Pith. sign in

REVIEW 2 major objections 4 minor 23 references

Checks on QED and strong-isospin breaking corrections to $a_{\mu}^{\mathrm{HVP}}$

T0 review · 2 major / 4 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read This paper reports cross-checks establishing that the QED and strong-isospin corrections to the muon magnetic moment from the collaboration's 2020 lattice computation are reproduced by direct simulations at physical quark masses.

desk verdict A useful internal audit of BMW's QED/SIB corrections, but the valence-QED 'verification' is a single-configuration check that is too weak to carry the claim. read the letter →

arxiv 2502.07572 v1 pith:GKJHTNRD submitted 2025-02-11 hep-lat

Andreas Risch This is my paper · ORCID
classification hep-lat
keywords muong-2hadronicvacuumpolarizationlatticeQCDQEDcorrectionsstrongisospinbreakingstaggeredfermionslow-modeaveragingdecomposition
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

With the muon g-2 experiment now at 0.20 ppm precision, the QED and strong-isospin breaking corrections to $a_\mu^{\mathrm{HVP}}$, the hadronic vacuum polarization contribution to the muon's magnetic moment, must be controlled tightly, and this paper reports a series of cross-checks on those corrections as published in the collaboration's 2020 lattice computation. The claim, stated in the conclusion, is that the earlier results are verified: direct computations at physical quark masses, enabled by low-mode averaging and frequency-splitting estimators, reproduce the previous chiral-extrapolation values for the strong-isospin, valence-QED, and disconnected contributions, and the isospin decomposition scheme itself agrees with two independent schemes when applied to kaon masses. The point of the verification is that the new methods are a genuinely different route to the same numbers, so the agreement is evidence that these electromagnetic and quark-mass corrections are physical results rather than artifacts of the extrapolation procedure.

What carries the argument

The carrying object is the isospin-breaking decomposition of section 4: with the physical point parametrized by $(\hat{M}w_0, M_{ss}w_0, \Delta M^2 w_0^2, \alpha)$, any observable splits as $[O]_{phys} = [O]_{iso} + [O]_{sib} + [O]_{qed}$, where the strong-isospin and electromagnetic pieces are obtained by turning off $\Delta M^2$ and the electric charge. The corrections to $a_\mu^{\mathrm{light}}$ are expressed through the derivatives $[a_\mu^{\mathrm{light}}]'_m = m_l\,\partial[a_\mu^{\mathrm{light}}]/\partial\delta m\,|_{\delta m=0}$ and $[a_\mu^{\mathrm{light}}]''_{20} = \tfrac{1}{2}\,\partial^2[a_\mu^{\mathrm{light}}]/\partial e_v^2\,|_{e_v=0}$. The verification is carried by replacing the per-configuration chiral extrapolations in the valence mass $\kappa$ with direct low-mode-averaged computations at $\kappa = 1$, and by replacing the discrete mass derivative with an exact derivative combined with a frequency-splitting estimator; a three-scheme comparison of the kaon masses (the paper's scheme, the GRS scheme, and a Cottingham-formula decomposition) tests the scheme ambiguity of the decomposition itself.

What would settle it

Compute the valence-QED derivative $[a_\mu^{\mathrm{light}}]''_{20}$ with low-mode averaging on a statistically meaningful number of configurations across several ensembles and compare it with the chirally extrapolated value of reference [2]; a difference exceeding the combined statistical errors at the percent level would show that the single-configuration check in fig. 7 missed a real discrepancy.

