{"id":"a7d9a554-42b0-429e-95ab-aa2fb486ab67","arxiv_id":"2411.19170","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"With the h to bb decay, the rare H to h gamma gamma process could give a 5-sigma signal at the HL-LHC for mH up to about 950 GeV in the type-II 2HDM.","lead":"The paper studies a rare heavy-Higgs decay, H to h plus two photons, in four types of two-Higgs-doublet models, and estimates whether the HL-LHC could detect it through the h to b-quark pair decay. It finds that the type-II model offers the best chances, with a 5-sigma signal possible for heavy Higgs masses up to about 950 GeV.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 5σ reaches in Figs. 7–9 are for the combined pp→H→b¯bγγ signal, which the paper itself says is dominated by H→hh→b¯bγγ; they never quantify a 5σ excess of H→hγγ events, so the abstract's claim that this rare decay is observable is not established.","rationale":"The reader's weakest_assumption concerns direct-search exclusions of the benchmark points; that is a legitimate external check, and it is not addressed in the paper for tanβ=0.2, 0.3, 1, 3, 5. However, my concern is more load-bearing because it targets the internal relation between the computed significance and the central claim. Even if every benchmark is allowed by Fig. 1, a 5σ observation of b¯bγγ from H production is not by itself a 5σ observation of H→hγγ when about 80% of the signal is the already-known H→hh cascade. The paper's own Fig. 5 and the accompanying text acknowledge the dominance of the cascade and the O(20%) size of the three-body contribution, but no statistical test is presented to show that this correction is detected. The one-loop amplitude itself appears carefully derived, with FeynCalc/Package-X cross-checks and a full amplitude in App. C, and the background simulation is detailed; the issue is the interpretation of the computed significance. A conditional verdict is appropriate: the authors should either compute the discriminating significance or soften the claim to the combined b¯bγγ channel. This does not change the reader's CONDITIONAL verdict, hence UNCHANGED.","tokens_in":17946,"tokens_out":11194,"duration_ms":112690,"concrete_test":"For each quoted benchmark and mass point, generate (i) the cascade-only sample H→hh→b¯bγγ and (ii) the full sample including H→hγγ, applying identical cuts. Build a binned likelihood in M(b¯b), M(γγ), M(hγγ), and leading-photon pT; profile the H→hh normalization over a conservative ±20% theory/systematics uncertainty, and compute the significance of the full signal relative to the cascade-only hypothesis. If the excess significance is below 5σ for the masses quoted in the abstract, revise the claim to \"combined final state\" instead of \"observation of H→hγγ.\"","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV defines the significance as NS/sqrt(NS+NB+(κNB)^2), where NS is the number of signal events after the b¯b, γγ, and hγγ invariant-mass cuts, and the same section states that \"the dominant contribution has its source in the cascade decay H→hh→γγb¯b\" while the H→hγγ box/reducible amplitudes add only an O(20%) excess over that cascade. Figures 7–9 therefore show the discovery potential for the combined b¯bγγ final state, not for the rare three-body decay H→hγγ. The abstract and conclusions nevertheless claim that H→hγγ \"could be observed at the 5σ level.\" To substantiate that claim one must discriminate the excess from the H→hh cascade, which is the process already constrained by the ATLAS/CMS searches quoted in Fig. 1. The paper does not do this: Fig. 5 displays the non-uniform ratio between the full and cascade-only distributions, but no shape-based test statistic, likelihood ratio, or excess-only significance is computed. The sentence \"neglecting the hγγ subprocess would amount to a sizable unitarity violation\" is not a statistical argument for detectability. Thus, even for benchmark points that pass the current H→hh→b¯bγγ limits, the quoted 5σ reach may only establish the known cascade, leaving the claimed observation of H→hγγ unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the decay H -> h gamma gamma in the four 2HDM types and argues that the b bbar gamma gamma final state can yield a 5-sigma discovery at HL-LHC for heavy-Higgs masses up to 950 GeV (type-II), 650 GeV (Lepton-Specific and Flipped), and 350 GeV (type-I). Section III derives the one-loop amplitude, presented in closed form in App. C, and tests the