{"id":"51a0ceb8-0699-44bc-9e58-3338a2ab03b0","arxiv_id":"2507.20910","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Nucleon electromagnetic radii and magnetic moments are extracted from lattice QCD in the continuum limit directly at the physical pion mass, with disconnected contributions included.","lead":"This paper computes the electric and magnetic form factors of the proton and neutron using lattice QCD at three lattice spacings with physical quark masses, and takes the continuum limit without any chiral extrapolation. A generalist should read it because it provides a first-principles benchmark for the proton radius and magnetic moments that can be compared against scattering experiments and the proton radius puzzle.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The final AIC model average is dominated by one z-exp(a^2) fit; dipole and no-a^2 fits carry near-zero weight yet differ by ~0.1 fm^2 in the proton radius, so the quoted model systematic likely understates model dependence.","rationale":"Read in good faith: the paper is a careful, state-of-the-art lattice calculation with a detailed excited-state analysis, model-averaged ground-state extraction, disconnected contributions, and comparison to experiment. The central claim—a direct continuum limit at physical pion mass with disconnected contributions—is plausible, and the results agree with PDG values within errors. The load-bearing weakness is not the lattice technology but the final model-averaging step: with three lattice spacings and two Q^2 parameterizations, the AIC weights collapse onto a single model, so the quoted model systematic is essentially the within-model Q^2_cut scatter. The reader's weakest assumption (linear a^2 from only three spacings) is part of this; I would sharpen it to the fragility of the AIC weighting, since the dipole and no-a^2 fits are viable but assigned ~0 weight. This does not invalidate the calculation—the results are correctable—but it means the precision of the headline radii and moments is less robust than stated. The internal typo in mu_p's systematic error (Eq. 46 vs Table XIV) and the single-error neutron electric radius are minor and do not affect this assessment. A conditional acceptance with a request to report a flat-model spread or a broader BMA is appropriate.","tokens_in":43888,"tokens_out":9282,"duration_ms":115354,"concrete_test":"Recompute the final model-averaged <r_E^2>^p using a flat (equal-weight) average over all 12 fits in Table XIV with reduced chi^2 < 2, including the dipole and no-a^2 z-exp entries. If the flat average differs from 0.739 fm^2 by more than the quoted systematic error of 0.039 fm^2 (a simple average of the listed values gives ~0.69 fm^2), then the AIC weights are suppressing genuine model dependence and the model systematic should be enlarged accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—the continuum-limit radii and moments in Table XIV/Fig. 19—rests on the AIC model average of Eq. (45). That average is effectively single-model: for <r_E^2>^p, the z-expansion with linear a^2 dependence at Q^2_cut=0.85 GeV^2 receives 82% of the weight, while the dipole-with-a^2 and z-expansion-without-a^2 fits receive ~0% (Table XIV). The excluded fits are not poor fits: the dipole continuum result is <r_E^2>^p = 0.650(52) fm^2 (Table VIII) and the no-a^2 z-exp values are 0.636–0.667 fm^2, whereas the adopted z-exp(a^2) gives 0.747–0.816 fm^2. The ~0.1 fm^2 offset is comparable to the statistical error and larger than the quoted model systematic of 0.039 fm^2. AIC weights become sharply peaked because chi^2 differences among acceptable fits are modest, and with only three lattice spacings the free a^2 slope in c_1(a^2) is weakly constrained; the weighting then assigns essentially zero probability to alternatives that are not excluded by the data. Thus the systematic error from Eq. (45) reflects scatter within a single model family across Q^2_cut choices and does not cover the dipole/z-exp model dependence, so the quoted precision of e.g. the proton charge radius is not robust.