Watch

Extended reading notes

Core claim

On its own terms, the central result is the conclusion's sentence: 'We verified the results published in the previous publication [2].' The verification compares two routes to the leading isospin-breaking corrections of $a_\mu^{\mathrm{HVP}}$: the strong-isospin derivative $[a_\mu^{\mathrm{light}}]'_m$ is now evaluated with an exact mass derivative and low-mode averaging at the physical valence mass ($\kappa = 1$) instead of by a linear chiral extrapolation from $\kappa = 3, 5, 7$, and the two routes agree (fig. 6); the valence-QED derivative $[a_\mu^{\mathrm{light}}]''_{20}$ is checked on a single gauge configuration, with the difference between methods compatible with zero (fig. 7); and the disconnected contribution $[a_\mu^{\mathrm{disc}}]'_m$ computed with a frequency-splitting estimator agrees with the previous stochastic discrete-derivative result (fig. 8). The paper also shows, through a three-scheme comparison of kaon masses, that the isospin-breaking decomposition underlying these corrections is not distorted by the choice of separation scheme at the quoted precision.

Load-bearing premise

The load-bearing assumption is that the valence-QED correction is adequately tested by a comparison on a single gauge configuration from one ensemble, which is too weak to catch a discrepancy at the few-percent level that matters at this precision.

Editorial extensions

If this is right

  • The strong-isospin correction $[a_\mu^{\mathrm{light}}]'_m$ computed with an exact mass derivative and low-mode averaging agrees with the chiral-extrapolation result of the 2020 computation, so that correction is confirmed under the new method.
  • The valence-QED correction $[a_\mu^{\mathrm{light}}]''_{20}$ shows a difference compatible with zero between the two methods on the tested configuration, and the disconnected strong-isospin contribution $[a_\mu^{\mathrm{disc}}]'_m$ computed with a frequency-splitting estimator agrees with the earlier discrete-derivative value.
  • The kaon-mass decomposition in the paper's isospin-breaking scheme agrees with the GRS scheme and with a Cottingham-formula decomposition, indicating that the separation-scheme ambiguity is under control at the current level of precision.
  • Autocorrelation times on the two finest ensembles are small enough ($\tau_{int} \lesssim 8$ configurations) that the 48-block jackknife blocking keeps autocorrelation effects under control.
  • Staggered taste violations decrease approximately as $a^4$ at small lattice spacings, supporting the continuum-extrapolation parametrization used in the updated analysis.

Reading between the lines

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

  • Extending the valence-QED cross-check from a single configuration to a statistically meaningful sample across several ensembles would upgrade the verification from a consistency statement into a quantitative test of the 2020 QED correction.
  • The three-scheme agreement is established for kaon masses; applying the same comparison directly to $a_\mu^{\mathrm{HVP}}$ would close the gap between the scheme-consistency check and the observable whose corrections are at issue.
  • The reported reduction in stochastic source vectors by one to two orders of magnitude suggests that a fully direct computation of the valence-QED contribution at physical quark masses across all ensembles is now computationally affordable, which would remove the last chiral extrapolation from the isospin-breaking corrections.
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

2 major / 4 minor

Summary. This conference proceedings paper reports cross-checks of the QED and strong-isospin breaking corrections to the hadronic vacuum polarisation contribution to the muon g-2 used in the BMW collaboration's 2020 computation. The paper presents two new ensembles at a=0.0483 fm, an autocorrelation analysis of the finest ensembles, a study of staggered taste violations, a three-scheme decomposition of the kaon mass, and comparisons between the old chiral-extrapolation-based estimators and new exact-derivative/low-mode-averaging estimators for the leading isospin-breaking corrections to a_mu^light and a_mu^disc. The central claim is that these checks confirm the results published in [2].

Significance. If the checks are fully valid, they strengthen confidence in the 2020 BMW QED/SIB corrections, which are important in the current g-2 context. The genuine strengths are the three-scheme kaon mass decomposition, which uses external GRS and Cottingham results, and the replacement of chiral extrapolation by exact mass derivatives with low-mode averaging for the connected SIB contribution, which is a methodological improvement. However, the valence-QED check in Fig. 7 is a single-configuration comparison, so it is much weaker than the other checks and cannot carry the weight of the conclusion's 'verified' claim. The paper should be transparent about the difference between internal consistency checks and independent verification.

