{"id":"9a57066d-0586-4f82-85f1-4b7f0b988bc4","arxiv_id":"2411.19861","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A two-loop scheme-conversion calculation shifts and tightens the world average of the kaon bag parameter to 0.7627(60), updating the Standard Model prediction for epsilon_K.","lead":"This paper computes missing two-loop corrections that convert lattice QCD results for the kaon mixing parameter into the standard continuum scheme, and uses them to update the world average to 0.7627(60). A smart generalist should read it because that single number feeds the Standard Model prediction of indirect CP violation in kaons, a key constraint on new physics.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-loop conversion factors in Table 3 depend on NNLO quark-field factors imported from Refs. [27-29], and the paper does not demonstrate that those factors share the exact MS/NDR convention used for the four-quark sector; any mismatch shifts every coefficient.","rationale":"I read the central claim as the two-loop RI-(S)MOM to MS matching calculation together with the derived world averages. The matching has several independent internal checks, including NLO reproduction, pole cancellation, and consistency with the analytic poles of [6], which support the four-quark part of the computation. The weakest link is not the loop technology but the factorization of the conversion factor: the bilinear sigma factors are external, squared, and untested within this manuscript. A mismatch does not need to be large to matter: at alpha_s(3 GeV) of about 0.24, a 1% error in sigma moves the conversion by roughly 2%, which is comparable to or larger than the quoted final uncertainty on B_hat_K. The lattice inputs are external and the averaging procedure is transparent, so the world averages would shift if the conversion coefficient shifts. Thus an independent check of the imported sigma factors is the single most valuable verification. This does not invalidate the paper, but it justifies making acceptance conditional on the compatibility of the imported bilinear renormalization constants, consistent with the reader's CONDITIONAL verdict.","tokens_in":25814,"tokens_out":8853,"duration_ms":87869,"concrete_test":"Independently re-derive the six NNLO values of sigma(MS,s) for s = gamma_mu and q-slash from the published expressions in Refs. [27-29] with the stated parameter substitutions, insert them together with the paper's lambda(MS,l) into Eqs. (2.17)-(2.18), and compare against Table 3 for f = 3. The comparison should agree within the quoted +/-0.08 numerical errors; alternatively, compute one full two-loop C^{S->MS}_{BK} with a single code that also evaluates the bilinear two-point functions, removing the import entirely.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The construction in Sec. 5.1 uses C^{S->MS}_{BK} = lambda(MS,l)[sigma(MS,s)]^2 (Eqs. 2.17-2.18). The four-quark projected amplitude lambda is computed in this paper, but the bilinear factors sigma(MS,s) are taken at NNLO from Gracey [27], Gorbahn-Jaeger [28] and Chetyrkin-Retey [29], with only the substitutions CA=Nc, TF=1/2, w=1, r=1 stated. Since sigma appears squared, a two-loop error or convention mismatch in any of these imported results propagates directly into every entry of Table 3 and into C^{S->RGI}_{BK} through Eq. (5.2). The paper neither lists the adapted sigma expressions nor verifies that the RI' and RI/SMOM conventions used in [27-29] coincide with the projectors and the BG evanescent MS scheme defined in Secs. 2.2 and 4.5. In particular, if the RI' field renormalization in [27] uses a gamma_mu condition different from Eq. (2.9), or if the massless limit of the results in [28,29] contains finite pieces not matching the Landau-gauge NDR scheme used here, all NNLO entries shift. This is an input assumption structurally separate from the two-loop four-quark calculation, and the central 'two-loop matching complete' claim inherits it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper computes the two-loop conversion factors between six RI-(S)MOM schemes and the MS-bar scheme for the Delta S = 2 four-quark operator that defines the kaon bag parameter. The four-quark projected amplitudes are computed from first principles in this paper, using a projector-based method that avoids gamma5 ambiguities, integration-by-parts reduction, and partly numerical evaluation of master integrals. The two-loop quark-field (bilinear) factors sigma(MS,s) are not computed here but are imported from Refs. [27-29]. The resulting conversion factors are then used to convert lattice bag parameters to the RGI quantity B_hat_K, to combine 3- and 4-flavour lattice results through a charm-threshold matching, and to produce updated world averages B_hat_K^{(f=3)} = 0.7627(60) and B_hat_K^{(f=4)} = 0.7759(84), together with an updated Standard Model prediction |epsilon_K| = 2.171(65)_pert.