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a lattice QCD calculation of the proton and neutron electromagnetic form factors using three Nf=2+1+1 twisted-mass clover ensembles at the physical pion mass with lattice spacings a=0.080, 0.068, and 0.057 fm. The authors perform a detailed multi-state excited-state analysis of the connected three-point functions, include disconnected isoscalar contributions computed with stochastic techniques, renormalize the local current in RI'/MOM, and extract radii and magnetic moments from dipole, z-expansion, and Galster-like fits to the Q^2 dependence, followed by a linear a^2 continuum extrapolation. The final results, including proton and neutron electric and magnetic radii and magnetic moments, as well as Zemach and Friar radii, agree with experiment within the quoted errors.","tokens_in":44313,"tokens_out":8195,"duration_ms":99559,"significance":"If the quoted systematic errors are robust, this is a valuable step: it is one of the first attempts to take the continuum limit directly at the physical pion mass for nucleon electromagnetic form factors, with a careful treatment of excited states and disconnected contributions. The main strengths are the physical-point ensembles, the extensive sink-source separation analysis, the stochastic disconnected-loop methodology, and the transparent comparisons with the Mainz collaboration and with experimental data. However, the central error budget is not fully established because the final model average and the three-point continuum extrapolation rest on assumptions that are only weakly tested by the data.","major_comments":[{"comment":"The final quoted systematic error for <r_E^2>^p is dominated by an AIC average in which a single z-expansion(a^2) fit with Q^2_cut=0.85 GeV^2 receives 82% of the weight, while the dipole(a^2) and z-expansion-without-a^2 fits receive essentially zero weight. The continuum values of those down-weighted fits are 0.650(52) fm^2 and 0.636-0.667 fm^2, respectively, compared with 0.747-0.816 fm^2 for the adopted z-exp(a^2) fits. The ~0.1 fm^2 offset is comparable to the statistical error and is several times larger than the quoted 0.039 fm^2 model systematic. These fits are not excluded by their reduced chi^2 values, so the AIC weighting of Eq. (45) does not cover the ansatz dependence. I recommend adding an explicit systematic from the spread over fit families, for example an envelope over all fits with acceptable chi^2 or a prior over model families, and checking the effect on the final errors.","section":"Sec. V.A, Table XIV, Eq. (45)"},{"comment":"The AIC weights used in Table XIV are computed with log(w_j) = -chi^2_j/2 + N_dof,j, where N_dof = N_data - N_params. Because Q^2_cut changes N_data, this formula adds a term linear in N_data to the log-weight. For the differences in data counts between Q^2_cut=0.4 and 1.0 GeV^2, this term overwhelmingly favors the largest Q^2_cut even when the fit quality per degree of freedom is comparable. The near-100% probabilities for single Q^2_cut values in Table XIV therefore partly reflect dataset-size effects rather than relative support of the fits. A fair model average over Q^2_cut should either use a common Q^2 range or a likelihood normalization that does not reward extra data points exponentially.","section":"Sec. IV.A.1, Eq. (27)"},{"comment":"The continuum extrapolation assumes a linear a^2 dependence and uses only three lattice spacings covering 0.057-0.080 fm. The one-step and two-step comparisons in Sec. V.D test the fitting procedure, not the functional form of the cutoff dependence. A curvature term such as a^4, or a residual O(a) effect, would shift the extrapolated radii and moments by an amount not included in the quoted systematics. I ask the authors to estimate this sensitivity, for example by adding an a^4 term, by dropping the coarsest ensemble, or by quoting a cutoff-scale systematic; without this, the claimed continuum-limit precision is underdetermined.","section":"Sec. V.A, Eqs. (35) and (41)"},{"comment":"The neutron electric radius is extracted from a single Galster-like fit with no a^2 dependence and is quoted as <r_E^2>^n = -0.147(48) fm^2 with no systematic error in the abstract. The statement that the data cannot resolve a lattice-spacing dependence does not imply that the discretization systematic is zero; a cutoff effect at the level of the other observables would directly bias this central quantity. I request a conservative estimate of the continuum systematic, for example from fits to individual ensembles or from a fit with an a^2 slope, and that this be included in the final error.","section":"Sec. V.F and Table XIV"}],"minor_comments":[{"comment":"The proton magnetic moment is quoted in Eq. (46) as mu_p = 2.849(92)(25), while the abstract, Table XIV, and the surrounding text give 2.849(92)(52). This is an inconsistency that must be corrected.","section":"Sec. VI.A, Eq. (46)"},{"comment":"There is a typo in 'the the heavy quark parameters' in the paragraph after Table I.","section":"Sec. III.B"},{"comment":"The ensemble label 'cB211.72.64' omits the zero in 'cB211.072.64' used elsewhere; please make the labels consistent.","section":"Table