major comments (2)
  1. [Section 5 (Fig. 7)] The sole check of the valence-QED contribution [a_light]''20 is a comparison between the old chiral-extrapolation estimator and the new κ=1 low-mode-averaging estimator on a single gauge configuration from one a=0.0787 fm ensemble. A difference compatible with zero on N=1 cannot constrain the ensemble-averaged bias of the previous chiral extrapolation; it only shows that the two procedures agree on that particular gauge-field realization. No error bar is shown for the plotted difference and no statement is given of the size of discrepancy this test could have detected. To support the Sec. 6 claim that the results of [2] are 'verified', the comparison should be repeated on many configurations and ideally at least one additional lattice spacing, or the authors should provide a quantitative sensitivity bound so the reader can judge what the test actually excludes.
  2. [Section 5 (Figs. 6 and 8)] The SIB comparisons are internal consistency checks rather than independent verification. In Fig. 6 the comparison is made at several lattice spacings, which is a strength, but both the extrapolated and κ=1 results are computed on the same ensembles with the same gauge configurations and within the same analysis framework; in Fig. 8 the two estimators are evaluated on the same data. These comparisons are valuable and can catch methodological bias, but they cannot exclude systematic effects common to both, such as scale setting, finite-volume effects, or shared chiral/continuum input. The text should state this limitation explicitly; the conclusion's wording 'We verified the results published in [2]' overstates the evidence unless this caveat is added.
minor comments (4)
  1. [Fig. 7] The 'diff.' curve would be far more informative with an error band or a table of the difference with its statistical error; as printed, the reader cannot assess whether the two methods agree within noise or merely because the comparison is underpowered.
  2. [Section 4] The sentence 'We define the physical value of M-hat^2 by neglecting next-to-leading order effects' should clarify that this is a scheme choice rather than a numerical approximation, since the scheme dependence is precisely what the kaon-mass comparison in Fig. 4 is intended to probe.
  3. [Section 2] The statement 'This value is within one percent of the most recent lattice average from FLAG [9–12]' should identify which FLAG review is meant; [9] is the FLAG 2021 review, so 'most recent' should be updated to the latest published version.
  4. [Abstract] The phrase 'BMW's collaboration 2020 computation' is awkward; consider 'the 2020 BMW collaboration computation'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the cross-checks compare independent estimators and external scheme decompositions, and the underpowered single-configuration valence-QED check is a sensitivity limitation, not a definitional reduction.

full rationale

The paper's central assertion, 'We verified the results published in the previous publication [2]', is supported by comparisons between the old chiral-extrapolation estimators used in [2] and new direct computational methods (exact mass derivative, low-mode averaging at kappa=1, frequency-splitting estimators) shown in Figs. 6-8. These new computations are not fitted to the old values; they are independent procedures targeting the same physical quantities, so no equation reduces to another by construction. The kaon-mass decomposition comparison (Fig. 4) uses external GRS and Cottingham-formula results from Refs. [17-20], providing independent anchors. The valence-QED check in Section 5 is indeed weak: it is performed 'on a single gauge configuration from one of the a=0.0787 fm ensembles' (Fig. 7), and a difference compatible with zero on one configuration cannot exclude a few-percent chiral-extrapolation bias at the ensemble level. However, this is a statistical-power and sensitivity limitation, not a circular step: the two methods being compared are not the same quantity by definition, and agreement is not enforced by fitting. Self-citations to [2] and [3] are natural for a same-collaboration follow-up and are not load-bearing arguments in the derivations; the comparisons carry independent content. The paper is not claiming an external independent determination, but the checks are not circular in the sense of pattern 1-6.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new free parameters and no invented entities. It depends on the BMW isospin decomposition scheme, on physical inputs imported from earlier BMW publications, and on standard lattice QCD modeling assumptions about staggered taste violations, autocorrelations, and estimator unbiasedness.