(71)_non-pert.(153)_param. x 10^{-3}.","tokens_in":26017,"tokens_out":6190,"duration_ms":61255,"significance":"If the results are correct, this is a valuable and timely piece of precision phenomenology. The two-loop RI-(S)MOM to MS-bar conversion for B_K is a natural completion of the NLO program, and the paper contains a number of concrete validation checks: the NLO results reproduce Refs. [17,19,20], the NNLO 1/epsilon poles cancel within the quoted numerical uncertainties and match the analytic pole structure of Ref. [6], and the residual scale dependence is mapped for all six schemes. The paper also makes a genuine effort to combine all available lattice inputs, including a 3/4-flavour matching at O(alpha_s^2). These strengths should be credited. The main reservation concerns not the four-quark calculation itself but the imported bilinear two-loop factors, which enter every NNLO conversion coefficient squared and whose convention compatibility with the MS-bar/NDR scheme used here is not demonstrated.","major_comments":[{"comment":"The NNLO coefficients in Table 3 are built from C^{S->MS}_{BK} = lambda(MS,l) [sigma(MS,s)]^2, with sigma(MS,s) taken at NNLO from Refs. [27-29] after the substitutions CA=Nc, TF=1/2, w=1, r=1. The paper does not list the adapted sigma expressions, nor does it demonstrate that the RI' and RI/SMOM field-renormalization conventions used in [27-29] coincide with the projectors and the BG evanescent MS-bar scheme defined in Secs. 2.2 and 4.5. Since sigma enters squared, any convention mismatch or finite scheme piece in those imported results shifts every entry of Table 3, propagates through Eq. (5.2), and moves the final averages in Eqs. (5.11)-(5.12). This is a load-bearing input assumption, structurally separate from the two-loop four-quark calculation. A concrete fix is to display the explicit NNLO sigma(MS,s) expansions used in this paper and to verify them in the same Landau-gauge, NDR, BG-evanescent convention, for example by checking that the full conversion factor reproduces the pole structure of Ref. [6] without relying on cancellations between the four-quark and bilinear pieces.","section":"Sec. 5.5, Table 8"},{"comment":"Only four of the six lattice entries entering the global average are actually converted with the new two-loop factors. The SWME 15A and Laiho 11 results are in the MS-bar scheme and are converted with the external factor 1.369 taken from FLAG [13], as indicated by the asterisk in Table 8. The abstract states that the world averages combine the complete set of lattice results, but the role of these external conversions should be stated more prominently, and the numerical sensitivity of the final average to using NNLO versus NLO conversion for these two entries should be quantified. As written, the claim that the precision of B_hat_K is improved to NNLO is only partially supported for these two inputs.","section":"Sec. 5.5, Table 8"},{"comment":"The correlation between the two RBC/UKQCD 24 SMOM projections is inferred from the chi^2 value of the two-projection average: the text states that chi^2 = 0.671 is used to estimate a 33.9% correlation. This is an indirect, model-dependent way to assign the covariance, and the same procedure is used for the f = 4 average with a correlation of 68.2%. Since the quoted uncertainties in Eqs. (5.11)-(5.12) depend directly on this correlation, the paper should report the range of central values and errors obtained as the correlation is varied over its allowed interval, or obtain the covariance from the lattice collaboration's published error budget.","section":"Sec. 5.5, Eq. (5.8)"}],"minor_comments":[{"comment":"The coefficients written as 3/9 f, 4/9 f, and 12/9 f should be reduced to 1/3 f, 4/9 f, and 4/3 f, and the definition of f should be restated in the equation caption to avoid ambiguity.","section":"Eq. (5.1)"},{"comment":"The abstract refers to a 'PDG rescaling factor of 1.28', but the text in Sec. 5.5 only introduces it as multiplication of the error by sqrt(chi^2/dof). The explicit connection between the two statements should be made in the text.","section":"Abstract and Sec. 5.5"},{"comment":"The table caption uses the notation (delta_alpha_s/delta_nu) to report the ratio of uncertainties, but this notation is not defined in the text; a phrase such as 'ratio of the alpha_s uncertainty to the scale-variation uncertainty' would be clearer.","section":"Table 5"},{"comment":"The