{"id":"89cc3c99-a669-4db6-a7be-376dc65dc285","arxiv_id":"1908.03755","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A mu-tau-philic second Higgs doublet can fit both the muon g-2 and tau LFU anomalies, but LHC searches require the new CP-even Higgs mass above 560 GeV.","lead":"This paper studies a two-Higgs-doublet model where one new Higgs doublet interacts only with muons and taus. It shows the model can explain the muon g-2 anomaly and tau-decay lepton universality deviations, and that LHC searches push the new Higgs mass above 560 GeV.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The real weak point is not tau->mu gamma (that amplitude vanishes at one loop here) but the large radiative correction to the muon mass from the same rho coupling, which the paper never constrains.","rationale":"The reader's verdict is CONDITIONAL and I agree that the paper needs additional checks, but not for the reason stated. The reader's weakest_assumption was the omission of tau->mu gamma; that specific constraint is not effective in this model. With only off-diagonal mu-tau scalar couplings, a one-loop dipole transition tau->mu gamma would require the internal charged-lepton propagator to be simultaneously a muon at one scalar vertex and a tau at the other, which is impossible without flavor-changing fermion self-energies. The same argument suppresses one-loop Z->tau mu. Thus adding tau->mu gamma would not change the parameter scan.\n\nThe genuine soft spot is the radiative muon mass correction. The same coupling rho that drives Eq. (12) also generates a one-loop shift to m_mu proportional to rho^2 m_tau and logarithmically enhanced. For the LHC-surviving region rho > 0.68 and mH ~ 560 GeV, the shift is of order 0.2 GeV, roughly twice the muon mass. The paper's Sec. III.A list of 'relevant theoretical and experimental constraints' does not include any muon-mass naturalness condition. If one imposes |delta_m_mu| < m_mu as a standard perturbativity/naturalness requirement, the allowed rho drops below the LHC-allowed value and the claimed simultaneous solution disappears or requires an extreme cancellation. This is not a hard inconsistency, so I keep the reader's CONDITIONAL verdict rather than moving to REJECT, but the condition should be: verify the muon-mass correction is acceptably small, or explicitly argue that the fine-tuning is acceptable. The proposed concrete test settles whether this concern lands.","tokens_in":11892,"tokens_out":41735,"duration_ms":454415,"concrete_test":"Add a muon-mass renormalization check to the scan: compute the finite one-loop correction delta_m_mu from Eq. (8) for every sample that survives the constraints of Sec. III.A (e.g., delta_m_mu = rho^2 m_tau/(16 pi^2) [ln(mH^2/m_tau^2) + ln(mA^2/m_tau^2) - 3], or with the exact loop functions from the same diagrams that give Eq. (12) but with zero photon). Then impose |delta_m_mu| < m_mu and rerun the scan shown in Fig. 2. If no surviving samples with mH > 560 GeV and rho > 0.68 remain, the claimed LHC-allowed region is excluded by the muon-mass naturalness requirement, and the central claim must be weakened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's proposed tau->mu gamma constraint is actually not a valid gap: because the only neutral-scalar Yukawas are off-diagonal mu-tau, any one-loop l_j->l_i gamma diagram would need the internal charged-lepton propagator to be a different flavor at the two scalar vertices; in the mass basis that is impossible. The same argument suppresses one-loop Z->tau mu. So the omitted cLFV constraints are not the load-bearing issue.\n\nThe load-bearing issue is the one-loop correction to the muon mass generated by the same Eq. (8) coupling. For a surviving point such as mH=560 GeV, mA~590 GeV, rho~0.68 (the region claimed to pass the LHC bound in Sec. III.B), the finite correction is roughly delta_m_mu ~ rho^2 m_tau/(16 pi^2) [F(mH)+F(mA)], with F~ln(m_phi^2/m_tau^2)-3/2 ~ 10. This gives delta_m_mu ~ 0.2 GeV, about twice the physical muon mass. The paper's constraint list in Sec. III.A includes vacuum stability, unitarity, perturbativity, oblique parameters, Z decays and LHC searches, but never imposes that this radiative correction not exceed the observed muon mass. Requiring |delta_m_mu| < m_mu would push rho below about 0.45 for mH~560 GeV, excluding all LHC-allowed samples with rho>0.68. At minimum the surviving region relies on a >100% cancellation between the bare muon mass and this loop correction, which is a severe fine-tuning; at worst the one-loop prediction for delta_a_mu from the same vertex is not under control once the associated mass shift is larger than the physical mass.