assumptions (6)
  • domain assumption The decomposition of any observable into isospin-symmetric, strong-isospin-breaking, and QED pieces via the physical points in eq. (1) is a valid definition at the required precision.
    Section 4 defines the QCD+QED, QCD, and QCDiso physical points and states that separation scheme ambiguities exist; the paper checks scheme dependence only for kaon masses, not for a_mu^HVP.
  • domain assumption Neglecting NLO chiral corrections so that [Mhat]phys = M_pi0 = 134.9768(5) MeV does not bias the decomposition.
    Section 4 explicitly defines [Mhat]phys by neglecting NLO effects; this choice affects all decomposed observables.
  • domain assumption Staggered taste-violation corrections scale as a^4 (or alpha_s^3 a^2) and can be controlled in the continuum extrapolation using Delta_KS(A).
    Section 3 reports the scaling and states that Delta_KS(A) is used in the continuum extrapolation in [3]; a wrong scaling form would bias the extrapolation.
  • domain assumption The exact mass derivative and low-mode averaging at kappa=1 give unbiased estimates of the mass derivative of a_mu^light, comparable to the chiral extrapolation.
    Section 5 replaces the old chiral extrapolation with these techniques; the verification assumes their correctness.
  • domain assumption Integrated autocorrelation times from Q^2, energy density, and Omega correlators on the finest ensemble bound autocorrelations of all observables used in the analysis.
    Section 3 computes tau_int for these observables and argues the block length is sufficiently large; this assumes they are representative of every quantity entering the HVP calculation.
  • domain assumption The physical point values imported from [2], namely [Delta M^2]phys = 13170(420) MeV^2, [Mss]phys = 689.89(49) MeV, and [w0]phys = 0.17245(51) fm, are correct.
    Section 4 imports these values from earlier BMW work; any error in these inputs would shift the isospin decomposition and the inferred corrections.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Checks on QED and strong-isospin breaking corrections to $a_{\mu}^{\mathrm{HVP}}$." pith.science (2026). https://pith.science/paper/GKJHTNRD

@misc{pith2026250207572,
  author       = {Pith},
  title        = {Pith review of: Checks on QED and strong-isospin breaking corrections to $a_\mu^\mathrmHVP$},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GKJHTNRD}},
  note         = {Machine review of arXiv:2502.07572}
}
read the original abstract

At the current levels of precision reached in the measurement of the muon g-2 by Fermilab, it is essential to control QED and strong isospin breaking corrections to the HVP contribution to the muon g-2. Here we present a number of cross-checks performed on the results for those corrections presented in the BMW's collaboration 2020 computation.

Figures

Figures reproduced from arXiv: 2502.07572 by the authors.

Figure 1
Figure 1. Landscape of ensembles. Left: The horizontal and vertical axes are the squared pseudo-scalar masses 𝑀2 ll and 𝑀2 ss, which serve as proxies for the quark mass 𝑚l and 𝑚s , in units of the 𝑤0-scale normalised with the corresponding physical values. The isospin symmetric physical point is shown in black. The uncertainties of this point result from the determination of 𝑀ll, 𝑀ss and 𝑤0 in physical units. Different colour… view at source ↗
Figure 2
Figure 2. Normalised autocorrelation functions 𝜌(𝜏) and integrated autocorrelation times 𝜏 int in units of configurations on one of the two ensembles with finest lattice spacing (𝛽 = 4.1479, 𝑎 = 0.0483 fm). Left: Energy density 𝐸(𝑤 2 0 ) and the squared topological charge 𝑄 2 (𝑤 2 0 ) evaluated at a gradient flow time of 𝑡fl = 𝑤 2 0 . Right: ΩVI correlation function with point and smeared sources evaluated at 𝑡 = 1.5 fm [PIT… view at source ↗
Figure 3
Figure 3. Taste violations as a function of the lattice spacing 𝑎 for axial-vector 𝐴 and tensor 𝑇 tastes. In the staggered fermion discretisation lattice artefacts are related to the taste symmetry vi￾olation of staggered fermions. As a result, pseudo-scalar mesons on the lattice become heavier than in the continuum and their masses depend on their taste quantum numbers. The masses of the pions 𝑀2 𝜋 (𝜉) = 𝑀2 ll + ΔKS(𝜉) are c… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Decomposition of the neutral and charged kaon masses in three different schemes: (𝑀ˆ 2 , Δ𝑀2 , 𝑀2 ss, 𝑤0, 𝛼) scheme, GRS scheme and Cottingham formula based scheme. -21 -20 -19 -18 -17 -16 -15 e.m. isospin breaking [alight µ ]’’20 extrapolation measurement 2.5 3.0 3.5 …
Figure 5
Figure 5. Figure 5: Extrapolation procedure for [𝑎 light 𝜇 ] ′ 𝑚 and [𝑎 light 𝜇 ] ′′ 20 for 𝛽 = 3.7000 in the previous work [2]. 5. Verification of isospin breaking contributions In the most recent work [3] only the isospin symmetric contribution to 𝑎 light 𝜇 and 𝑎 disc 𝜇 were updated, wh…
Figure 7
Figure 7. Figure 7: [𝑎 light 𝜇 ] ′′ 20 computed on a single configura￾tion at 𝑎 = 0.0787 fm as a function of the upper limit of integration 𝑡𝑐. Comparison of computations based on a chiral extrapolation and based on low-mode av￾eraging applied at 𝜅 = 1. 𝑚𝑙 is fixed to the sea quark mass, …
Figure 8
Figure 8. Figure 8: Comparison between the standard stochastic estimator and frequency-splitting estimator of the strong isospin breaking contribution [𝑎 disc 𝜇 ] ′ 𝑚. In the previous setup [2] the SIB contribution to 𝑎 disc 𝜇 was computed performing a discrete derivative with respect to …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 3 canonical work pages

