{"id":"3e163dd1-42be-4346-b638-e3f800874f94","arxiv_id":"2506.23390","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"Lattice simulations with machine-learning fits suggest that a real singlet plus SU(2) doublet scalar model stays interacting, with enhanced propagators and a weakly momentum-dependent Yukawa vertex.","lead":"This paper runs lattice simulations of a scalar theory with a real singlet and an SU(2) doublet, including all quartic and Yukawa interactions, at 192 parameter points. It reports that the coupled theory may escape the triviality of a single scalar field and finds a 0+ spectrum spanning from micro-electronvolts to teraelectronvolts.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The non-triviality conclusion is not established: finite-L, one-cutoff dressing functions, partly ML-extrapolated, cannot demonstrate absence of triviality without a continuum limit.","rationale":"The reader's weakest_assumption (the ML fits) is real, and I agree that the deep-IR enhancement is extrapolated rather than directly measured. However, my primary objection is more fundamental: even a perfect estimate of the dressing function at one finite lattice spacing cannot establish 'absence of triviality,' because triviality is a property of the continuum limit. The paper has a single volume (18^4) and no continuum extrapolation, so the strongest claim is underdetermined. This is not a disagreement with current consensus; it is a missing control within the paper's own logic. I would keep the reader's REJECT verdict: the manuscript is better read as a pilot study with interesting data, not as a demonstration of a non-trivial scalar sector. I credit the large parameter scan and the transparent listing of ML coefficients, but those do not supply the missing continuum-limit evidence. The proposed test is standard and would settle whether the claim survives.","tokens_in":39774,"tokens_out":7560,"duration_ms":90892,"concrete_test":"Choose one representative parameter point (e.g., ms=1e-6, alpha=1, lambda=1, gamma=1) and compute the renormalized scalar dressing function (p^2+ms^2)Ds(p) at fixed physical p from raw lattice data on L=12,18,24,32, tuning the bare ms or beta so the Higgs-ball mass stays at 125 GeV and varying a by roughly a factor of 2. Extrapolate L->infinity and then a->0 with jackknife errors. If the extrapolated dressing function tends to 1, or the renormalized quartic/Yukawa couplings flow to zero, the 'absence of triviality' claim is unsupported; if it remains finite and >1, the central claim survives. Compare raw data at the lowest accessible p with the ML prediction (Eq. 20) at the same p to separate ansatz bias from physical IR enhancement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Section 3.4's inference that 'variety of scalar propagators ... strong sign of absence of triviality,' plus the abstract's 'continuum-like scaling window.' For this to hold, the renormalized dressing functions must remain non-trivial as the regulator is removed. The paper never provides that limit: all simulations are on L=18^4, one lattice spacing per parameter point, no L->infinity or a->0 extrapolation, no statistical errors, and no independent variation of the regulator at fixed physical input. A finite-cutoff lattice scalar theory is interacting even if its continuum limit is trivial, so enhanced dressing functions at a single cutoff cannot by themselves signal absence of triviality. Figure 3 only maps ln a against ln Lambda ms; it does not establish a scaling window. This problem is compounded by the polynomial ansatze (Eqs. 20-21): the 'infrared enhancement' appears below the directly accessed momenta, and the paper assigns ML-vs-lattice deviations to 'numerical (lattice) artifacts' (Section 3.4) without testing the ansatz. Even perfect fits would not convert a one-cutoff observation into a continuum non-triviality statement.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a Euclidean lattice model containing a real scalar singlet and an SU(2)-preserving complex doublet, with quartic and Yukawa interactions, on a single 18^4 lattice at 192 parameter points. Using multi-hit Metropolis and smeared operators, the authors compute Higgs-ball and scalar-ball masses, 0+ operator masses, field expectations, propagators, and the amputated Yukawa vertex. They fit degree-5 polynomial 'machine learning' functions to the lattice spacing, the scalar and Higgs propagators, and the vertex. The central claims are that the propagator dressing functions are infrared-enhanced, the Yukawa vertex depends only weakly on momenta, and this indicates absence of triviality and