NNLO coefficients are quoted with numerical errors from the integral evaluation, but the corresponding conversion factors in Table 4 are given without propagated uncertainties. Please state explicitly whether the errors in Table 4 include the Table 3 coefficient errors and the alpha_s error separately.","section":"Sec. 5.1, Table 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is clearly within the scope of the journal and the four-quark part of the calculation appears technically sound and well validated. My recommendation is driven by a single structural concern: the NNLO conversion factors rely on imported bilinear two-loop factors whose convention compatibility with the MS-bar/NDR and BG-evanescent scheme is asserted but not demonstrated. Because these factors enter squared, this is not a cosmetic issue. If the authors can provide the explicit sigma(MS,s) expressions and a cross-check in the stated conventions, the paper would likely be suitable for publication after a straightforward revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real content here is the two-loop matching calculation itself. Prior work stopped at NLO, so the conversion factors in Table 3 are new, and the internal checks are genuine: the NLO results reproduce Aoki/Ciuchini/Buras, the NNLO 1/epsilon poles cancel within uncertainties and match Buras-Weisz, and the residual scale dependence is mapped for all six schemes. The paper is also honest about what is perturbative and what is lattice input. The B_K update to 0.7627(60) is a modest but real improvement, and the epsilon_K re-evaluation is straightforward and not oversold. No circularity problem: the lattice data are external and the threshold matching uses published two-loop anomalous dimensions.\n\nThe soft spots are real but not fatal. First, as the stress-test note says, the NNLO conversion factors inherit the NNLO quark-field sigma factors from Gracey, Gorbahn-Jaeger, and Chetyrkin-Retey, and the paper only states substitutions (CA = Nc, TF = 1/2, w = r = 1) without listing the adapted expressions or demonstrating that the RI' and RI/SMOM conventions in those papers match the projectors and the BG evanescent scheme used here. Because sigma enters squared, any convention mismatch shifts every entry of Table 3. This is a legitimate caveat, not a demonstrated error. The cited papers are standard, the substitutions are conventional, and the NLO reproduction would catch gross mismatches, but the burden is on the authors to make that compatibility explicit. A referee should ask for the sigma expressions in an appendix or for a direct cross-check. Second, the claim of an NNLO world average is not uniform: SWME 15A and Laiho 11 are still converted with the FLAG one-loop conversion factor, as the asterisks in Table 8 admit. The paper is transparent about this, but the abstract's 'NNLO' label overstates the average. Third, the drop of Z(2,1)_QEF from A' is asserted without demonstration; minor. Fourth, the numerical NNLO coefficients come with no code or data release, which makes independent verification harder than it should be for a precision claim, though this is hardly unique in the field.\n\nBottom line: this deserves serious refereeing. It is a technically involved, honest paper with a new result and a plausible world average. The referee should focus on the convention compatibility of the imported bilinear factors and on how the mixed-order inputs are billed. With those addressed, the paper is publishable. I would take it to our next reading group and would cite it once the averaging caveats are clear.\n\nRecommendation: send to peer review.","headline":"Genuinely new two-loop RI-(S)MOM to MS matching for B_K, with a solid internal validation, but the headline NNLO claim is diluted by mixed-order lattice inputs and an unverified external convention assumption.","tokens_in":26749,"tokens_out":2920,"would_cite":true,"duration_ms":27107,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper completes the two-loop RI-(S)MOM to $\\overline{\\mathrm{MS}}$ matching for the $\\Delta S=2$ operator and reports $\\hat{B}_K^{(f=3)}=0.7627(60)$, $\\hat{B}_K^{(f=4)}=0.7759(84)$, with an updated Standard Model prediction…","keywords":["kaon bag parameter","RI-SMOM scheme","RI-MOM scheme","MS-bar scheme","two-loop matching","four-quark operator","epsilon_K","lattice QCD"],"falsifier":"Recompute the two-loop off-shell box master integrals with an independent method and re-extract the NNLO coefficients in Table 3; if any coefficient shifts outside its quoted uncertainty, or if the final $\\hat{B}_K$ averages change when the evanescent-operator basis is changed, the claimed two-loop