{"id":"97920bb8-0d5f-422e-a112-d0218f423a4b","arxiv_id":"2412.07263","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A GEVP weighting that skips the costly pion-nucleon diagonal three-point function significantly reduces N pi excited-state contamination for isovector pseudoscalar and axial nucleon matrix elements at m_pi=131 MeV.","lead":"This lattice QCD paper tests a cheaper way to remove pion-nucleon contamination from nucleon matrix element calculations, using a generalized eigenvalue problem with nucleon and pion-nucleon operators. It finds the method cleans up the axial and pseudoscalar channels at physical pion mass, while the nucleon sigma term is unaffected.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The omitted diagonal Npi three-point function is the load-bearing assumption: Eq. (5) sets d_Npi,Npi=0 based on an asserted but unquantified decay hierarchy that is only weakly satisfied at physical pion mass; a direct computation of is required.","rationale":"The paper is a proceedings contribution and its strongest evidence is the parity-zero test (Fig. 3) and the timelike axial agreement with Ref. [9] (Fig. 4). These are genuine successes, and I agree they support the qualitative conclusion for those channels. However, the method's novelty is precisely the omission of the diagonal Npi three-point function; the assertion that it is subdominant is not demonstrated. The exponential hierarchy is real asymptotically, but at the physical pion mass the relevant gap is only m_pi and the separations are 1-2 fm, so the suppression is moderate and is not quantified. The reader's CONDITIONAL verdict is appropriate: the abstract's blanket claim of minimized contamination for all five currents is also not supported by the body, which states in Sec. 5 that no significant improvement is observed for the majority of cases. My proposed test is a direct computation of the omitted correlator, which would settle whether the residual bias is negligible. I do not see an internal inconsistency or a reason to reject; the issue is an unquantified systematic that must be bounded before the method can be adopted.","tokens_in":6848,"tokens_out":7418,"duration_ms":75491,"concrete_test":"On the same cA2.09.48 ensemble, compute the omitted diagonal three-point function for the axial and pseudoscalar currents on a subset of ~100-200 configurations. Reconstruct both I_d (Eq. 5 with d_Npi,Npi=0) and the full GEVP combination I (Eq. 4) and compare the extracted ratios at the source-sink separations used in Figs. 3-5. If the difference I_d - I, normalized by , is larger than the statistical error or shows a trend with t_s, the central assumption fails and the reported improvements are not controlled. A cheaper cross-check is to repeat the analysis on the heavier-pion ensemble of Ref. [4], where the energy gap is larger and the suppression stronger; a significant shift there would already invalidate the claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. 4 introduces I_d in Eq. (5) with d_Npi,Npi=0, then fixes the other weights in Eq. (6) to eliminate the off-diagonal and contaminations. The cost of this choice is an uncontrolled residual: the GEVP-weighted diagonal term v_{0,Npi} v*_{0,Npi} is omitted entirely. The paper's justification is the 'key observation' in Sec. 4 that this diagonal contamination decreases faster than the off-diagonal terms as time separations increase. No derivation or numerical estimate is given. A spectral decomposition shows the diagonal-to-off-diagonal ratio is e^{-(E_Npi - E_N)(t_s - t_ins)}; at physical pion mass E_Npi - E_N ~ m_pi ~ 0.14 GeV, so even a 1 fm source-sink-minus-insertion separation suppresses the diagonal term only by a factor ~0.5 relative to the off-diagonal term. Moreover, in the plateau plots of Figs. 3-5 the insertion time is varied at fixed t_s, in which case the diagonal term is independent of t_ins while the off-diagonal terms decay with t_ins, so the claimed hierarchy does not improve along the plotted direction. Since the entire method is motivated by avoiding the expensive three-point function, the residual bias from this term is the central systematic uncertainty; the successful parity-zero test for the pseudoscalar and the agreement of the timelike axial channel with the continuum reference do not constrain it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This proceedings paper proposes a GEVP-based method to reduce N-pi excited-state