VII"},{"comment":"The disconnected contributions are extracted from a single fit-range combination, with no model averaging over fit ranges. The text states that varying the fit ranges gives effects suppressed by the larger statistical errors, but the error budget should state explicitly that this contribution to the disconnected systematic is not included in Eq. (45).","section":"Sec. IV.B and Table VII"},{"comment":"The horizontal axis labels for the proton electric and magnetic radii are written as 'r2_E^p [fm]' and 'r2_M^p [fm]' but the plotted quantities are the square roots of the mean-square radii. Please relabel the axes to avoid ambiguity.","section":"Fig. 19"}],"recommendation":"major_revision","confidential_remarks":"The calculation is substantial and the central physics results are plausible, but the manuscript's headline precision is not yet supported: the AIC model-average and the three-spacing linear continuum extrapolation need either stronger justification or an enlarged systematic error. The issues are fixable within the scope of the paper, so I do not recommend rejection. I would also ask the editor to ensure the quoted neutron electric radius is not presented as a continuum-limit result without an explicit cutoff systematic."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a real step forward. Alexandrou et al. give the first continuum-limit extraction of the proton and neutron electromagnetic form factors, radii, and magnetic moments directly at the physical pion mass, using three N_f=2+1+1 twisted-mass ensembles with a=0.080, 0.068, and 0.057 fm, including disconnected contributions. The excited-state analysis is unusually careful: multiple source-sink separations, separate excited-state energies for two- and three-point functions, and a check that the nucleon-pion state is handled. The disconnected part uses deflation, dilution, and hierarchical probing, all described in enough detail to be reproduced. The final numbers agree with experiment and with the Mainz continuum extrapolation. That is a solid outcome.\n\nThe soft spot is the model-averaged systematic error. The AIC weighting in Table XIV is sharply peaked: for the proton charge radius, the z-expansion with a linear a^2 dependence at Q^2_cut=0.85 GeV^2 receives 82% of the weight, while the dipole(a^2) and z-expansion without a^2 carry essentially zero weight. Those excluded fits are not bad fits; they give <r_E^2>^p around 0.65 fm^2 versus 0.75 fm^2 for the adopted fit. The quoted systematic of 0.039 fm^2 from Eq. (45) reflects scatter within one model family, not the spread between model families. That spread is comparable to the statistical error and larger than the systematic. So the stated precision on the radii is not robust. A revision should either include the dipole and no-a^2 fits in a more conservative model average, or quote a systematic from the full spread. This is correctable, but it matters for anyone using these numbers.\n\nTwo smaller points. The neutron electric radius is quoted as -0.147(48) fm^2 with no systematic error, although the abstract promises both errors in all cases. That should be fixed. And the continuum limit is a straight line in a^2 through three points; that is the best one can do with three spacings, but it should be flagged as a caveat when comparing precision to experiment.\n\nOverall, this is a serious calculation and deserves a proper referee. I would send it out, and ask the authors to rework the model-averaging systematics before publication.","headline":"A serious lattice calculation that likely gives the first physical-pion continuum limit for nucleon EM form factors, but the model-averaged systematic error understates the spread between fit families and needs reworking.","tokens_in":44800,"tokens_out":4991,"would_cite":true,"duration_ms":51978,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81T25","81V05"],"pacs":["12.38.Gc","13.40.Gp","14.20.Dh"],"model":"deepseek-v4-flash","headline":"This paper computes proton and neutron electromagnetic form factors directly at the physical pion mass in the continuum limit of lattice QCD, with disconnected contributions included, yielding radii and magnetic moments for both nucleons.","keywords":["lattice QCD","nucleon electromagnetic form factors","continuum limit","physical pion mass","disconnected contributions","proton radius","neutron charge radius","z-expansion"],"falsifier":"Compute the same observables on a fourth, finer ensemble, for example at $a\\approx0.04$ fm, or refit the present data with an $a^4$ term added to the $z$-expansion coefficients; if $\\sqrt{\\langle r_E^2\\rangle^p}$ shifts