narrow-width approximation against it. Section IV defines benchmark points after imposing theoretical and some experimental constraints, simulates pp -> b bbar gamma gamma at 14 TeV with MadGraph/Pythia/MadAnalysis, applies b-tagging, photon isolation and invariant-mass cuts, and computes significances with N_S/sqrt(N_S+N_B+(kappa N_B)^2) for integrated luminosities between 300 and 3000 fb^-1. The paper further claims that the virtual H -> h gamma gamma contributions produce an O(20%) excess over the two-body H -> hh -> b bbar gamma gamma cascade and could help discriminate 2HDM types.","tokens_in":18281,"tokens_out":7187,"duration_ms":85046,"significance":"If the predictions hold, the channel would add a new final state to the LHC heavy-Higgs search program and could provide a complementary handle on 2HDM type discrimination. The paper has real strengths: the one-loop amplitude is documented in analytic form and cross-checked with FeynCalc and PackageX; the narrow-width approximation is tested rather than assumed; the collider study includes fake rates, b-tagging efficiencies, and detector emulation; and the parameter scan enforces perturbativity, vacuum stability, unitarity, oblique parameters and b -> s gamma constraints. The main caveats are that the quoted discovery significances are dominated by the already-constrained H -> hh -> b bbar gamma gamma cascade rather than by an isolated H -> h gamma gamma signal, and that the benchmark points are not explicitly checked against the observed limits in Fig. 1. Both issues are addressable by reframing the claims or adding the missing validation.","major_comments":[{"comment":"The benchmark points used for the 5-sigma mass reaches are not validated against the 95% CL limits shown in Fig. 1. The text quotes exclusions only for tan beta = 0.1 (type-I, Lepton-Specific and Flipped) and tan beta = 8 (type-II), but BMP1, BMP2 and BMP3 use tan beta = 0.3, 0.2 and 1 for type-I/LS/Flipped and tan beta = 3, 5 for type-II. Since sigma(pp -> H -> hh -> b bbar gamma gamma) grows toward small tan beta in the first three types, the quoted limits do not imply that the BMP cross sections are allowed; the paper should overlay the predictions for each benchmark point on the observed and expected limit curves in Fig. 1 and restrict the mass reach to regions not excluded by current data.","section":"Section IV, benchmark points and Fig. 1"},{"comment":"The significance plotted in Figs. 7-9 is computed from the total number of signal events in the combined b bbar gamma gamma final state, which the same section states is dominated by the cascade H -> hh -> b bbar gamma gamma, with the genuine H -> h gamma gamma virtual contributions adding only an O(20%) excess. The quoted 5-sigma reaches therefore establish, at best, sensitivity to the combined final state, most of which is the H -> hh process already constrained by the searches reproduced in Fig. 1; they do not by themselves establish that the rare decay H -> h gamma gamma is observable. To support the abstract's claim, the authors must either reformulate the claims as sensitivity to the b bbar gamma gamma final state with H -> h gamma gamma as a subdominant correction, or provide a shape-based test statistic, such as a binned likelihood over the M(h gamma gamma) or pT(gamma) distributions, that isolates the H -> h gamma gamma excess from the H -> hh cascade. Fig. 5 shows non-uniform ratios but no statistical separation, and the statement that neglecting the h gamma gamma subprocess would cause a sizable unitarity violation is not an argument for detectability.","section":"Section IV, 'Signal significance'; Abstract and Conclusions"}],"minor_comments":[{"comment":"The phrase 'neglecting the h gamma gamma subprocess would amount to a sizable unitarity violation' is imprecise: omitting a loop-induced contribution is a model-completeness issue, not a violation of S-matrix unitarity in the presented calculation; the wording should be revised.","section":"Section IV, 'Signal significance'"},{"comment":"The statement that the narrow-width approximation is accurate at the O(20%) level should specify the parameter region for which this holds, since the size of the box and reducible contributions will depend on M_H, tan beta and the 2HDM type.","section":"Section III, around Eq. (3.3)"},{"comment":"The pre-cut background cross sections in Table V are enormous, especially b bbar jj, but the paper does not provide a cut-flow table listing N_S