a 'continuum-like scaling window' in the parameter space.","tokens_in":40144,"tokens_out":5657,"duration_ms":61503,"significance":"If the central claims were established, the paper would be a significant non-perturbative indication that adding an SU(2) doublet and Yukawa couplings to a scalar singlet evades the triviality of pure φ^4 theory. The paper does provide a broad parameter scan, explicit ML coefficient tables in Appendices A–D that could serve as inputs to Dyson–Schwinger studies, and a careful sign-stability check of the action over a billion histories. However, the main physics conclusion is not backed by the presented evidence: the data are all at one lattice spacing per point with no continuum extrapolation, no statistical errors are reported, and the ML fits are fitted to the same data from which the physics is inferred. The claimed deep-infrared behavior is an extrapolation of polynomial ansatze rather than a measured lattice signal.","major_comments":[{"comment":"The claim that 'the variety of scalar propagators in the parameter space is a strong sign of absence of triviality' is not supported by the data presented. All simulations are on a single lattice volume L=18^4, with one lattice spacing per parameter point, and no L→∞ or a→0 extrapolation is performed. A lattice scalar theory with finite bare couplings is interacting at any finite cutoff even if its continuum limit is trivial, so dressing functions rising above 1 at a single cutoff cannot discriminate between trivial and non-trivial continuum limits. The abstract's 'continuum-like scaling window' likewise has no scaling analysis: Figure 3 merely plots ln a versus ln Λms and contains no demonstration of a scaling regime.","section":"Section 3.4, Eqs. (20)–(21), Figs. 14–17"},{"comment":"The machine-learning procedure is circular for the main conclusions. The functions in Eqs. (20) and (21) are fitted to the same lattice data points from which the infrared enhancement and weak momentum dependence of the vertex are inferred; no train/test split, cross-validation, or independent data set is described. The statement that 'the observed deviations stem from numerical (lattice) artifacts with a likely small contribution from the limitations of the representative functions' is an assumption, not a test. In particular, the claimed infrared enhancement below the lowest directly accessed momenta is an extrapolation of a degree-5 polynomial, and the flatness of the Yukawa vertex in Figs. 18–21 is largely a property of the fitted ansatz rather than of the raw lattice data.","section":"Section 3.4, after Fig. 17, and Eqs. (20)–(22)"},{"comment":"The scale-setting procedure relies on the unjustified identification of the lowest non-zero mass from the three-exponential fit of the Higgs-ball operator with the physical Higgs mass M_h = 125.09 GeV. The paper offers no evidence that this fitted mass indeed corresponds to the physical Higgs state rather than a lattice artifact or a different operator state. The subsequent conversion of lattice masses to physical units produces values spanning from μeV to hundreds of TeV (Figs. 5–11), which the paper attributes to 'extraordinary rise ... due to the low values of lattice spacing.' Without an independent scale-setting check or a continuum extrapolation, these mass assignments are not reliable.","section":"Section 3.1, Eqs. (6)–(7)"},{"comment":"No statistical errors are reported for any quantity. The Yukawa vertex is described as 'one of the noisiest quantities,' and the differences between lattice and ML vertex values in Figs. 18–21 are of comparable magnitude to the claimed weak momentum dependence. Without error bars, the classification of expectation values in Figs. 12–13 and the conclusion that the vertex is 'weakly depending upon the field momenta' cannot be assessed. The paper should report at least jackknife or bootstrap errors for the underlying correlation functions and propagated errors for the dressing functions and vertex.","section":"Section 2, Table 1, and Section 3.5"}],"minor_comments":[{"comment":"The vertex function FΓ is written with β among its variables even though β is held fixed throughout the study; the notation should be clarified so that the reader knows which parameters are actually varied.","section":"Eq. (22)"},{"comment":"The renormalization conditions do not state the numerical value of the renormalization point µ used in the fits; the scheme is not fully reproducible without this information.","section":"Eqs. (14)–(15)"},{"comment":"The legends such as '1,1' and the axis labels like 'mass [GeV]+0' are ambiguous; the meaning of the plotted