by more than the quoted $0.023$ fm systematic error, the linear-in-$a^2$ extrapolation is the fragile step.","tokens_in":1963,"feed_emoji":"⚛️","tokens_out":2252,"duration_ms":150153,"temperature":0.7,"pith_summary":"The paper computes the electromagnetic form factors of the proton and neutron from lattice QCD, the first-principles discretization of the strong force, using three ensembles of twisted-mass clover-improved fermions (a standard lattice discretization of quarks) with quark masses set to their physical values and lattice spacings $a=0.080$, $0.068$, and $0.057$ fm. Because all three ensembles sit at the physical pion mass, the continuum limit can be taken directly at that mass, without a chiral extrapolation. The calculation includes disconnected quark-loop contributions to the isoscalar current, which are often neglected, and controls excited-state contamination with multi-state fits over sink-source separations up to 1.5 fm. From the continuum-limit form factors the paper extracts electric and magnetic radii, magnetic moments, and the Zemach and Friar radii, the higher moments that enter hydrogen spectroscopy. If these results hold, lattice QCD becomes a genuinely first-principles check on electron-scattering and muonic-hydrogen measurements of nucleon size.","feed_headline":"Lattice QCD pins down proton and neutron form factors in the continuum","feed_subtitle":"Three lattice spacings and full disconnected diagrams give proton charge radius 0.860 fm and magnetic moment 2.849, matching experiment.","key_machinery":"The load-bearing mechanism is a one-step continuum extrapolation: for each ensemble the Sachs form factors are extracted at many values of $Q^2$ by multi-state fits to the two- and three-point nucleon correlation functions, and then all three ensembles are fit together to a $z$-expansion whose coefficients depend linearly on $a^2$, so that radii, moments, and continuum limit come out of one combined fit. Disconnected quark-loop contributions are computed with the local vector current renormalized in the RI'/MOM scheme; the paper confirms that the singlet and nonsinglet renormalization factors coincide, allowing the disconnected and connected pieces to be added safely. Excited-state systematics are handled by allowing different excited-state energies in the two- and three-point functions and by Akaike-weighted model averaging over fit ranges. Dipole and Galster-like parameterizations are used as cross-checks, and the one-step and two-step continuum extrapolations agree.","core_discovery":"The central claim is that proton and neutron electromagnetic form factors can be obtained in the continuum limit directly at the physical pion mass. With $N_f=2+1+1$ twisted-mass clover-improved fermions, the paper obtains the electric and magnetic Sachs form factors $G_E(Q^2)$ and $G_M(Q^2)$ for both nucleons on three ensembles at $a=0.080$, $0.068$, and $0.057$ fm, including disconnected contributions renormalized with the flavor-singlet vector current. The final model-averaged results are $\\sqrt{\\langle r_E^2\\rangle^p}=0.860(38)(23)$ fm, $\\langle r_E^2\\rangle^n=-0.147(48)$ fm$^2$, $\\sqrt{\\langle r_M^2\\rangle^p}=0.870(53)(15)$ fm, $\\sqrt{\\langle r_M^2\\rangle^n}=0.913(67)(19)$ fm, $\\mu^p=2.849(92)(52)$, and $\\mu^n=-1.819(76)(29)$, with the $z$-expansion fits carrying the largest model-averaging weight. The neutron electric form factor, which is hard to measure directly, comes out more precise than the experimental one, and the computed Zemach and Friar radii agree with other determinations.","pith_inferences":["If the linear-in-$a^2$ assumption is right, a fourth ensemble near $a\\approx0.04$ fm would be the cleanest test; visible curvature there would push every continuum radius outside its quoted error.","The SVD-based use of nonzero sink momenta for the disconnected diagram adds many $Q^2$ points at no extra inversions, a technique that could be applied to other isoscalar nucleon matrix elements such as the axial or scalar charges.","Because the three ensembles have similar physical volumes, finite-volume effects are constrained mainly by comparison with other calculations rather than by the data itself; a dedicated larger-volume study at the finest spacing would quantify them.","Interpreting the systematic error as dominated by model and $Q^2_{\\rm cut}$ choice, the quickest precision gain may come from denser low-$Q^2$ coverage, for example from larger volumes, rather than from more statistics at existing momenta."],"forward_implications":["Lattice QCD results for nucleon form factors can now be compared with electron-scattering and muonic-hydrogen