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper considers a two-Higgs-doublet model in which the second doublet Φ2 is inert and carries only off-diagonal μ-τ Yukawa couplings. Assuming real, symmetric couplings ρμτ=ρτμ=ρ, the authors compute the one-loop contribution to the muon anomalous magnetic moment and the tree/loop corrections to tau-lepton universality ratios, and scan the parameter space 300 GeV < mH < 800 GeV, mH < mA = mH± < mH + 200 GeV, 0.1 < ρ < 1.0. After imposing theoretical constraints, oblique parameters, Z-decay universality, and a suite of LHC multilepton searches implemented with CheckMATE and related tools, they find a surviving region with small mass splittings Δm ≡ mA(mH±) − mH < 50 GeV and ρ > 0.36, and claim that LHC searches require mH > 560 GeV and ρ > 0.68. The central claim is that both the muon g-2 and tau LFU anomalies can be simultaneously explained while passing current direct searches.","tokens_in":12426,"tokens_out":43165,"duration_ms":463831,"significance":"If correct, the model would be an economical simultaneous explanation of the muon g-2 and tau LFU discrepancies, and the detailed LHC recasting with cut-flow tables would be a useful reference for similar LFV-Higgs models. The one-loop g-2 and tau LFU formulas are standard, the scan logic is clear, and the LHC analysis uses public tools in a transparent way. However, the paper omits a decisive constraint: the same ρ coupling generates τ→μγ at one loop, and the surviving parameter space appears to predict Br(τ→μγ) at the 10^{-6}–10^{-5} level, far above the BaBar bound of 4.4×10^{-8}. In addition, the Z4 charge assignment as printed does not allow the defining interaction of Eq. (8). I also checked the radiative-muon-mass concern and find that it does not land: the one-loop correction to mμ vanishes because the two off-diagonal Yukawa vertices project the internal tau line with P_R and P_L, annihilating the tau-mass term. The strengths of the paper do not compensate for the missing charged-lepton flavor-violating constraint.","major_comments":[{"comment":"The scan constraints listed in Sec. III.A do not include any charged-lepton flavor-violating observable, in particular the very strong limit Br(τ→μγ) < 4.4×10^{-8} (BaBar). This is a load-bearing omission. The ρ coupling of Eq. (8) generates τ→μγ at one loop: the transition τ_L → μ_R requires only a single flavor-changing Yukawa vertex, with the photon emitted from the internal tau (or muon) line. The H and A contributions cancel in the mA=mH limit, but the g-2-favored region has mA−mH = 10–50 GeV, ρ ≈ 0.7–1.0, and mH ≈ 560–800 GeV, so the cancellation is incomplete. Using the standard one-loop dipole formula with L = ln(mH^2/mτ^2) − 3/2 ≈ 10, the residual rate is approximately Br(τ→μγ) ≈ [α mτ^5/(4 Γτ)] [ρ L (mA^2−mH^2)/(16π^2 mH^4)]^2. At the illustrative point mH = 600 GeV, Δm = 30 GeV, ρ = 0.7, this gives Br(τ→μγ) ≈ 2×10^{-6}, more than one order of magnitude above the experimental bound. The surviving samples shown in Fig. 2 (ρ > 0.68, mH > 560 GeV) therefore appear to be excluded once this constraint is imposed. I request a full one-loop computation of τ→μγ (and also of Z→τμ, which receives a similar one-loop contribution) and a rerun of the scan with these constraints included.","section":"Sec. III.A (constraint list); Sec. III.B/Fig. 2"},{"comment":"The Z4 charge assignment in Table I is inconsistent with the interaction in Eq. (8). With the entries as printed, q(Lμ) = i, q(τR) = −i, q(Φ2) = −1, so q(Lμ Φ2 τR) = i · (−1) · (−i) = −1, and similarly q(Lτ Φ2 μR) = −1. Neither product is invariant under the Z4 symmetry. Unless the table is a typographical error and the correct charges are supplied, the model's defining LFV Yukawa interaction is forbidden by the very symmetry that is introduced to justify it. This should be corrected and the charges verified explicitly.","section":"Table I and Eq. (8)"}],"minor_comments":[{"comment":"There is a missing parenthesis in the