curves and the notation for parameter tuples should be defined explicitly.","section":"Figs. 5–11"},{"comment":"The sentence 'The simulations and calculations are performed in C++ environment, and ROOT CERN is used to produce the diagrams' is an implementation detail that would be better placed in a footnote than in the main text.","section":"Section 2"},{"comment":"There are several typographical and grammatical errors, including 'the theory remain interactive' in Section 3.4 (should be 'remains interactive') and inconsistent use of ligatures in words such as 'effects'; a careful proofread is needed.","section":"Throughout"}],"recommendation":"reject","confidential_remarks":"The paper is best read as a preliminary exploratory survey. The central physics claim (absence of triviality) is not supported because the lattice results are at a single cutoff and the ML fits are not validated against independent data. Addressing these points would require new simulations at multiple lattice spacings and volumes, a proper error analysis, and a ML validation protocol—efforts beyond a routine revision. The manuscript could be reconsidered if reframed as a finite-cutoff data resource with raw data and analysis code made available, but in its current form the main conclusions outrun the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a new lattice parameter scan for a scalar singlet plus SU(2) doublet model, and it is worth reading as an exploratory study. But the abstract's nontriviality claim — that the model evades scalar triviality — is not backed by the data shown. The reader's skeptic is right; the stress-test note holds up on reading.\n\nWhat is actually new: nobody has run this particular scalar-only model with all quartic plus singlet-doublet Yukawa interactions, so the 192-point scan of propagators, the vertex, and the 0+ operators is a new dataset. The paper is transparent about the action, the update algorithm, and the scale-setting choice, and the ML fit coefficients are tabulated in appendices, which is a reproducible artifact even without raw data. The sign-of-action discussion is honest and checked.\n\nThe soft spot is load-bearing. Section 3.4 concludes absence of triviality from dressing functions that are 'significantly above 1,' but at the lowest directly accessed momenta they are typically 1.05–1.2, and the deep-infrared enhancement comes from the polynomial ansatze in Eqs. 20–21. Those fits are trained on the same lattice data and then deviations are assigned to 'numerical (lattice) artifacts' without an independent test. There are no statistical errors, one lattice size (18^4), no continuum extrapolation, and no independent regulator variation; a finite-cutoff scalar theory is interacting even if its continuum limit is trivial. So 'variety of scalar propagators' cannot carry the nontriviality conclusion. The same problem affects the flat Yukawa vertex and the claimed continuum-like scaling window in Fig. 3, which is only ln a against ln(Λ ms). The 0+ masses from meV to TeV are dominated by the scale-setting convention that fixes the lowest Higgs-ball mass to 125.09 GeV; I would not quote those as physical predictions.\n\nNone of this is a takedown of the numerical work. The paper has real exploratory value: the model is new, the scan is broad, and the ML parametrizations could feed DSE studies if labeled as parametrizations, not measurements. I agree with the reader that it should be reframed as a pilot study and sent back for major revision: add error bars, at least one second lattice size or a genuine a→0 statement, an independent scale observable, and validation of the ML fits on held-out momenta or parameter points.\n\nWho this is for: lattice scalar practitioners and BSM model-builders who want an existing dataset and ansatze for a two-scalar model. I would not cite the triviality claim, but I would send it to a serious referee rather than desk-reject: the model scan is not available elsewhere, and a revision could be valuable.","headline":"A genuinely new exploratory lattice scan of a scalar singlet plus SU(2) doublet model whose central nontriviality claim rests on single-cutoff, ML-extrapolated dressing functions; right to send back for major revision, not to desk-reject.","tokens_in":40582,"tokens_out":3519,"would_cite":false,"duration_ms":39831,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["11.15.Ha","12.60.Fr"],"model":"deepseek-v4-flash","headline":"This paper claims that a real scalar singlet coupled to an SU(2)-symmetric complex doublet through quartic and Yukawa interactions remains an interacting quantum field theory, escaping