contamination in nucleon matrix elements while avoiding the computationally most expensive three-point function, . The authors build a two-operator basis (N, N-pi), solve the GEVP, define a weighted combination I_d of the three cheaper three-point functions (Eq. 5), set the diagonal N-pi-N-pi weight to zero, and choose the remaining weights (Eqs. 6-7) to cancel the off-diagonal N-N-pi contaminations. The method is applied to scalar, vector, pseudoscalar, axial, and tensor currents on an N_f=2 twisted-mass ensemble at m_pi=131 MeV with 1228 configurations. The paper reports that the GEVP improves the isovector pseudoscalar and axial channels, most notably the timelike axial charge, which moves into agreement with the continuum-limit result of Ref. [9], while it does not improve the sigma-term ratios. A parity-zero pseudoscalar ratio that should vanish is used as a consistency test.","tokens_in":7136,"tokens_out":18059,"duration_ms":183055,"significance":"If the key assumption in Sec. 4 were controlled, this would be a practically valuable method: it reduces the dominant N-pi contamination using only the three cheaper three-point functions, and the parity-zero pseudoscalar test is a genuine operator-level check. The authors also include disconnected contributions for isovector operators, which is a nontrivial technical step. The central weakness is that the method drops the diagonal N-pi-N-pi three-point function on the basis of an asserted and unquantified decay hierarchy; the provided tests do not directly bound the residual. The value of the paper therefore depends on whether that residual can be shown to be negligible, either by a direct estimate or by a clear model-based argument.","major_comments":[{"comment":"The decision to set d_{Npi,Npi}=0 rests on the claim that the diagonal contamination decreases faster than the off-diagonal terms as the time separations increase. This is not derived or quantified. In a spectral decomposition relative to the ground N-N term, the Npi-Npi term is suppressed as exp[-(E_Npi-E_N) t_s], the N-Npi term as exp[-(E_Npi-E_N) t_ins], and the Npi-N term as exp[-(E_Npi-E_N)(t_s-t_ins)]. In the plateau plots of Figs. 3-5, t_s is fixed and t_ins is varied, so the omitted diagonal term is independent of t_ins while the off-diagonal terms decay; the plotted improvement with t_ins therefore does not demonstrate suppression of the residual. At m_pi=131 MeV the energy gap is only about m_pi, so exp[-(E_Npi-E_N) t_s] is not negligible for the t_s values shown. Please provide a quantitative estimate of this residual, for example by computing on a subset of configurations or by including it in a model fit; without this, the central claim that I_d isolates is not established.","section":"Sec. 4, Eq. (5)"},{"comment":"The two external checks do not constrain the omitted diagonal term. The parity-zero pseudoscalar test (Fig. 3) is a null-channel check: it demonstrates cancellation of the contaminating contributions relevant to that operator, but it does not measure the size of the diagonal term for a general operator O, and for the pseudoscalar channel that diagonal term may vanish for the same parity reason. The agreement of the timelike axial ratio with the continuum-limit band of Ref. [9] (Fig. 4, second row) is suggestive, but it is obtained at a single lattice spacing with a different action and including disconnected contributions that were not present in Ref. [9]; it cannot by itself establish that the omission of the diagonal term is negligible. Please clarify what these tests can and cannot establish, and if possible add a channel where the omitted term has a known nonzero value.","section":"Sec. 5, Figs. 3-4"}],"minor_comments":[{"comment":"The weights d_N,N, d_N,Npi, and d_Npi,N are stated without derivation. Because the cancellation property of I_d depends on these formulas, please either give the derivation or point to the specific section of Ref. [4] where it appears.","section":"Sec. 4, Eqs. (6)-(7)"},{"comment":"The abstract states that Npi contamination is minimized for the scalar, vector, pseudoscalar, axial, and tensor currents, but Section 5 reports no significant improvement for the majority of cases, including the sigma-term ratios. Please align the abstract with the body of the paper.","section":"Abstract and Sec. 5"},{"comment":"In Eq. (1), the limit \"t_s - t_sink -> infinity\" is not defined; the sink time is t_s, so this should likely be \"t_s - t_ins -> infinity\" or a similar expression for the source-sink separation.","section":"Eq. (1)"},{"comment":"There is a typo in \"nu_{jk}(t,t+0)\"; the second argument should be t_0. Also, the reference list entry for Ref. [8] begins with \"A. Collaboration\"; this should read \"ALPHA Collaboration\" or similar.","section":"Eq. (2) and references"},{"comment":"The text before Fig. 5 contains \"thre-point functions\"; it should be \"three-point functions\".","section":"Sec. 5, before Fig. 5"},{"comment":"The phrase \"when the source-sink and source-insertion time separations increase\" is ambiguous; please specify whether t_s and t_ins are varied independently, at fixed t_s, or with t_s-t_ins fixed, since the spectral suppression factors are different in each case.","section":"Sec. 4"},{"comment":"Please state how the statistical uncertainty of the GEVP eigenvectors and of the weights d in Eq. (6) is propagated into the final ratios; the plots show bands for the model averages but not the treatment of the eigenvector errors.","section":"Sec. 5, Figs. 4-5"}],"recommendation":"major_revision","confidential_remarks":"The paper is a proceedings contribution that relies on a companion PRD paper for details. The main issue is that the cost-saving omission of is the central systematic uncertainty and the current evidence does not bound it. If the authors can produce a quantitative estimate (even on a subset) or a model-based bound, I would support publication; as it stands, the central claim is not yet established. The parity-zero test is a nice internal check but is not sufficient."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThis proceedings has one genuinely useful trick and one genuinely shaky justification. The trick: use GEVP with nucleon and Npi interpolators but set the diagonal Npi-Npi weight to zero, so you never compute the most expensive three-point function. That is a real saving. The paper also does two honest checks: the parity-zero pseudoscalar ratio (Fig. 3) goes to zero with the GEVP combination, which is a clean signal that the dominant Npi contamination is removed, and the timelike axial channel moves into agreement with the continuum limit of Ref. [9]. The weights in Eqs. (6)-(7) are derived from two-point GEVP eigenvectors and are current-independent, so the method is not circularly tuned to the target matrix elements.\n\nThe soft spot is the 'key observation' in Sec. 4. The paper claims the diagonal contamination decays faster than the off-diagonal terms. That is not derived and, in the plateau plots where t_s is fixed and t_ins is varied, the opposite is true: the diagonal term is constant in t_ins while the off-diagonal terms decay, so the ratio grows as t_ins approaches the sink. The suppression only works if you increase t_s at fixed t_ins, which is not the plotted direction. The practical saving may still be fine because the GEVP eigenvector component v_{0,Npi} is small, but the paper never quantifies that residual. A direct computation of (or an estimate from the eigenvector overlaps) is the obvious missing piece.\n\nSecond: the abstract says contamination is minimized for all five currents, but the body says most cases show no significant improvement. That is a direct contradiction; the abstract should be corrected.\n\nThird: results are at a single lattice spacing, and the full analysis is in the authors' PRD (Ref. [4]). This proceedings is a condensation, which is fine, but it should not be read as the primary source.\n\nOverall, the method is plausible and the axial/pseudoscalar improvement is supported by the checks. The load-bearing assumption is under-analyzed, not obviously wrong. A serious referee should ask for the missing systematic estimate and a corrected abstract. I would not cite this proceedings over the PRD, but I would send it to review rather than desk reject: the method deserves scrutiny and the fixes are tractable.","headline":"Useful cost-saving trick undercut by an unquantified zeroing of the diagonal Npi term and an abstract that overclaims.","tokens_in":7764,"tokens_out":3636,"would_cite":false,"duration_ms":35689,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81T25","81V05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Nucleon matrix elements can be cleaned of pion-nucleon contamination without computing the costliest correlation function.","keywords":["lattice QCD","nucleon matrix elements","generalized eigenvalue problem","pion-nucleon excited