{"id":"4e484c1b-d8de-48da-8d52-e4fec79bfbae","arxiv_id":"2411.13091","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A hypercentral quark model with a deep neural network plus particle swarm optimizer is used to predict masses and semileptonic decays of doubly heavy baryons, but missing parameters and code prevent reproduction.","lead":"The authors solve a six-dimensional Schrodinger equation for heavy baryons by training a neural network on shooting-method solutions and refining energies with particle swarm optimization, then quote a spectrum and semileptonic decay widths for doubly heavy Xi and Omega baryons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Mass-table validation is circular: the Killingbeck parameters α, β, τ are never reported and the text states single-heavy experimental masses were used to optimize the calculation, so the small errors in Tables 1–2 do not independently validate the predicted doubly-heavy masses.","rationale":"Both the reader and I flag the unreported potential calibration as the weakest point. The paper is a phenomenological calculation, not a formal proof, and the missing parameters make numerical reproduction impossible. The central predictions (doubly heavy masses) are extrapolations from a fit to the very data used for validation. My concrete check would settle the overfitting question by testing whether the calibration generalizes. I am not raising an external-consensus objection; the DNN/PSO machinery is irrelevant to the physics, but that is not a fatal flaw. The decay-width section's reliance on an external IW function is a secondary concern: it means the widths do not test the model, but the masses remain the primary output. Since the issues are fixable with additional reporting and a cross-validation test, the original CONDITIONAL verdict stands.","tokens_in":11374,"tokens_out":6476,"duration_ms":65233,"concrete_test":"Perform a leave-one-out cross-validation of the calibration: remove the Ω_b mass (6.045 GeV) from the fitting set, re-optimize α, β, τ (and any other free parameters) on the remaining single-heavy masses, then predict Ω_b. If the predicted Ω_b deviates from 6.045 GeV by more than the 0.03% quoted in Table 2, the reported agreement is overfit and the doubly-heavy predictions in Tables 3–4 require explicit error bars before they can guide experiment.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim—that the DNN+PSO solution of Eq. (5) yields trustworthy masses for unmeasured doubly heavy baryons—rests on potential parameters that are never given. Section 2 defines V(x)=αx²+βx−τ/x (Eq. 6), but Sections 3–5 report no numerical values for α, β, τ, no fitting procedure, and no uncertainty. The Introduction and Section 5 state that existing experimental single-heavy masses were used to optimize the calculation. Hence the <0.5% 'errors' in Tables 1 and 2 are calibration residuals on the training data, not independent tests. The DNN is trained on shooting-method solutions of the same equation, so any systematic error in the shooting method or in the potential form propagates directly into Tables 3–4. The reported MSE (≈6.4×10⁻⁵) is a training loss in energy, not a predictive uncertainty on masses. Consequently, the predicted values Ω_cc=3.750 GeV, Ξ_bc=7.050 GeV, Ω_bb=10.400 GeV have no stated error bar and no demonstrated extrapolation validity. A secondary issue: the decay-width calculation in Section 6 does not use the model wave functions; it simply adopts the IW function Eq. (18) from Ref. [69] with Λ_B=2.5–3.5 GeV, so Table 5 does not actually test the hypercentral/DNN model.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript applies a hypercentral quark model with a Killingbeck potential, V(x)=αx^2+βx-τ/x (Eq. 6), to S-, P-, and D-wave states of singly and doubly heavy Ξ and Ω baryons. The hyperradial Schrödinger equation (Eq. 5) is solved by generating training labels with a shooting method and fitting a two-hidden-layer deep neural network, then refining the energy estimates with particle swarm optimization. From the resulting eigenvalues the authors tabulate masses (Tables 1–4) and, using a single Isgur–Wise function from Ref. [69], compute semileptonic b→c decay widths and branching ratios (Tables 5–6). The central new numerical results are the unmeasured doubly heavy masses in Tables 3–4, e.g. Ω_cc=3.750 GeV, Ξ_bc=7.050 GeV, Ω_bb=10.400 GeV, and the decay widths in Table 5.","tokens_in":11610,"tokens_out":15648,"duration_ms":140806,"significance":"If the results were reliable, the unmeasured doubly heavy baryon masses and semileptonic widths would be directly useful for ongoing LHCb, ATLAS, and CMS searches, and the DNN+PSO workflow would offer a practical way to speed up eigenvalue calculations in