the triviality of pure phi^4 theory, as shown by…","keywords":["scalar singlet","SU(2) doublet","lattice scalar field theory","triviality","Yukawa vertex","field propagators","machine learning correlation functions","0+ states"],"falsifier":"Run the same parameter points on a larger lattice, say $24^{4}$ or $32^{4}$, and check whether the propagator dressing functions computed directly at the lowest accessible momenta still rise significantly above 1 and whether the Yukawa vertex remains flat; if the enhancement shrinks or the vertex develops momentum dependence as the volume grows, the paper's central claim fails.","tokens_in":39545,"feed_emoji":"⚛️","tokens_out":7073,"duration_ms":73790,"temperature":0.7,"pith_summary":"This paper tries to establish that a scalar field theory containing a real singlet and an SU(2) symmetric complex doublet, with all quartic couplings and a Yukawa coupling between them, is not trivial. Using lattice simulations at 192 points in parameter space and supervised machine learning to fit the correlation functions, the authors report scalar and Higgs propagators whose dressing functions rise significantly above 1, infrared enhancement relative to tree level, and a scalar-Yukawa vertex that depends weakly on field momenta across a wide range. Such behavior, if correct, means the triviality of the standalone $phi^{4}$ interaction is mitigated once the singlet is coupled to the doublet, making the model a viable nonperturbative window into Higgs-portal and dark-matter scalar sectors. The paper also maps a rich 0+ spectrum, from sub-eV to hundreds of TeV, with a relative scarcity of states near hundreds of GeV.","feed_headline":"Singlet-plus-doublet scalars dodge phi^4 triviality","feed_subtitle":"Propagators rise in the infrared, the Yukawa vertex stays flat—signs the theory is genuinely interacting.","key_machinery":"The central objects are the renormalized scalar and Higgs propagators and the amputated scalar-Yukawa vertex. The propagators are written as tree-level forms plus polynomial corrections in the couplings, the momentum, and the scalar mass, while the vertex is written as the Yukawa coupling times a polynomial in the momentum invariants and the couplings. Supervised machine learning fits these polynomial forms to the lattice data; the fitted functions supply the deep-infrared behavior where the lattice itself cannot resolve it, and their outputs are what show dressing functions above 1 and a flat vertex. The paper's claim that the theory remains interacting stands on those fitted functions.","core_discovery":"The core discovery claimed is that the theory remains interacting in the presence of all allowed SU(2)-preserving scalar interactions. In the adopted renormalization scheme, the renormalized scalar and Higgs propagators are enhanced compared to their tree-level forms, with dressing functions rising markedly above 1, particularly for the lightest scalar masses. The variety of propagator shapes across the parameter space is interpreted as a strong sign of the absence of triviality, meaning the Yukawa and doublet interactions prevent the pure $phi^{4}$ sector from becoming free. The amputated scalar-Yukawa vertex, computed over the full two-dimensional momentum plane, is found to depend only weakly on the field momenta; combined with the propagator enhancement, this indicates a genuinely interacting scalar sector. In addition, the spectrum of 0+ operators ranges from ultralight masses to hundreds of TeV, with a notable scarcity around hundreds of GeV.","pith_inferences":["Editorial inference: the polynomial machine-learning fits are only as trustworthy as their extrapolation into the deep infrared, so a direct test on a larger lattice that resolves those momenta without fitting would either confirm the infrared enhancement or expose it as a fitting artifact.","Editorial inference: the flatness of the Yukawa vertex could be a consequence of the renormalization condition at the chosen scale, and checking the vertex at several renormalization points or against a perturbative comparison would show whether the flatness is physical or scheme-induced.","Editorial inference: the reported scarcity of 0+ states near hundreds of GeV may depend on the choice of the Higgs-ball operator and the smallest nonzero lattice mass for setting the scale, so an independent scale-setting operator would test whether the gap is a genuine spectral feature."],"forward_implications":["If the claim holds, a scalar singlet coupled to a doublet provides a nonperturbative counterexample to the expectation that scalar-only theories are trivial, opening a path to