second term of Eq. (12): it should read (log(mA^2/mτ^2) − 3/2)/mA^2 rather than log(mA^2/mτ^2 − 3/2)/mA^2.","section":"Eq. (12)"},{"comment":"The function H(x) has a removable singularity at x = 1, and the paper sets mA = mH± so that xA = 1. The limiting value H(1) = −2 should be stated explicitly so that the numerical evaluation is unambiguous.","section":"Eq. (17)"},{"comment":"The text sometimes uses 'LUF' for 'LFU' (e.g., in the Sec. III heading and in the discussion of Fig. 1). This should be corrected for consistency.","section":"Throughout"}],"recommendation":"reject","confidential_remarks":"The missing τ→μγ constraint is, in my assessment, decisive: the surviving region claimed in Sec. III.B appears to predict a τ→μγ rate one to two orders of magnitude above the BaBar limit. The authors should recompute this observable and rerun the scan; if the allowed region disappears, the paper's central claim is falsified. The Z4 table inconsistency, if not a typo, is also fatal. These are not presentation issues and require a substantially revised analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper does exactly one genuinely new thing: it takes the mu-tau-philic inert doublet from Abe-Toma-Tsumura and recasts ATLAS/CMS multilepton searches to derive a concrete bound, m_H > 560 GeV. The authors are upfront that the simultaneous fit to muon g-2 and tau LFU is already in Ref. [23]; the recast is the contribution, and the appendix cut flow is a nice, transparent piece of work.\n\nThe scan logic is standard and the presentation is clear. The LHC simulation uses public tools and identifies the sensitive signal regions (SR-A44, SR-C18). The note added about the overlapping paper is honest.\n\nNow the soft spots. The reader’s flagged gap — tau->mu gamma — is not actually a gap. Because the only neutral-scalar Yukawas are the off-diagonal mu-tau entries, the one-loop dipole amplitude vanishes; there is no way to close the flavor loop. I verified the index structure. So that constraint omission is harmless.\n\nThe real problem is the radiative correction to the muon mass generated by the same rho coupling. The one-loop diagram with a tau in the internal line gives delta m_mu ~ rho^2 m_tau/(16 pi^2) times a log of order 10 for the masses in play. For rho~0.7 and mH~560 GeV that is several tens of MeV; for rho near 1 it is of order the physical muon mass. The paper never requires this mass shift to be smaller than m_mu. That is a naturalness problem, and more importantly it signals the one-loop delta a_mu calculation is not under control in the large-rho part of the allowed region: the same vertex produces a mass correction comparable to the muon mass itself. At minimum the surviving points with rho>0.68 rely on a strong cancellation between the bare and loop contributions.\n\nMinor: the recast is not validated against official supplementary material, though the cut-flow table mitigates that.\n\nVerdict: this deserves peer review. The LHC bound is a real result for the model, and the muon-mass issue can be addressed in revision. I would not desk-reject it.","headline":"A solid recast paper that should be reviewed, but the large radiative correction to the muon mass is the real soft spot, not tau->mu gamma.","tokens_in":12832,"tokens_out":6432,"would_cite":true,"duration_ms":68474,"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":"A second Higgs doublet that couples only to muon and tau flavours can simultaneously explain the muon g-2 and tau-decay lepton-universality anomalies while surviving LHC multilepton searches, provided the heavy CP-even Higgs is above 560…","keywords":["two-Higgs-doublet model","muon g-2","lepton flavour universality","tau decays","mu-tau flavour violation","LHC multilepton searches","inert Higgs doublet","flavour-changing Higgs couplings"],"falsifier":"Compute the one-loop branching ratio $B(\\tau\\to\\mu\\gamma)$ for the surviving points with $\\rho>0.36$, $m_H\\in[560,800]$ GeV and $\\Delta m<50$ GeV, and compare it with the current experimental upper bound of a few $10^{-8}$; any point exceeding that bound falsifies the paper's claim that the two anomalies can be simultaneously explained in this model.","tokens_in":2149,"feed_emoji":"⚛️","tokens_out":10170,"duration_ms":196565,"temperature":0.7,"pith_summary":"This paper proposes