measurements without any chiral extrapolation.","The proton electric radius from first principles, $\\sqrt{\\langle r_E^2\\rangle^p}=0.860(38)(23)$ fm, has errors that cover both the electron-scattering and muonic-hydrogen values, so it provides an independent data point without resolving the proton radius puzzle.","The neutron electric form factor, measured only indirectly in experiment, is obtained more precisely from lattice QCD than from experiment, making lattice QCD the more accurate source for that quantity.","The same continuum-limit form factors yield Zemach and Friar radii, giving lattice-based input for hydrogen hyperfine-splitting and Lamb-shift determinations of the proton radius.","Adding a fourth, finer ensemble or larger physical volumes, as the paper suggests, is the specific next step that would shrink the dominant systematic errors in the radii."],"supporting_citations":[{"why":"provides the lattice spacings and pion masses that set the physical scale of all three ensembles, and the symmetry argument that singlet and nonsinglet vector renormalization factors coincide.","marker":"[16]"},{"why":"earlier form-factor calculation on the coarsest ensemble, extended here to three lattice spacings with a full excited-state analysis.","marker":"[8]"},{"why":"the continuum-extrapolated nucleon form-factor calculation used as the primary comparison for the final radii and moments.","marker":"[6]"},{"why":"supplies the z-expansion parameterization with a-squared-dependent coefficients and the Akaike model-averaging procedure.","marker":"[38]"},{"why":"gives the generalized one-end trick used to compute the disconnected quark-loop contractions.","marker":"[23]"},{"why":"provides hierarchical probing to reduce stochastic noise in the disconnected loop trace.","marker":"[28]"},{"why":"shows that pion-nucleon excited states can contaminate the three-point function, motivating separate excited-state energies in the fits.","marker":"[35]"},{"why":"defines the RI'/MOM scheme used to renormalize the local vector current for the disconnected diagram.","marker":"[30]"},{"why":"gives the accepted experimental values against which the final radii and magnetic moments are compared.","marker":"[43]"}],"fun_headline_variants":["Proton charge radius 0.860 fm from continuum limit lattice QCD","Lattice QCD continuum limit yields nucleon radii and magnetic moments","Proton and neutron electromagnetic radii from lattice QCD at physical pion mass","Continuum-limit lattice QCD: proton radius 0.860 fm, mu^p=2.849"],"cache_read_input_tokens":46720,"weakest_assumption_plain":"All lattice artifacts are assumed to be captured by a straight-line dependence on the square of the lattice spacing ($a^2$) across only three lattice spacings, with the lattice spacing values taken from an earlier paper; a bend in that line would shift every quoted radius and moment.","fun_headline_variants_meta":{"raw":{"variants":["Proton charge radius 0.860 fm from continuum limit lattice QCD","Lattice QCD continuum limit yields nucleon radii and magnetic moments","Proton and neutron electromagnetic radii from lattice QCD at physical pion mass","Continuum-limit lattice QCD: proton radius 0.860 fm, mu^p=2.849"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0009,"raw_usage":{"total_tokens":4016,"prompt_tokens":1225,"completion_tokens":2791,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":841,"completion_tokens_details":{"reasoning_tokens":2702}},"tokens_in":841,"tokens_out":2791,"duration_ms":23637,"temperature":1.0,"reasoning_tokens":2702,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T13:09:00.543789+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same observables on a fourth, finer ensemble, for example at $a\\approx0.04$ fm, or refit the present data with an $a^4$ term added to the $z$-expansion coefficients; if $\\sqrt{\\langle r_E^2\\rangle^p}$ shifts by more than the quoted $0.023$ fm systematic error, the linear-in-$a^2$ extrapolation is the fragile step.","supporting_citations":[{"cited_title":"Evaluation of disconnected quark loops for hadron structure using GPUs","cited_arxiv_id":"1309.2256","evidence_quote":"provides hierarchical probing to reduce stochastic noise in the disconnected loop trace."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"shows that pion-nucleon excited states can contaminate the three-point function, motivating separate excited-state energies in the fits."},{"cited_title":"Galster, H","cited_arxiv_id":null,"evidence_quote":"gives the accepted experimental values against which the final radii and magnetic moments are compared."}],"review_version":1}