interacting Higgs-portal models beyond perturbation theory.","The machine-learned propagator and vertex functions give a parameter-space map that can be used as input to nonperturbative bound-state calculations, since they reduce the need for arbitrary ansatze.","The reported 0+ spectrum, spanning from ultra-light to hundreds of TeV with a gap near hundreds of GeV, would give specific, testable predictions for scalar extensions of the Standard Model and for ultralight dark matter searches.","The observed classification of field expectation values into distinct trajectories, with no conclusive phase transition, implies that the model has structural changes in parameter space that future larger-lattice studies can pin down."],"supporting_citations":[{"why":"Establishes that pure phi^4 theory is trivial, the problem the paper's model is designed to escape.","marker":"[11]"},{"why":"Supplies the propagator renormalization scheme and the definition of the amputated Yukawa vertex that the paper adopts.","marker":"[12]"},{"why":"Provides the earlier nonperturbative study of an interacting scalar-Higgs system and the Higgs-ball idea used for scale setting.","marker":"[10]"},{"why":"Provides the lattice action and Monte Carlo methodology used for the simulations.","marker":"[8]"},{"why":"Provides the supervised machine learning approach used to fit the propagators, scale, and vertex.","marker":"[24]"},{"why":"Gives the physical context for interpreting the reported ultra-light 0+ masses as ultralight scalar candidates.","marker":"[27]"}],"fun_headline_variants":["Scalars resist triviality via Yukawa and doublet sectors","IR-enhanced propagators signal genuine scalar interaction","Yukawa vertex flat, propagators rise: interacting scalars","Doublet-singlet model escapes phi^4 free-field fate","Nonperturbative scalars: no triviality, stable Yukawa"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The deep-infrared behavior of the propagators and the flatness of the Yukawa vertex come from the machine-learning fits of Eqs. (20)-(22), and the paper assumes those polynomial forms are faithful and that all deviations from the lattice data are numerical artifacts.","fun_headline_variants_meta":{"raw":{"variants":["Scalars resist triviality via Yukawa and doublet sectors","IR-enhanced propagators signal genuine scalar interaction","Yukawa vertex flat, propagators rise: interacting scalars","Doublet-singlet model escapes phi^4 free-field fate","Nonperturbative scalars: no triviality, stable Yukawa"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000724,"raw_usage":{"total_tokens":3215,"prompt_tokens":881,"completion_tokens":2334,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":497,"completion_tokens_details":{"reasoning_tokens":2248}},"tokens_in":497,"tokens_out":2334,"duration_ms":17089,"temperature":1.0,"reasoning_tokens":2248,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:44:21.953194+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same parameter points on a larger lattice, say $24^{4}$ or $32^{4}$, and check whether the propagator dressing functions computed directly at the lowest accessible momenta still rise significantly above 1 and whether the Yukawa vertex remains flat; if the enhancement shrinks or the vertex develops momentum dependence as the volume grows, the paper's central claim fails.","supporting_citations":[{"cited_title":"Φ 4 theory is trivial","cited_arxiv_id":null,"evidence_quote":"Establishes that pure phi^4 theory is trivial, the problem the paper's model is designed to escape."},{"cited_title":"Two- and three-point functi ons in Landau gauge Yang-Mills-Higgs theory","cited_arxiv_id":null,"evidence_quote":"Supplies the propagator renormalization scheme and the definition of the amputated Yukawa vertex that the paper adopts."},{"cited_title":"Spectroscopic analysis of t he phase dia- gram of Yang-Mills-Higgs theory","cited_arxiv_id":null,"evidence_quote":"Provides the earlier nonperturbative study of an interacting scalar-Higgs system and the Higgs-ball idea used for scale setting."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the lattice action and Monte Carlo methodology used for the simulations."},{"cited_title":"Machine Learning - An Algorithmic Perspective","cited_arxiv_id":null,"evidence_quote":"Provides the supervised machine learning approach used to fit the propagators, scale, and vertex."},{"cited_title":"Ostriker, Scott Tremaine, and Edwa rd Wit- ten","cited_arxiv_id":null,"evidence_quote":"Gives the physical context for interpreting the reported ultra-light 0+ masses as ultralight scalar candidates."}],"review_version":1}