed neural networks.Comput. Methods Appl. Mech. Eng., 358:112623, 2020. doi:10.1016/j.cma.2019.112623

    arXiv 2020

  42. [42]

    Hughes, Daniel R

    Jacob Mathews, Marco Francisquez, John W. Hughes, Daniel R. Hatch, Bei Zhu, and Ben J. Rogers. Uncovering turbulent plasma dynamics via deep learning from partial observations.Phys. Rev. E, 104(2):025205, 2021. doi: 10.1103/PhysRevE.104.025205

    doi:10.1103/physreve.104.025205 2021

  43. [43]

    Spencer, Alexander G

    David Pfau, James S. Spencer, Alexander G. D. G. Matthews, and W. M. C. Foulkes. Ab initio solution of the many-electron schr¨ odinger equation with deep neural networks.Phys. Rev. Res., 2(3):033429, 2020. doi: 10.1103/PhysRevResearch.2.033429

    doi:10.1103/physrevresearch.2.033429 2020

  44. [44]

    Deep potential molecular dynamics: A scalable model with the accuracy of quantum mechanics.Phys

    Linfeng Zhang, Jiequn Han, Han Wang, Roberto Car, and Weinan E. Deep potential molecular dynamics: A scalable model with the accuracy of quantum mechanics.Phys. Rev. Lett., 120:143001, 2018. doi:10.1103/PhysRevLett.120.143001

    doi:10.1103/physrevlett.120.143001 2018

  45. [45]

    Karniadakis

    Liu Yang, Xuhui Meng, and George E. Karniadakis. B-pinns: Bayesian physics-informed neural networks for forward and inverse pde problems with noisy data.J. Comput. Phys., 425:109913, 2021. doi:10.1016/j.jcp.2020.109913

    arXiv 2021

  46. [46]

    Jagtap and George E

    Ameya D. Jagtap and George E. Karniadakis. Extended physics-informed neural networks (xpinns): A generalized space- time domain decomposition based deep learning framework for nonlinear partial differential equations.Commun. Comput. Phys., 28(5):2002–2041, 2020. doi:10.4208/cicp.OA-2020-0164

    doi:10.4208/cicp.oa-2020-0164 2002

  47. [47]

    Karniadakis

    Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George E. Karniadakis. Learning nonlinear operators via deep- onet based on the universal approximation theorem of operators.Nat. Mach. Intell., 3:218–229, 2021. doi:10.1038/s42256- 021-00302-5

    doi:10.1038/s42256- 2021

  48. [48]

    Karniadakis

    Xuhui Meng and George E. Karniadakis. A composite neural network that learns from multi-fidelity data: Application to function approximation and inverse pde problems.J. Comput. Phys., 401:109020, 2020. doi:10.1016/j.jcp.2019.109020

    arXiv 2020

  49. [49]

    Addressing the non-perturbative regime of the quantum anharmonic oscillator by physics-informed neural networks.New J

    Lorenzo Brevi, Antonio Mandarino, and Enrico Prati. Addressing the non-perturbative regime of the quantum anharmonic oscillator by physics-informed neural networks.New J. Phys., 26(10):103015, 2024. doi:10.1088/1367-2630/ad7e2e

    work page doi:10.1088/1367-2630/ad7e2e 2024

  50. [50]

    A tutorial on the use of physics-informed neural networks to compute the spectrum of quantum systems.Technologies, 12(10):174, 2024

    Lorenzo Brevi, Antonio Mandarino, and Enrico Prati. A tutorial on the use of physics-informed neural networks to compute the spectrum of quantum systems.Technologies, 12(10):174, 2024. doi:10.3390/technologies12100174

    doi:10.3390/technologies12100174 2024

  51. [51]

    Yongji Wang, Mehdi Bennani, James Martens, S´ ebastien Racani` ere, Sam Blackwell, Alex Matthews, Stanislav Nikolov, Gonzalo Cao-Labora, Daniel S. Park, Martin Arjovsky, Daniel Worrall, Chongli Qin, Ferran Alet, Borislav Kozlovskii, Nenad Tomaˇ sev, Alex Davies, Pushmeet Kohli, Tristan Buckmaster, Bogdan Georgiev, Javier G´ omez-Serrano, Ray Jiang, and Ch...

    arXiv 2025

  52. [52]

    Physics-informed neural networks viewpoint for solving the dyson–schwinger equations of quantum electrodynamics.SciPost Phys