the hypercentral quark model. The paper also has the merit of stating explicit predictions for many states not yet measured, and it compares those predictions with several other models. However, at present the central claims are not sufficiently supported: the potential parameters are not reported, the validation against known masses is partly circular, no predictive uncertainty is attached to the predicted masses, and the decay-width calculation does not use the model wave functions. These issues are correctable in principle, but they affect the main results.","major_comments":[{"comment":"The Killingbeck potential in Eq. (6) is defined in terms of constants α, β, and τ, but no numerical values are reported anywhere in Sections 2–5. The grand angular quantum number γ entering Eq. (5) is also never specified for the S-, P-, and D-wave states. Without these inputs, Eq. (7) cannot be solved and the numbers in Tables 1–4 cannot be reproduced. Since the Introduction states that existing experimental single-heavy-baryon masses were used to optimize the mass evaluations, the missing calibration protocol is a central gap rather than a presentation issue.","section":"Section 2, Eq. (6); Section 5"},{"comment":"The validation in Tables 1 and 2 is partly circular. The text states that existing experimental single-heavy-baryon masses help optimize the mass evaluations, so the quoted errors of 0.03–0.48% against those same states are calibration residuals, not independent tests. The closing paragraph of Section 7 similarly states that the validity of the scheme will be determined by consistency with existing experimental data, which is not an independent test. This does not by itself invalidate the predicted doubly heavy masses, but it means Tables 1 and 2 cannot serve as evidence for the reliability of Tables 3 and 4. An independent cross-check, such as withholding one benchmark state from the fit and predicting it, should be provided.","section":"Introduction, p. 2; Section 7; Tables 1 and 2"},{"comment":"The DNN is trained on energy labels produced by the shooting method, so it is a surrogate for that numerical procedure rather than an independent solution of Eq. (7). The MSE values reported in Section 7 are training and validation losses, not predictive uncertainties on baryon masses, and no error bars are attached to the predicted masses in Tables 3 and 4. The PSO refinement described in Section 4 is driven by the same target energies, so it cannot independently improve the eigenvalues. A comparison of the DNN+PSO output with direct shooting-method eigenvalues on a test set is needed to support the claimed gain in accuracy.","section":"Section 3; Section 4; Section 7; Tables 3 and 4"},{"comment":"The semileptonic decay calculation does not use the hypercentral wave functions or the DNN solution. The form factors are set to the single Isgur–Wise function η(ω) of Eq. (18), adopted from Ref. [69], with Λ_B in the range 2.5–3.5 GeV. Consequently, Tables 5 and 6 do not test the model presented in Sections 2–5. The Section 7 claim that the paper introduces a new form of the Isgur–Wise function is not supported by the text, since Eq. (18) is taken from Ref. [69]. The uncertainties in Table 5 also reflect only the Λ_B range and ignore the mass-model uncertainties.","section":"Section 6, Eqs. (17)–(18); Tables 5 and 6"},{"comment":"The numerical method is described too incompletely to reproduce. Section 3.1 specifies only that the network has two hidden layers, without giving the number of neurons, activation function, learning rate, number of epochs, or regularization. The input features of the dataset are not defined. Section 4 describes PSO verbally, without the position and velocity update equations, swarm size, or termination criterion. Because the DNN+PSO combination is the methodological novelty of the paper, these details are essential.","section":"Sections 3.1, 3.2, and 4"},{"comment":"The model neglects spin-dependent interactions, yet Tables 1 and 2 compare P-wave and D-wave predictions with specific experimental masses, for example the Σ_c P-wave state at 2.792 GeV. Without spin-orbit and tensor terms, the model cannot describe the multiplet structure of orbital excitations, and the choice of experimental partner for each computed state is not explained. This weakens the excited-state comparison and leaves the meaning of the P- and D-wave predictions in Tables 3 and 4 ambiguous.","section":"Section 2; Tables 1 and 2"}],"minor_comments":[{"comment":"The word 'Jacoobi' should be 'Jacobi'.","section":"Section 2, Eq. (1)"},{"comment":"The title contains 'widhts' and the text after Eq. (10) contains 'as fallows'; both should be corrected to 'widths' and 'as follows'.","section":"Section 6 