and N_B after each stage of the analysis; such a table is needed to make the significances in Figs. 7-9 reproducible.","section":"Table V and Section IV"},{"comment":"The claim that Lepton-Specific and Flipped 2HDMs 'would be the most difficult versions to distinguish' because their sensitivities differ by about 5% is not a demonstrated discrimination statement; without an explicit hypothesis test, a 5% cross-section difference with 5% systematic uncertainty is not evidence of distinguishability.","section":"Section V, Conclusions"}],"recommendation":"major_revision","confidential_remarks":"The central physics calculation appears solid and the collider study is reasonably detailed, but the manuscript currently markets a rare-decay discovery while the quoted significance is dominated by the H -> hh cascade. If the authors reframe the main claim as combined sensitivity to the b bbar gamma gamma final state with an O(20%) H -> h gamma gamma contribution, and validate the benchmark points against Fig. 1, the result would be publishable; in its present form the abstract overstates what the analysis establishes."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: the 5σ reach numbers are for the combined b¯bγγ final state, where the H→hh cascade dominates by a factor of ~4. The paper never separates H→hγγ from the cascade, so the abstract's wording that the rare decay \"could be observed at 5σ\" is not supported by the analysis shown. That matters for how you read the headline, but the underlying calculation is real.\n\nWhat's new: a full one-loop amplitude for H→hγγ in the unitary gauge, with the exact expression in App. C, cross-checked with FeynCalc and PackageX. The narrow-width approximation is tested and found good to O(20%), and they give a clean argument for type discrimination via bottom/tau loops (Table IV). The collider study is competently done: background list, cuts, systematic uncertainty treatment, and it correctly notes the cascade dominates.\n\nSoft spots, in order:\n- The significance is NS/sqrt(NS+NB+...) for the combined signal. The figures are labeled H→hγγ but include the cascade. No shape-based test or likelihood ratio is used to isolate the three-body component. The non-uniform ratio in Fig. 5 is suggestive but not a test statistic. So a 5σ excess in Figs. 7-9 would not establish observation of H→hγγ; it could be entirely the already-constrained H→hh process. The conclusions should either be reframed to \"enhanced b¯bγγ final state\" or add a discrimination analysis.\n- Benchmark points are not validated against the Fig. 1 95% CL limits. The text only quotes exclusions for tanβ=0.1 and 8, but the BMPs use 0.2, 0.3, 1, 3, 5 with cos(β−α)=0.01. Given the cascade dominates the signal, those points might be excluded by the very same search that gives Fig. 1. This is checkable and must be shown.\n- The sentence about \"unitarity violation\" for neglecting the subprocess is just wrong—neglecting a subprocess in a rate estimate has nothing to do with S-matrix unitarity. It's a minor overclaim, but it should be deleted.\n- Ref [42] appeared before this version and studies the same decay; the note added acknowledges it but the paper doesn't compare numbers or quantify the difference. For a paper claiming a new final state, that's a serious gap in the literature treatment, even if concurrent.\n\nWho this is for: 2HDM practitioners who want the one-loop expression and a first look at the b¯bγγ final state. It deserves a serious referee, but the referee should ask for a major revision: reframe the reach claims, validate the benchmarks, and either compare with [42] or justify why the overlap doesn't change the results.","headline":"A real one-loop calculation and collider study, but the 5σ reach claims are for the cascade-dominated b¯bγγ final state, not for H→hγγ itself.","tokens_in":18829,"tokens_out":4056,"would_cite":true,"duration_ms":35321,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Rare decay $H\\to h\\gamma\\gamma$ with $h\\to b\\bar b$ could reach 5$\\sigma$ at HL-LHC for $M_H$ up to 950 GeV (type-II), 650 GeV (lepton-specific/flipped), 350 GeV (type-I) 2HDMs.","keywords":["two-Higgs-doublet model","rare Higgs decay","H to h gamma gamma","b bbar gamma gamma final state","High-Luminosity LHC","one-loop amplitude","2HDM type discrimination"],"falsifier":"Recompute the leading-order cross section $\\sigma(pp\\to H\\to hh\\to b\\bar b\\gamma\\gamma)$ for each benchmark point and compare