  1. [2]

    Borsanyi, Z

    S. Borsanyi, Z. Fodor, J.N. Guenther, C. Hoelbling, S.D. Katz, L. Lellouch et al.,Leading hadronic contribution to the muon magnetic moment from lattice QCD, Nature593 (2021) 51 [2002.12347]

    arXiv 2021

  2. [1]

    Muon g-2 collaboration, Measurement of the Positive Muon Anomalous Magnetic Moment to 0.20 ppm, Phys. Rev. Lett.131 (2023) 161802 [2308.06230]

    arXiv 2023

  3. [3]

    Boccaletti, S

    A. Boccaletti, S. Borsanyi, M. Davier, Z. Fodor, F. Frech, A. Gerardin et al.,High precision calculation of the hadronic vacuum polarisation contribution to the muon anomaly, 2407.10913. 8 Checks on QED and strong-isospin breaking corrections to𝑎HVP 𝜇 Andreas Risch

    arXiv

  4. [4]

    Frech, K

    F. Frech, K. Szabo and F. Stokes,Reconstruction of the vector meson propagator using a generalized eigenvalue problem,PoSLATTICE2024(2025) 133

    work page 2025

  5. [5]

    Luscherand P

    M. Luscherand P. Weisz,On-shell improvedlattice gauge theories,Commun. Math.Phys.98 (1985) 433

    work page 1985

  6. [6]

    Morningstar and M.J

    C. Morningstar and M.J. Peardon,Analytic smearing of SU(3) link variables in lattice QCD, Phys. Rev. D69(2004) 054501 [hep-lat/0311018]

    arXiv 2004

  7. [7]

    Borsányi, S

    S. Borsányi, S. Dürr, Z. Fodor, C. Hoelbling, S.D. Katz, S. Krieg et al.,High-precision scale setting in lattice QCD,JHEP 09(2012) 010 [1203.4469]

    arXiv 2012

  8. [8]

    McNeile, C.T.H

    C. McNeile, C.T.H. Davies, E. Follana, K. Hornbostel and G.P. Lepage,High-Precision c and b Masses, and QCD Coupling from Current-Current Correlators in Lattice and Continuum QCD, Phys. Rev. D82 (2010) 034512 [1004.4285]

    arXiv 2010

Show all 23 references
  1. [9]

    Flavour Lattice A veraging Group (FLAG)collaboration, FLAG Review 2021, Eur. Phys. J. C82(2022) 869 [2111.09849]