title and text"},{"comment":"The entries '7.4457' and '11.1177' appear to have an extra digit; please correct them to the intended precision or explain the additional significant figure.","section":"Table 4"},{"comment":"Reference [28] is empty and should be filled in, since the Introduction cites it for the Ξ_cc lifetime.","section":"References"},{"comment":"Reference [48] has a garbled author list ('... Challenger, Mishraa'), which makes the citation difficult to locate.","section":"References"},{"comment":"The text says the calculation is performed close to the zero recoil point, but Eq. (16) integrates over the full range 1 ≤ ω ≤ ω_max; the justification for using Eq. (17) away from ω=1 should be stated.","section":"Abstract and Section 6"},{"comment":"The Ω_bb → Ω_bcℓν̄ row appears to lack the LFQM [40] entry; either provide it or mark it as not available.","section":"Table 6"},{"comment":"The reported validation MSE (6.196×10^-5) is smaller than the training MSE (6.690×10^-5) without comment; this unusual behavior should be discussed.","section":"Section 7"},{"comment":"The sentence claiming that the experimental Ξ_cc mass can be obtained in Table 3 is confusing because Table 3 has no experimental column; the match between 3.620 GeV and the LHCb value of 3.621 GeV should be stated explicitly.","section":"Section 5, Table 3"},{"comment":"No code or data repository is provided; for a computational paper, this makes independent verification difficult.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is not publishable in its current form because the main numerical results rest on unreported potential parameters and partly circular validation. I would be willing to review a revised version that reports α, β, τ and the fitting procedure, adds an independent validation test and uncertainty estimates, and either connects the decay calculation to the model wave functions or explicitly presents it as an Isgur–Wise based estimate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Frankly, this is a usable but not transformative paper. The authors solve the hypercentral Schrödinger equation with a Killingbeck potential using a deep neural network plus particle swarm optimization, and produce masses and semileptonic widths for doubly heavy Ξ and Ω baryons. The numbers are close to many other model predictions and to their own earlier work, which is reassuring but also means the DNN+PSO machinery doesn't change the physics. What's genuinely useful: a clear presentation of the model, comparison tables against several approaches, and an honest statement that single-heavy baryon masses were used to optimize the calculation. That honesty is the paper's best feature.\n\nThe big gap is reproducibility. The Killingbeck parameters α, β, τ are never reported. No fitting procedure, no uncertainties. Without them, Tables 1 and 2 are calibration residuals, not independent tests. And because the DNN is trained on shooting-method solutions of the same equation, it can't fix any systematic error in the potential or the solver. The MSE values quoted are training losses, not predictive errors. So the predicted doubly heavy masses in Tables 3 and 4 have no demonstrated error bars. The decay-width section is even more detached: the form factors are reduced to a single Isgur-Wise function taken from Ref. [69], so Table 5 mostly tests that IW parametrization, not the hypercentral/DNN model.\n\nThese are fixable in principle. The authors know their model; they can supply the missing parameters, give the fitting details, and clearly separate calibration from prediction. If they do, the paper becomes checkable. As it stands, a serious referee would need those artifacts before any independent verification.\n\nMinor: the text has a few typos and one empty reference [28], but nothing that changes the science. The claims are appropriately modest.","headline":"Solid phenomenological numbers, but missing potential parameters and a calibration-validation mix-up mean the predictions can't be independently checked as they stand.","tokens_in":12232,"tokens_out":2733,"would_cite":false,"duration_ms":26353,"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":"This paper claims that a hybrid deep-neural-network and particle-swarm-optimization solution of the hypercentral quark model predicts the masses of single and doubly heavy $\\Xi$ and $\\Omega$ baryons, and that semileptonic $b\\to c$ decays…","keywords":["doubly heavy baryons","hypercentral quark model","Killingbeck potential","deep neural network","particle swarm optimization","semileptonic decay widths","Isgur-Wise function","mass spectrum"],"falsifier":"Measure the ground-state mass of $\\Omega_{cc}$: the paper predicts 3.750 GeV, and a measurement far from this