it with the observed 95% CL upper limits in Fig. 1; any point above the black observed curve is already excluded and its corresponding 5-$\\sigma$ reach is invalid. A direct experimental check is a recast of the existing 13 TeV $b\\bar b\\gamma\\gamma$ search data that sets observed limits in the $M_{\\gamma\\gamma b\\bar b}$ distribution for the benchmark points.","tokens_in":17743,"feed_emoji":"⚛️","tokens_out":19310,"duration_ms":148199,"temperature":0.7,"pith_summary":"The paper studies a Higgs-sector process not previously proposed as a search channel: a heavy CP-even Higgs boson $H$ decaying into the 125 GeV Standard Model-like Higgs $h$ plus two photons, $H\\to h\\gamma\\gamma$, in a two-Higgs-doublet model (2HDM). If $h$ then decays to a bottom-quark pair, the visible final state is two $b$-jets and two photons, and the paper argues this rare decay could be discovered at the High-Luminosity LHC, reaching $5\\sigma$ for $M_H$ up to 950 GeV in type-II, 650 GeV in lepton-specific and flipped, and 350 GeV in type-I models at 3000 fb$^{-1}$. The interest is practical: the three-body decay adds roughly 20% more events to the same $b\\bar b\\gamma\\gamma$ final state produced by the two-body cascade $H\\to hh$, so a search that ignores it would mis-model the yield, and the extra component carries kinematic information that helps tell the four 2HDM types apart.","feed_headline":"Rare Higgs three-body decay could reach 5 sigma at HL-LHC","feed_subtitle":"The bottom-quark pair plus two-photon final state could expose heavy Higgs states up to 950 GeV.","key_machinery":"The machinery is the exact one-loop invariant amplitude for $H(p)\\to\\gamma(k_1)\\gamma(k_2)h(k_3)$, computed in the unitary gauge and reduced to Passarino-Veltman scalar functions; the full polarization-summed $|M|^2$ is given in Appendix C. The amplitude separates into box form factors for fermions, $W$ bosons and charged Higgses and a reducible-diagram form factor, with the type dependence carried entirely by coupling products such as $g_h^f g_H^f (g_\\gamma^f/q_f)^2$ listed in Table IV. In the alignment limit these products make bottom and tau loops $\\tan\\beta$-enhanced in the type-II, lepton-specific and flipped models, which is what turns the decay rate into a model-discriminating observable.","core_discovery":"The central claim is that the loop-level decay $H\\to h\\gamma\\gamma$ is a real, observable process in two-Higgs-doublet models with natural flavor conservation, and that its signature through $h\\to b\\bar b$ competes with the usual $H\\to hh\\to b\\bar b\\gamma\\gamma$ search. The one-loop amplitude is built from box diagrams with fermions, $W$ bosons and charged Higgses plus reducible diagrams, and the narrow-width approximation $\\Gamma(H\\to h\\gamma\\gamma)\\simeq\\Gamma(H\\to hh)\\,\\mathrm{BR}(h\\to\\gamma\\gamma)$ is accurate only at the 20% level, so the genuine three-body contribution cannot be discarded without biasing the rate and the kinematics. Because the bottom-quark and tau couplings differ among the four 2HDM types, the size of the $H\\to h\\gamma\\gamma$ rate discriminates type-I, type-II, lepton-specific and flipped models. With the analysis cuts of Section IV, the projected significance reaches $5\\sigma$ at $3000~\\mathrm{fb}^{-1}$ for $M_H$ up to $950$ GeV in the type-II model, $650$ GeV in the lepton-specific and flipped models, and $350$ GeV in the type-I model.","pith_inferences":["Editorial: the same $b\\bar b\\gamma\\gamma$ dataset could be recast to set model-independent limits on $\\mathrm{BR}(H\\to h\\gamma\\gamma)$ for any narrow heavy scalar, since the genuine three-body decay shifts the invariant-mass and $p_T$ distributions relative to the $hh$ cascade.","Editorial: a shape-based analysis of the two-photon $p_T$ or the $M_{\\gamma\\gamma}$ line shape could isolate the three-body component and verify the claimed O(20%) excess without assuming the total signal rate.","Editorial: because $h\\to\\tau\\tau$ could replace $h\\to b\\bar b$, the lepton-specific 2HDM could also be probed through $\\tau\\tau\\gamma\\gamma$ events, extending the discrimination into the lepton sector; this is not studied in the paper."],"forward_implications":["A $5\\sigma$ observation of the $b\\bar b\\gamma\\gamma$ resonance would be direct evidence of a second Higgs doublet and of the loop-level $H\\to h\\gamma\\gamma$ transition.","Searches for $H\\to b\\bar b\\gamma\\gamma$ should include the three-body component: at