matching is falsified.","tokens_in":25486,"feed_emoji":"⚛️","tokens_out":9912,"duration_ms":86357,"temperature":0.7,"pith_summary":"The paper completes the two-loop matching between the RI-(S)MOM and $\\overline{\\mathrm{MS}}$ renormalization schemes for the $\\Delta S=2$ four-quark operator that controls neutral-Kaon mixing, and uses that matching to convert all published 3- and 4-flavour lattice determinations of the Kaon bag parameter to a common scheme. The central quantitative claims are the conversion factors in Eq. (5.1) with the NNLO coefficients in Table 3, the world averages $\\hat{B}_K^{(f=3)}=0.7627(60)$ and $\\hat{B}_K^{(f=4)}=0.7759(84)$, and the updated value $|\\epsilon_K|=2.171(65)(71)(153)\\times 10^{-3}$. A sympathetic reader should care because $\\hat{B}_K$ is the main non-perturbative input into indirect CP violation in the Kaon system; moving that input to NNLO and combining 3- and 4-flavour lattice results tightens the constraints on the CKM matrix and on new physics. The paper also notes a PDG rescaling factor of 1.28, reflecting a mild tension among lattice inputs after the NNLO conversion.","feed_headline":"Two-loop precision for kaon mixing yields B_hat_K = 0.7627(60)","feed_subtitle":"All 3- and 4-flavor lattice results now combine after full two-loop scheme conversion, tightening the epsilon_K prediction.","key_machinery":"The central object is the amputated four-point Green's function $\\Lambda$ for the operator $Q$, with momentum-subtraction conditions defined by the projectors $P_{(\\gamma_\\mu)}$ and $P_{(q\\!\\!/)}$. The key mechanism is a projection-first calculation: the paper defines dimensionally unambiguous projectors $\\Pi_\\mu,\\Pi_{11},\\Pi_{12},\\Pi_{22}$ that act before tensor reduction, avoiding $\\gamma_5$ and Levi-Civita ambiguities, and it constructs evanescent operators that vanish under all projectors. This decomposes the two-loop amplitude into coefficients $A_i,\\tilde{A}_i$ in front of tree-level matrix elements, from which the (S)MOM projections are recovered. The scheme conversion itself uses the relation $C_{B_K}^{S\\to\\overline{\\mathrm{MS}}}=\\lambda(\\overline{\\mathrm{MS}},l)\\,[\\sigma(\\overline{\\mathrm{MS}},s)]^2$, where $\\lambda$ is the projected four-point amplitude and $\\sigma$ is the scalar two-point amplitude.","core_discovery":"The paper establishes that the two-loop RI-(S)MOM-to-$\\overline{\\mathrm{MS}}$ conversion factors for the $\\Delta S=2$ bag parameter are now complete, with the explicit NNLO coefficients listed in Table 3 for all six momentum-subtraction variants, evaluated in Landau gauge at $N_c=3$. These factors take the form $C_{B_K}^{S\\to\\overline{\\mathrm{MS}}}(\\mu,\\nu)=1+(\\alpha_s/4\\pi)C_{B_K}^{S,\\mathrm{NLO}}+(\\alpha_s/16\\pi^2)[C_{B_K}^{S,\\mathrm{NNLO}}+\\ldots]$, and they are combined with the existing two-loop anomalous dimensions and charm-threshold matching to convert every lattice value of $B_K$ to the renormalization-group-invariant bag parameter. Combining all available 3- and 4-flavour lattice results, the paper obtains the averages $\\hat{B}_K^{(f=3)}=0.7627(60)$ and $\\hat{B}_K^{(f=4)}=0.7759(84)$, and from the 3-flavour average it updates the Standard Model prediction for indirect CP violation in the Kaon system to $|\\epsilon_K|=2.171(65)(71)(153)\\times 10^{-3}$.","pith_inferences":["An extension of the paper's approach is that the same projection-first technique could produce two-loop RI-(S)MOM conversions for the other four-quark operators in $\\Delta F=2$ effective Hamiltonians, where such NNLO conversions are not yet available.","If the imported bilinear conversion constants carry an undetected convention mismatch, the spread among the six schemes after conversion would reveal it: a consistent matching must make all schemes agree on $\\hat{B}_K$, so a residual scheme-dependent spread beyond quoted errors would point to that input.","The quoted PDG rescaling factor of 1.28 is a testable diagnostic: if the mild lattice tension persists after future lattice inputs are added, the averaged error should be rescaled again, whereas a resolved tension would confirm that the NNLO conversion absorbed the previous scheme discrepancy.","The 4-flavour average $\\hat{B}_K^{(f=4)}=0.7759(84)$ is directly usable for future phenomenology that treats the charm quark non-perturbatively, so improvements in long-distance lattice calculations should translate almost entirely into a smaller $\\epsilon_K$ uncertainty."],"forward_implications":["The NNLO conversion factors complete the scheme-conversion chain for $\\hat{B}_K$ at $O(\\alpha_s^2)$, so any existing or future lattice result in any of the six RI-(S)MOM variants