value would falsify the mass spectrum. A second, independent check is to measure $\\Gamma(\\Xi_{bc}\\to\\Xi_{cc}\\ell\\bar\\nu_\\ell)$: the predicted width is $4.47^{+1.0}_{-1.2}\\times10^{-14}$ GeV, so a measured width outside the $\\Lambda_B$-induced band of about $3.3$ to $5.5\\times10^{-14}$ GeV would falsify the decay treatment.","tokens_in":11090,"feed_emoji":"⚛️","tokens_out":9844,"duration_ms":91605,"temperature":0.7,"pith_summary":"The paper tries to show that a machine-learning solver can turn the hypercentral quark model into a practical predictor for heavy baryons that have not yet been seen. It combines a deep neural network with particle swarm optimization to find energy eigenvalues of the six-dimensional Schrödinger equation, then uses those eigenvalues to predict ground and excited masses of single and doubly heavy $\\Xi$ and $\\Omega$ baryons. On the decay side, it reduces the $b\\to c$ semileptonic form factors to one exponential Isgur–Wise function and computes widths and branching ratios for doubly heavy states. If the predictions are right, experiments would find $\\Omega_{cc}$ near 3.75 GeV, $\\Xi_{bc}$ near 7.05 GeV, and the double-bottom states near 10.2–10.4 GeV, with semileptonic widths in the $10^{-14}$ GeV range. These masses and widths are currently unmeasured, so the paper offers concrete targets for future searches.","feed_headline":"Predicted masses for five unseen baryons run 3.6-10.4 GeV","feed_subtitle":"A deep-learning quark-model solver reproduces the known Xi_cc and predicts masses and decay rates for the rest.","key_machinery":"The machinery is the reduction of the three-quark problem to a single hyperradial equation. Jacobi coordinates $\\rho,\\lambda$ combine into the hyperradius $x=\\sqrt{\\rho^2+\\lambda^2}$, and the wavefunction obeys a one-dimensional equation with centrifugal term $\\gamma(\\gamma+4)/x^2$ under the Killingbeck potential $V(x)=\\alpha x^2+\\beta x-\\tau/x$. A feed-forward neural network trained on a shooting-method dataset gives initial energy eigenvalues, and particle swarm optimization refines them; this is the step that produces the mass tables. For the decays, the load-bearing object is the universal Isgur–Wise function $\\eta(\\omega)=\\exp(-3(\\omega-1)m_{bb}^2/\\Lambda_B^2)$, with $\\Lambda_B\\in[2.5,3.5]$ GeV, which reduces all form factors near zero recoil to a single exponential and converts the widths into integrals of $\\eta^2(\\omega)$.","core_discovery":"The paper's central claim is that solving the hypercentral Schrödinger equation with a deep neural network plus particle swarm optimization yields the mass spectra of single and doubly heavy $\\Xi$ and $\\Omega$ baryons in their ground, $P$-wave, and $D$-wave states, and that the same model, combined with a single exponential Isgur–Wise function near zero recoil, determines the semileptonic $b\\to c$ decay widths of the doubly heavy states. The paper reports ground-state predictions $\\Xi_{cc}=3.620$ GeV (matching the measured value), $\\Omega_{cc}=3.750$ GeV, $\\Xi_{bc}=7.050$ GeV, $\\Omega_{bc}=6.900$ GeV, $\\Xi_{bb}=10.200$ GeV, and $\\Omega_{bb}=10.400$ GeV, plus excited-state masses up to 11.118 GeV for the $D$-wave $\\Omega_{bb}$. It also reports widths such as $\\Gamma(\\Xi_{bc}\\to\\Xi_{cc}\\ell\\bar\\nu_\\ell)=4.47^{+1.0}_{-1.2}\\times10^{-14}$ GeV and branching fractions near $10^{-2}$. These are the targets the paper offers to future experiments.","pith_inferences":["The same DNN+PSO pipeline would transfer to other three-body spectra if the Killingbeck constants $\\alpha,\\beta,\\tau$ were reported; as it stands, the unmeasured predictions cannot be independently reproduced from the paper alone.","Because the calculation neglects spin-dependent interactions, each quoted mass should be read as a spin-averaged multiplet center; comparing a specific observed resonance with the tables will require adding hyperfine splitting.","Ratios of the four semileptonic widths are more robust than the absolute widths, since the unknown potential and the $\\Lambda_B$ uncertainty partly cancel; those ratios are a sharper test of the universal Isgur–Wise ansatz than any single channel.","If one semileptonic width is measured, it fixes the effective $\\Lambda_B$ and turns the whole width table into definite predictions, which would also constrain the CKM element $V_{cb}$."],"forward_implications":["The unmeasured ground states $\\Omega_{cc}$, $\\Xi_{bc}$, $\\Omega_{bc}$, $\\Xi_{bb}$, and $\\Omega_{bb}$ become concrete mass targets at 3.750, 7.050, 6.900, 10.200, and 10.400 GeV, respectively, with the paper's $\\Xi_{cc}$ already reproducing the measured 3621 MeV state.","The $D$-wave $\\Omega_{bb}$ prediction near 11.118 GeV is a sharp discriminator: the paper puts it