the benchmark points it raises the expected yield by about 20% over the $H\\to hh$ cascade and shifts the $p_T$ and invariant-mass distributions.","The size of the $H\\to h\\gamma\\gamma$ rate distinguishes 2HDM types, since bottom and tau loops are $\\tan\\beta$-enhanced in type-II, lepton-specific and flipped models but suppressed in type-I.","The channel is primarily a High-Luminosity probe: at 300 fb$^{-1}$ only the type-II model reaches about $3\\sigma$ for favorable benchmark points, while $5\\sigma$ requires 1000-3000 fb$^{-1}$ depending on the model.","For the favored type-II benchmark, the $5\\sigma$ reach extends to $M_H\\simeq 950$ GeV at 3000 fb$^{-1}$."],"supporting_citations":[{"why":"It supplies the current 95% CL observed limits on pp to H to hh to b bbar gamma gamma that constrain MH in each 2HDM type and define the surviving benchmark region.","marker":"[7]"},{"why":"It is the earlier dark-sector study of H to h gamma gamma that motivates applying the same three-body decay to heavy Higgs bosons in a 2HDM.","marker":"[14]"},{"why":"It defines the four natural-flavor-conserving 2HDM types and their Yukawa couplings used throughout the analysis.","marker":"[15]"},{"why":"It provides the parameterization of the 2HDM scalar potential in terms of v, the scalar masses, tan beta, cos(beta-alpha), and m12^2.","marker":"[18]"},{"why":"It is the model generator used to build the 2HDM UFO files and effective loop-induced vertices for event generation.","marker":"[21]"},{"why":"It is the generator used to produce the parton-level signal and background events for the collider study.","marker":"[22]"},{"why":"It provides the b to s gamma rate expressions used to constrain the charged-Higgs mass in the scanned parameter space.","marker":"[24]"},{"why":"It supplies the Passarino-Veltman scalar functions in terms of which the one-loop H to h gamma gamma amplitude is decomposed.","marker":"[28]"}],"fun_headline_variants":["5-sigma reach for H to h gamma gamma at HL-LHC","Heavy Higgs decay into h gamma gamma observable","Rare Higgs three-body decay could hit 5 sigma","New channel to spot heavy Higgs up to 950 GeV","H to h gamma gamma: probe for 2HDM types"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projected discovery masses assume that the benchmark parameter points used in the analysis are not already ruled out by current LHC searches for a heavy Higgs decaying to two lighter Higgs bosons; the paper verifies that assumption explicitly for only two of the parameter choices, so if any benchmark is in fact excluded, its quoted 5-sigma reach does not hold.","fun_headline_variants_meta":{"raw":{"variants":["5-sigma reach for H to h gamma gamma at HL-LHC","Heavy Higgs decay into h gamma gamma observable","Rare Higgs three-body decay could hit 5 sigma","New channel to spot heavy Higgs up to 950 GeV","H to h gamma gamma: probe for 2HDM types"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000982,"raw_usage":{"total_tokens":4207,"prompt_tokens":1021,"completion_tokens":3186,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":3112}},"tokens_in":637,"tokens_out":3186,"duration_ms":31781,"temperature":1.0,"reasoning_tokens":3112,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:27:38.513641+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the leading-order cross section $\\sigma(pp\\to H\\to hh\\to b\\bar b\\gamma\\gamma)$ for each benchmark point and compare it with the observed 95% CL upper limits in Fig. 1; any point above the black observed curve is already excluded and its corresponding 5-$\\sigma$ reach is invalid. A direct experimental check is a recast of the existing 13 TeV $b\\bar b\\gamma\\gamma$ search data that sets observed limits in the $M_{\\gamma\\gamma b\\bar b}$ distribution for the benchmark points.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the current 95% CL observed limits on pp to H to hh to b bbar gamma gamma that constrain MH in each 2HDM type and define the surviving benchmark region."},{"cited_title":"Gamma-ray signals from multicomponent scalar dark matter decays","cited_arxiv_id":"1909.13292","evidence_quote":"It provides the parameterization of the 2HDM scalar potential in terms of v, the scalar masses, tan beta, cos(beta-alpha), and m12^2."},{"cited_title":"Ciuchini, G","cited_arxiv_id":null,"evidence_quote":"It supplies the Passarino-Veltman scalar functions in terms of which the one-loop H to h gamma gamma amplitude is decomposed."}],"review_version":1}