a minimal addition to the Standard Model: a second Higgs doublet that has no ordinary Yukawa couplings, only flavour-changing couplings between the muon and the tau. If that doublet is real, the model can explain two long-standing anomalies at once: the 3.7 sigma excess in the muon anomalous magnetic moment and the roughly 2 sigma hints of lepton-flavour universality violation in tau decays. The paper then confronts the same parameter region with LHC multilepton searches and finds that the two anomalies can coexist with data only when the extra scalars are nearly degenerate and the heavy CP-even Higgs is heavier than about 560 GeV. This matters because it converts two independent experimental puzzles into one testable prediction about new scalar states at the TeV scale.","feed_headline":"One extra Higgs doublet fits muon g-2 and tau-decay anomalies","feed_subtitle":"A mu-tau-only second Higgs doublet passes LHC multilepton searches only if the heavy CP-even Higgs exceeds 560 GeV.","key_machinery":"The central object is the inert second Higgs doublet $\\Phi_2$, which changes sign under a $Z_4$ discrete symmetry, has zero vacuum expectation value, and carries the only new Yukawa interaction $\\mathcal{L}_{\\mathrm{LFV}} = \\sqrt{2}\\rho_{\\mu\\tau}L_\\mu^L\\Phi_2\\tau_R + \\sqrt{2}\\rho_{\\tau\\mu}L_\\tau^L\\Phi_2\\mu_R + \\mathrm{h.c.}$, with $\\rho_{\\mu\\tau}=\\rho_{\\tau\\mu}=\\rho$ and all parameters real. Because $\\Phi_2$ has no quark or diagonal-lepton couplings, the neutral scalars $H$ and $A$ decay dominantly to $\\tau^\\pm\\mu^\\mp$, and the charged scalar $H^\\pm$ decays to $\\tau^\\pm\\nu_\\mu$ or $\\mu^\\pm\\nu_\\tau$, when the mass splittings are small. That same $\\rho$ generates the muon g-2 shift through one-loop $H$ and $A$ diagrams, corrects $\\tau\\to\\mu\\nu\\bar\\nu$ through tree-level $H^\\pm$ exchange, and shifts the $Z\\to\\tau^+\\tau^-$ and $Z\\to\\mu^+\\mu^-$ widths through loops; the condition $m_A>m_H$ is what makes the g-2 contribution positive.","core_discovery":"The paper's central claim is that a single real mu-tau Yukawa coupling $\\rho$ can account for both anomalies. The one-loop contribution to the muon anomalous magnetic moment, $\\delta a_\\mu = \\frac{m_\\mu m_\\tau \\rho^2}{8\\pi^2}\\left[\\frac{\\log(m_H^2/m_\\tau^2)-3/2}{m_H^2}-\\frac{\\log(m_A^2/m_\\tau^2)-3/2}{m_A^2}\\right]$, is positive only for $m_A>m_H$, while the tree-level charged-Higgs exchange in $\\tau\\to\\mu\\nu\\bar\\nu$ increases $\\Gamma(\\tau\\to\\mu\\nu\\bar\\nu)$ by $\\delta_{\\mathrm{tree}}=4m_W^4\\rho^4/(g^4 m_{H^\\pm}^4)$, pulling the $g_\\tau/g_e$ ratio toward its measured value. The complementary parameter dependence forces the simultaneous fit into a narrow band with $\\Delta m\\equiv m_A-m_{H^\\pm}<50$ GeV and $\\rho>0.36$. Applying LHC multilepton searches cuts this band to $m_H>560$ GeV and $\\rho\\gtrsim 0.68$, with the most restrictive bins being three-lepton and hadronic-tau plus two-lepton signal regions.","pith_inferences":["An unlisted constraint the paper does not apply is $\\tau\\to\\mu\\gamma$: the same $\\rho$ coupling that produces the g-2 and tau-decay corrections generates this radiative decay at one loop, and the surviving region (large $\\rho$, scalar masses near 500-800 GeV) is precisely where it can be sizable; imposing the current limit could shrink or eliminate the allowed band.","Because the surviving parameter space is so degenerate, a dedicated search for $\\tau\\mu$ plus missing transverse momentum, rather than generic multilepton bins, may be a more sensitive probe of this model than the searches used here.","The mechanism is generic to any new scalar with mu-tau flavour-violating Yukawa couplings: the same positivity condition $m_A>m_H$ and the same $\\rho$-versus-$\\Delta m$ tension would reappear in other models with an additional scalar doublet, so the 560 GeV lower bound is a useful target for model-building."],"forward_implications":["If the model is correct, the extra scalars should appear as a nearly degenerate trio with $m_A\\simeq m_{H^\\pm}$ and $m_A-m_H<50$ GeV, decaying mostly to $\\tau\\mu$, $\\tau\\nu$, and $\\mu\\nu$ final states.","The LHC multilepton limit is the controlling constraint: the paper finds $m_H>560$ GeV and $\\rho\\gtrsim 0.68$, so dedicated searches in three-lepton and hadronic-tau channels