states","excited-state contamination","axial charge","pseudoscalar form factors","twisted-mass fermions"],"falsifier":"On the same ensemble, compute the omitted three-point function $\\langle J_{N\\pi}(t_s) O(t_{\\rm ins}) \\bar J_{N\\pi}(0)\\rangle$ and compare the combination $I_d$ with the full GEVP combination $I$ over the same separations. If the difference is comparable to the statistical errors, or if the improved ratios keep drifting with $t_s$, the assumption that the diagonal term decays faster is falsified. In a cheaper version, check the plateau: the GEVP-improved ratio should become time-independent beyond the fitted range; a residual slope or a plateau that sits outside the continuum-limit band would signal leftover contamination.","tokens_in":6557,"feed_emoji":"⚛️","tokens_out":10710,"duration_ms":107023,"temperature":0.7,"pith_summary":"This paper tries to establish that a generalized eigenvalue problem (GEVP) built from a nucleon interpolator and a pion-nucleon interpolator can suppress the dominant excited-state contamination in nucleon matrix elements while skipping the most expensive three-point correlation function: the one with pion-nucleon interpolators at both source and sink. On a physical-pion-mass ensemble ($m_\\pi = 131$ MeV), the authors form an improved nucleon operator from GEVP eigenvectors and use current-independent weights so that the off-diagonal terms $\\langle N|O|N\\pi\\rangle$ and $\\langle N\\pi|O|N\\rangle$ cancel. They set the diagonal $\\langle N\\pi|O|N\\pi\\rangle$ term to zero, relying on the observation that it falls faster than the off-diagonal contamination as the time separations grow. The payoff is a clear reduction of excited-state effects for the isovector pseudoscalar and axial currents, while the nucleon $\\sigma$-term ratios are essentially unchanged, indicating that its contamination is not dominated by the lowest $N\\pi$ state.","feed_headline":"Cleaner nucleon charges without the costly pion-nucleon diagram","feed_subtitle":"A two-operator GEVP removes excited-state contamination in axial and pseudoscalar currents at physical pion mass.","key_machinery":"The carrying object is the two-operator GEVP on the correlation matrix $C_{jk}(t) = \\langle J_j(t) \\bar J_k(0) \\rangle$ with basis $\\{J_N, J_{N\\pi}\\}$. Solving it gives an eigenvector $v_0$ that defines an improved interpolator $\\tilde J_N = v_{0,N} J_N + v_{0,N\\pi} J_{N\\pi}$. The three-point combination $I_d$ uses weights fixed by the eigenvector matrix, with $d_{N\\pi,N\\pi}=0$, so the expensive $\\langle J_{N\\pi} O \\bar J_{N\\pi}\\rangle$ correlator is never evaluated. These weights cancel the off-diagonal contaminations, and the paper argues that the diagonal term is suppressed at the time separations used; to make that argument stable, the analysis fixes $t-t_0$ rather than a small reference time $t_0$, because the eigenvectors show strong $t_0$ dependence.","core_discovery":"On the paper's own terms, the central claim is that the combination $I_d$ of three-point functions, built from GEVP weights with the diagonal pion-nucleon term omitted, still isolates $\\langle N|O|N\\rangle$ because the diagonal $\\langle N\\pi|O|N\\pi\\rangle$ contamination decays faster than the off-diagonal terms. The weights are $d_{N,N}=1-W^*W$, $d_{N,N\\pi}=1+W^*$, and $d_{N\\pi,N}=1+W$, with $W = 1/(v_{0,N}[v^{-1}]_{N,0}) - 1$, so they are fixed by the two-point correlation matrix and do not depend on the inserted current. The decisive observation is that $\\langle N\\pi|O|N\\pi\\rangle$ decays faster than the off-diagonal terms as the source-sink and source-insertion separations increase. At $m_\\pi = 131$ MeV the method makes the parity-zero pseudoscalar ratio consistent with zero, brings the timelike component of the isovector axial charge into agreement with the continuum-limit reference, and flattens the time dependence of the isovector pseudoscalar form factor, while leaving the $\\sigma$-term extraction unchanged.","pith_inferences":["Because the diagonal term is never computed, the same construction can be transplanted to other baryons or to $N\\pi\\pi$ systems whenever one multihadron channel dominates and its diagonal three-point function is the expensive one.","A direct test of the method's key assumption would be a single calculation of the omitted diagonal three-point function on a small ensemble; comparing $I_d$ with the full $I$ would convert the decay-rate observation into