can be converted to $\\overline{\\mathrm{MS}}$ at the same perturbative accuracy.","The averages $\\hat{B}_K^{(f=3)}=0.7627(60)$ and $\\hat{B}_K^{(f=4)}=0.7759(84)$ combine all 3- and 4-flavour lattice inputs, using charm-threshold matching at $O(\\alpha_s^2)$, and supersede previous flavour-separated averages with a smaller uncertainty.","The updated $|\\epsilon_K|$ prediction shifts to $2.171(65)(71)(153)\\times 10^{-3}$ and reduces the non-perturbative uncertainty from roughly 3.5% to about 3.28%, making long-distance contributions the dominant non-perturbative error.","Residual scale dependence is smallest for the SMOM scheme with slashed-momentum projectors at both 3 and 4 flavours, so future lattice determinations using that scheme will incur the smallest conversion uncertainty.","The same conversion factors also yield $\\overline{\\mathrm{MS}}$ and renormalization-group-invariant bag parameters for $D$-meson mixing from the existing four-flavour lattice input."],"supporting_citations":[{"why":"Supplies the one-loop quark-field and operator conversion coefficients that serve as the NLO input and validation baseline for the two-loop calculation.","marker":"[17]"},{"why":"Provides the three-loop quark-field conversion factor used for the NNLO $\\sigma$ contribution in the RI-prime scheme with the stated colour-charge substitutions.","marker":"[27]"},{"why":"Provides the two-loop bilinear conversion coefficient in the SMOM scheme, used with the parameters $w=1$, $r=1$.","marker":"[28]"},{"why":"Supplies the three- and four-loop quark-field renormalization constants that are converted into the NNLO factors $\\sigma(\\overline{\\mathrm{MS}},s)$.","marker":"[29]"},{"why":"Defines the NNLO Wilson-coefficient and evanescent-operator scheme, gives the RG evolution operator in Eqs. (5.3)-(5.5), and supplies the charm-threshold matching used for flavour conversion.","marker":"[7]"},{"why":"Gives the two-loop operator anomalous dimension whose pole structure is used to check the cancellation of ultraviolet divergences in the matching calculation.","marker":"[6]"},{"why":"Provides the prior world-average value and the lattice-input compilation that the new averages update and supersede.","marker":"[13]"},{"why":"Is the primary lattice determination in the two SMOM schemes, supplying the bag-parameter values converted at NNLO.","marker":"[32]"},{"why":"Supplies the RI-MOM lattice bag parameter that is converted through the new NNLO matching.","marker":"[34]"}],"fun_headline_variants":["Two-loop matching gives B_K = 0.7627(60) from all lattice data","NNLO scheme conversion tightens kaon CP violation","B_K world average updated to 0.7627(60) at NNLO","All 3- and 4-flavor lattice B_K combined at two loops","Two-loop RI-(S)MOM to MSbar conversion done for B_K"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes that the three-loop quark-field renormalization constants it imports from other papers use exactly the same conventions as its own four-quark operator scheme; if that assumption is wrong, all the NNLO conversion numbers move.","fun_headline_variants_meta":{"raw":{"variants":["Two-loop matching gives B_K = 0.7627(60) from all lattice data","NNLO scheme conversion tightens kaon CP violation","B_K world average updated to 0.7627(60) at NNLO","All 3- and 4-flavor lattice B_K combined at two loops","Two-loop RI-(S)MOM to MSbar conversion done for B_K"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000954,"raw_usage":{"total_tokens":4155,"prompt_tokens":1119,"completion_tokens":3036,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":735,"completion_tokens_details":{"reasoning_tokens":2932}},"tokens_in":735,"tokens_out":3036,"duration_ms":20968,"temperature":1.0,"reasoning_tokens":2932,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:47:16.339818+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the two-loop off-shell box master integrals with an independent method and re-extract the NNLO coefficients in Table 3; if any coefficient shifts outside its quoted uncertainty, or if the final $\\hat{B}_K$ averages change when the evanescent-operator basis is changed, the claimed two-loop matching is falsified.","supporting_citations":[{"cited_title":"Precise MS-bar light-quark masses from lattice QCD in the RI/SMOM scheme","cited_arxiv_id":"1004.3997","evidence_quote":"Provides the two-loop bilinear conversion coefficient in the SMOM scheme, used with the parameters $w=1$, $r=1$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the two-loop operator anomalous dimension whose pole structure is used to check the cancellation of ultraviolet divergences in the matching calculation."}],"review_version":1}