    Rodrigo Carmo Terin. Physics-informed neural networks viewpoint for solving the dyson–schwinger equations of quantum electrodynamics.SciPost Phys. Core, 8:054, 2025. doi:10.21468/SciPostPhysCore.8.3.054. URLhttps://scipost.org/ 10.21468/SciPostPhysCore.8.3.054

    work page doi:10.21468/scipostphyscore.8.3.054 2025

  53. [53]

    The GINN framework: a stochastic QED correspondence for stability and chaos in deep neural networks

    Rodrigo Carmo Terin. The GINN framework: a stochastic QED correspondence for stability and chaos in deep neural networks. 8 2025

    2025

  54. [54]

    An extrapolation method for a volterra integral equation with weakly singular kernel

    Pedro Lima and Teresa Diogo. An extrapolation method for a volterra integral equation with weakly singular kernel. Applied Numerical Mathematics, 24(2):131–148, 1997. ISSN 0168-9274. doi:https://doi.org/10.1016/S0168-9274(97)00016-

    work page doi:10.1016/s0168-9274(97)00016- 1997

  55. [55]

    Rane, and Kshitij Patil

    Gobinda Rakshit, Akshay S. Rane, and Kshitij Patil. Richardson extrapolation for the iterated galerkin solution of urysohn integral equations with green’s kernels.International Journal of Computer Mathematics, 99(8):1538–1556, 2022. doi:10.1080/00207160.2021.1986215. URLhttps://doi.org/10.1080/00207160.2021.1986215

    arXiv 2022

  56. [56]

    Kenneth G. Wilson. Confinement of quarks.Phys. Rev. D, 10:2445–2459, Oct 1974. doi:10.1103/PhysRevD.10.2445. URL https://link.aps.org/doi/10.1103/PhysRevD.10.2445

    doi:10.1103/physrevd.10.2445 1974

  57. [57]

    John B. Kogut. An introduction to lattice gauge theory and spin systems.Rev. Mod. Phys., 51:659–713, Oct 1979. doi:10.1103/RevModPhys.51.659. URLhttps://link.aps.org/doi/10.1103/RevModPhys.51.659

    doi:10.1103/revmodphys.51.659 1979

  58. [58]

    Monte carlo study of quantized su(2) gauge theory.Phys

    Michael Creutz. Monte carlo study of quantized su(2) gauge theory.Phys. Rev. D, 21:2308–2315, Apr 1980. doi: 10.1103/PhysRevD.21.2308. URLhttps://link.aps.org/doi/10.1103/PhysRevD.21.2308

    doi:10.1103/physrevd.21.2308 1980

  59. [59]

    Dee, Ehsan Khatami, and Steven Johnston

    Shaozhi Li, Philip M. Dee, Ehsan Khatami, and Steven Johnston. Accelerating lattice quantum monte carlo simula- tions using artificial neural networks: Application to the holstein model.Phys. Rev. B, 100:020302, Jul 2019. doi: 10.1103/PhysRevB.100.020302. URLhttps://link.aps.org/doi/10.1103/PhysRevB.100.020302

    work page doi:10.1103/physrevb.100.020302 2019

  60. [60]

    Second International Conference on the Numerical Solution of Volterra and Delay Equations

    URLhttps://www.sciencedirect.com/science/article/pii/S0168927497000160. Second International Conference on the Numerical Solution of Volterra and Delay Equations

  61. [66]

    E. E. Salpeter and H. A. Bethe. A relativistic equation for bound-state problems.Phys. Rev., 84:1232–1242, Dec 1951. doi:10.1103/PhysRev.84.1232. URLhttps://link.aps.org/doi/10.1103/PhysRev.84.1232

    doi:10.1103/physrev.84.1232 1951

  62. [1980]

    doi:10.1103/PhysRevD.22.2542

    doi:10.1103/physrevd.22.2542

  63. [1989]

    doi:10.1103/PhysRevD.40.3474

    doi:10.1103/physrevd.40.3474

  64. [1998]

    doi:10.1142/S0217732398001119

    work page doi:10.1142/s0217732398001119

  65. [1999]

    doi:10.1103/PhysRevC.60.055214

    doi:10.1103/physrevc.60.055214

  66. [2019]

    doi:10.1016/j.jcp.2018.10.045. 9

    doi:10.1016/j.jcp.2018.10.045 2018