roughly 400 MeV above other model predictions, so a single future measurement would separate these schemes.","Semileptonic widths near $10^{-14}$ GeV and branching ratios near $10^{-2}$ suggest the $b\\to c$ modes are measurable enough to pin down $V_{cb}$ once doubly heavy baryons are produced.","Sub-percent agreement with the known single-heavy masses, together with the reported MSE values around $6\\times10^{-5}$, supports the model's internal consistency and makes its unmeasured predictions the ones to compare against."],"supporting_citations":[{"why":"Supplies the constituent quark masses and the decay-width methodology that this calculation inherits.","marker":"[55]"},{"why":"Provides the measured single-heavy baryon masses used as calibration benchmarks and error checks.","marker":"[57]"},{"why":"Establishes the hypercentral quark-model framework and the Killingbeck potential used in the Schrödinger equation.","marker":"[58]"},{"why":"Provides the measured $\\Xi_{cc}$ mass that the model reproduces as its main doubly heavy benchmark.","marker":"[27]"},{"why":"Supplies the universal Isgur–Wise form factor and the $\\Lambda_B\\in[2.5,3.5]$ GeV range used in all width calculations.","marker":"[69]"},{"why":"Provides the lifetimes $\\tau_{\\Xi_{bb}}$ and $\\tau_{\\Xi_{bc}}$ used to convert widths into branching ratios.","marker":"[76]"},{"why":"Provides the lifetimes $\\tau_{\\Omega_{bc}}$ and $\\tau_{\\Omega_{bb}}$ used in the branching-ratio table.","marker":"[77]"},{"why":"Gives the independent $\\Omega_{cc}-\\Xi_{cc}$ mass-difference prediction that the paper's 150 MeV result is compared with.","marker":"[79]"}],"fun_headline_variants":["Deep learning predicts masses for five double-heavy baryons","Neural network quark model predicts double-heavy baryon masses","Masses for five unseen baryons from a DNN-swarm quark model","AI quark model predicts Xi and Omega double-heavy baryon spectrum"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole prediction rests on the three constants of the chosen potential, which are fixed by matching measured single-heavy baryon masses but are never shown in the paper; if those constants do not carry over to the doubly heavy systems, every predicted mass and decay rate shifts.","fun_headline_variants_meta":{"raw":{"variants":["Deep learning predicts masses for five double-heavy baryons","Neural network quark model predicts double-heavy baryon masses","Masses for five unseen baryons from a DNN-swarm quark model","AI quark model predicts Xi and Omega double-heavy baryon spectrum"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000688,"raw_usage":{"total_tokens":3123,"prompt_tokens":952,"completion_tokens":2171,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":568,"completion_tokens_details":{"reasoning_tokens":2100}},"tokens_in":568,"tokens_out":2171,"duration_ms":15740,"temperature":1.0,"reasoning_tokens":2100,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:51:41.503445+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the ground-state mass of $\\Omega_{cc}$: the paper predicts 3.750 GeV, and a measurement far from this value would falsify the mass spectrum. A second, independent check is to measure $\\Gamma(\\Xi_{bc}\\to\\Xi_{cc}\\ell\\bar\\nu_\\ell)$: the predicted width is $4.47^{+1.0}_{-1.2}\\times10^{-14}$ GeV, so a measured width outside the $\\Lambda_B$-induced band of about $3.3$ to $5.5\\times10^{-14}$ GeV would falsify the decay treatment.","supporting_citations":[{"cited_title":"Ghalenovi, C","cited_arxiv_id":null,"evidence_quote":"Supplies the constituent quark masses and the decay-width methodology that this calculation inherits."},{"cited_title":"Ghalenovi, A","cited_arxiv_id":null,"evidence_quote":"Establishes the hypercentral quark-model framework and the Killingbeck potential used in the Schrödinger equation."},{"cited_title":"Aaij et al","cited_arxiv_id":null,"evidence_quote":"Provides the measured $\\Xi_{cc}$ mass that the model reproduces as its main doubly heavy benchmark."},{"cited_title":"Faessler, T","cited_arxiv_id":null,"evidence_quote":"Supplies the universal Isgur–Wise form factor and the $\\Lambda_B\\in[2.5,3.5]$ GeV range used in all width calculations."},{"cited_title":"Karliner and J","cited_arxiv_id":null,"evidence_quote":"Provides the lifetimes $\\tau_{\\Xi_{bb}}$ and $\\tau_{\\Xi_{bc}}$ used to convert widths into branching ratios."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the lifetimes $\\tau_{\\Omega_{bc}}$ and $\\tau_{\\Omega_{bb}}$ used in the branching-ratio table."},{"cited_title":"Ebert, R","cited_arxiv_id":null,"evidence_quote":"Gives the independent $\\Omega_{cc}-\\Xi_{cc}$ mass-difference prediction that the paper's 150 MeV result is compared with."}],"review_version":1}