should see an excess or push the bound higher.","The model predicts exact equality of the semihadronic ratios, $(g_\\tau/g_\\mu)_\\pi=(g_\\tau/g_\\mu)_K=g_\\tau/g_\\mu$, because the new physics only enters the pure leptonic vertices.","Extrapolating the most sensitive signal regions, the paper estimates that $m_H\\lesssim 645$ GeV would be excluded at $2\\sigma$ with 139 fb$^{-1}$ of LHC data, and $m_H\\lesssim 700$ GeV with 300 fb$^{-1}$."],"supporting_citations":[{"why":"Introduced the Z4-symmetric mu-tau-philic inert doublet model and the tree-level tau-decay corrections that this paper starts from.","marker":"[23]"},{"why":"Gives the one-loop formula for the muon g-2 contribution from the mu-tau LFV Higgs couplings, which fixes the sign condition m_A > m_H.","marker":"[18]"},{"why":"Provides the five tau-decay lepton-universality ratios and their correlation matrix used in the chi-square fit.","marker":"[3]"},{"why":"Implements the vacuum stability, unitarity and perturbativity constraints and computes the oblique parameters applied to the parameter scan.","marker":"[24]"},{"why":"The multilepton search whose SR-A44 and SR-C18 signal regions give the strongest exclusions and force m_H > 560 GeV.","marker":"[33]"},{"why":"The two-lepton search used to constrain the model, giving a weaker m_H > 501 GeV bound.","marker":"[46]"},{"why":"The stau-pair search shown to give no constraint for m_H > 300 GeV because its signal regions require exactly two taus.","marker":"[37]"}],"fun_headline_variants":["Mu-tau Higgs doublet fits g-2 and tau anomalies","Second Higgs doublet with mu-tau coupling passes LHC only above 560 GeV","Narrow mass window for mu-tau Higgs solves two anomalies","Single mu-tau Higgs coupling explains g-2 and tau data"],"cache_read_input_tokens":14848,"weakest_assumption_plain":"The whole result depends on the assumption that the constraints listed in the scan are the only ones that matter; in particular, the radiative decay of a tau into a muon and a photon, generated by the very same $\\rho$ coupling, is not imposed, and applying it could shrink or remove the allowed region.","fun_headline_variants_meta":{"raw":{"variants":["Mu-tau Higgs doublet fits g-2 and tau anomalies","Second Higgs doublet with mu-tau coupling passes LHC only above 560 GeV","Narrow mass window for mu-tau Higgs solves two anomalies","Single mu-tau Higgs coupling explains g-2 and tau data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000682,"raw_usage":{"total_tokens":3115,"prompt_tokens":981,"completion_tokens":2134,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":597,"completion_tokens_details":{"reasoning_tokens":2056}},"tokens_in":597,"tokens_out":2134,"duration_ms":15433,"temperature":1.0,"reasoning_tokens":2056,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:03:13.068412+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the one-loop branching ratio $B(\\tau\\to\\mu\\gamma)$ for the surviving points with $\\rho>0.36$, $m_H\\in[560,800]$ GeV and $\\Delta m<50$ GeV, and compare it with the current experimental upper bound of a few $10^{-8}$; any point exceeding that bound falsifies the paper's claim that the two anomalies can be simultaneously explained in this model.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced the Z4-symmetric mu-tau-philic inert doublet model and the tree-level tau-decay corrections that this paper starts from."},{"cited_title":"Davidson, G","cited_arxiv_id":null,"evidence_quote":"Gives the one-loop formula for the muon g-2 contribution from the mu-tau LFV Higgs couplings, which fixes the sign condition m_A > m_H."},{"cited_title":"Eriksson, J","cited_arxiv_id":null,"evidence_quote":"Implements the vacuum stability, unitarity and perturbativity constraints and computes the oblique parameters applied to the parameter scan."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The multilepton search whose SR-A44 and SR-C18 signal regions give the strongest exclusions and force m_H > 560 GeV."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The two-lepton search used to constrain the model, giving a weaker m_H > 501 GeV bound."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The stau-pair search shown to give no constraint for m_H > 300 GeV because its signal regions require exactly two taus."}],"review_version":1}