a quantitative bias estimate.","The strong reference-time dependence of the GEVP eigenvectors suggests that results should be checked with more than two interpolators, since a larger basis would make the diagonal suppression less dependent on the eigenvector convention."],"forward_implications":["The isovector pseudoscalar and axial channels no longer need the diagonal $\\langle J_{N\\pi}O\\bar J_{N\\pi}\\rangle$ three-point function to control $N\\pi$ contamination at physical pion mass, freeing computer time for more configurations or longer separations.","A GEVP-improved operator makes the timelike axial component agree with the continuum-limit comparison, supporting a reliable $g_A$ extraction from this ensemble.","The absence of improvement in the sigma-term ratios implies that its contamination is not dominated by the lowest $N\\pi$ state, redirecting future excited-state studies to other states.","The inclusion of disconnected isovector quark-loop diagrams, absent in the comparison continuum study, can soften the lattice-spacing dependence seen in the induced pseudoscalar form factor."],"supporting_citations":[{"why":"introduced the use of $N\\pi$ interpolators in nucleon three-point functions and demonstrated their effect on isovector pseudoscalar and axial channels, the pattern this work builds on.","marker":"[3]"},{"why":"the companion paper giving the full analysis and results at a heavier pion mass; the present proceedings refer to it for details of the method and fits.","marker":"[4]"},{"why":"provides the generalized-eigenvalue framework for energies and matrix elements, including the reference-time dependence that motivates the fixed $t-t_0$ prescription.","marker":"[8]"},{"why":"provides the continuum-limit axial and pseudoscalar form-factor results used as the comparison bands in the ratios.","marker":"[9]"}],"fun_headline_variants":["GEVP skips costly pion-nucleon term for cleaner nucleon matrix elements","Two-operator trick removes excited-state contamination in nucleon charges","Avoiding pion-nucleon diagram yields cleaner axial and pseudoscalar currents","Cost-saving GEVP cleans up nucleon matrix element extraction","Pion-nucleon contamination tamed by generalized eigenvalue method"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the pion-nucleon-to-pion-nucleon three-point correlation decays faster than the cross-terms involving one nucleon at the time separations used, so setting its weight to zero does not bias the matrix element; if that decay is not fast enough, the improved operator simply inherits the contamination it was meant to remove.","fun_headline_variants_meta":{"raw":{"variants":["GEVP skips costly pion-nucleon term for cleaner nucleon matrix elements","Two-operator trick removes excited-state contamination in nucleon charges","Avoiding pion-nucleon diagram yields cleaner axial and pseudoscalar currents","Cost-saving GEVP cleans up nucleon matrix element extraction","Pion-nucleon contamination tamed by generalized eigenvalue method"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000811,"raw_usage":{"total_tokens":3544,"prompt_tokens":918,"completion_tokens":2626,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":534,"completion_tokens_details":{"reasoning_tokens":2533}},"tokens_in":534,"tokens_out":2626,"duration_ms":20569,"temperature":1.0,"reasoning_tokens":2533,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T18:56:24.217179+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the same ensemble, compute the omitted three-point function $\\langle J_{N\\pi}(t_s) O(t_{\\rm ins}) \\bar J_{N\\pi}(0)\\rangle$ and compare the combination $I_d$ with the full GEVP combination $I$ over the same separations. If the difference is comparable to the statistical errors, or if the improved ratios keep drifting with $t_s$, the assumption that the diagonal term decays faster is falsified. In a cheaper version, check the plateau: the GEVP-improved ratio should become time-independent beyond the fitted range; a residual slope or a plateau that sits outside the continuum-limit band would signal leftover contamination.","supporting_citations":[{"cited_title":"Nucleon form factors and the pion-nucleon sigma term","cited_arxiv_id":"2301.07885","evidence_quote":"introduced the use of $N\\pi$ interpolators in nucleon three-point functions and demonstrated their effect on isovector pseudoscalar and axial channels, the pattern this work builds on."}],"review_version":1}