    2022 arXiv

  2. [10]

    Fermilab Lattice, MILC, TUMQCD collaboration,Up-, down-, strange-, charm-, and bottom-quark masses from four-flavor lattice QCD, Phys. Rev. D98(2018) 054517 [1802.04248]

    2018 arXiv

  3. [11]

    European Twisted Mass collaboration, Up, down, strange and charm quark masses with N𝑓 = 2+1+1 twisted mass lattice QCD,Nucl. Phys. B887 (2014) 19 [1403.4504]

    2014 arXiv

  4. [12]

    Chakraborty, C.T.H

    B. Chakraborty, C.T.H. Davies, B. Galloway, P. Knecht, J. Koponen, G.C. Donald et al., High-precision quark masses and QCD coupling from𝑛𝑓 = 4 lattice QCD,Phys. Rev. D91 (2015) 054508 [1408.4169]

    2015 arXiv

  5. [13]

    ALPHAcollaboration, Critical slowing down and error analysis in lattice QCD simulations, Nucl. Phys. B845 (2011) 93 [1009.5228]

    2011 arXiv

  6. [14]

    Sheikholeslami and R

    B. Sheikholeslami and R. Wohlert,Improved Continuum Limit Lattice Action for QCD with Wilson Fermions,Nucl. Phys. B259 (1985) 572

    1985

  7. [15]

    Joswig, S

    F. Joswig, S. Kuberski, J.T. Kuhlmann and J. Neuendorf,pyerrors: A python framework for error analysis of Monte Carlo data,Comput. Phys. Commun.288 (2023) 108750 [2209.14371]

    2023 arXiv

  8. [16]

    Bijnens and N

    J. Bijnens and N. Danielsson,Electromagnetic Corrections in Partially Quenched Chiral Perturbation Theory,Phys. Rev. D75 (2007) 014505 [hep-lat/0610127]

    2007 arXiv

  9. [17]

    Gasser, A

    J. Gasser, A. Rusetsky and I. Scimemi,Electromagnetic corrections in hadronic processes, Eur. Phys. J. C32(2003) 97 [hep-ph/0305260]

    2003 arXiv

  10. [18]

    Di Carlo, D

    M. Di Carlo, D. Giusti, V. Lubicz, G. Martinelli, C.T. Sachrajda, F. Sanfilippo et al., Light-meson leptonic decay rates in lattice QCD+QED,Phys. Rev. D100(2019) 034514 [1904.08731]. 9 Checks on QED and strong-isospin breaking corrections to𝑎HVP 𝜇 Andreas Risch

    2019 arXiv

  11. [19]

    Stamen, D

    D. Stamen, D. Hariharan, M. Hoferichter, B. Kubis and P. Stoffer,Kaon electromagnetic form factors in dispersion theory, Eur. Phys. J. C82(2022) 432 [2202.11106]

    2022 arXiv

  12. [20]

    Giusti, V

    D. Giusti, V. Lubicz, C. Tarantino, G. Martinelli, F. Sanfilippo, S. Simula et al.,Leading isospin-breaking corrections to pion, kaon and charmed-meson masses with Twisted-Mass fermions, Phys. Rev. D95 (2017) 114504 [1704.06561]

    2017 arXiv

  13. [21]

    Giusti, P

    L. Giusti, P. Hernandez, M. Laine, P. Weisz and H. Wittig,Low-energy couplings of QCD from current correlators near the chiral limit, JHEP04(2004) 013 [hep-lat/0402002]

    2004 arXiv

  14. [22]

    DeGrand and S

    T.A. DeGrand and S. Schaefer,Improving meson two point functions in lattice QCD, Comput. Phys. Commun.159(2004) 185 [hep-lat/0401011]

    2004 arXiv

  15. [23]

    Giusti, T

    L. Giusti, T. Harris, A. Nada and S. Schaefer,Frequency-splitting estimators of single-propagator traces,Eur. Phys. J. C79(2019) 586 [1903.10447]. 10

    2019 arXiv

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.