{"id":"92bace4a-1fed-497a-970b-f5206130e8d0","arxiv_id":"2506.09568","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":9,"one_line_summary":"A Gaussian-process reconstruction of Hubble data is used to pick power-law and exponential f(L_m) models, but the dynamical-system proof of a stable late-time attractor is invalid for these models because f_R is constant, making a key denominator zero.","lead":"The authors reconstruct the matter-side function f(L_m) in f(R,L_m) gravity from Hubble data and then test two proposed forms with dynamical systems. The late-time attractor they report is generic to their variables rather than specific to the models, and the stability derivation divides by zero for the models considered.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"For both proposed models f(R,L_m)=R/2+αf(L_m), f_R=1/2 so Eq. (26) divides by b=0; moreover Eq. (11) gives 3H²=(2b−1)αL_m^b, negative for their positive α and b<1/2, so the claimed late-time attractors are not solutions of the field equations.","rationale":"The GP reconstruction pipeline itself uses standard publicly available code and data, and that portion is not the main problem. The critical issue is internal consistency of the proposed models with the f(R,L_m) field equations. The reader correctly identified b=0 in Eq. (26); my independent check makes the failure stronger: even before writing the autonomous system, the proposed power-law and exponential forms with the stated signs cannot satisfy Eq. (11). This is an internal inconsistency, not a disagreement with an external consensus. Since the central claim of late-time attractors depends on these equations, rejection is warranted. No machine-checked proof or parameter-free derivation supports the dynamical conclusion.","tokens_in":11432,"tokens_out":13319,"duration_ms":138538,"concrete_test":"Compute b=f_RR R/f_R for f=R/2+\\alpha L_m^b (this gives b=0) and substitute f=R/2+\\alpha L_m^b into the 00-component of Eq. (4) at z=0 with the paper's \\alpha and b_1=0.018-0.025. If the algebra gives 3H^2=(2b_1-1)\\alpha L_m^b and the right-hand side is negative, then the model is not a solution; the eigenvalues of points B and P2 should then be recomputed from the actual constrained system without Eq. (26).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central stability result rests on Eq. (26), \\dot R/(6H^3)=xz/b with b=d\\ln f_R/d\\ln R=f_RR R/f_R. For both f_1 and f_2, the action is f(R,L_m)=R/2+f(L_m), so f_R=1/2 and f_RR=0, hence b=0 and Eq. (26) is undefined; equations (29) and (31), and therefore critical points B and P2, do not follow from the stated action. This is not an isolated technical slip: using the same f_R=1/2 in the Friedmann equation (11) gives 3H^2=(2b_1-1)\\alpha L_m^{b_1}. With the paper's positive \\alpha=(1-\\Omega_{m0})/[(1+2b_1)(6H_0^2)^{1-b_1}] and b_1 in [0.018,0.025], the right-hand side is negative for positive L_m, so no real cosmological solution exists for the power-law model. The exponential model with b_2>0 has the same structural problem in its Friedmann constraint. The claimed stable attractors therefore do not belong to the space of solutions of the theory under study.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to reconstruct the matter-Lagrangian function f(L_m) in f(R,L_m) gravity using Gaussian-process regression applied to Hubble data, assuming the separable form f(R,L_m)=R/2+f(L_m). From the reconstructed 1-sigma band the authors select a power-law model f1(L_m)=α L_m^{b1} with b1 in [0.018,0.025] and a square-root-exponential model f2(L_m)=α L_{m0}(1-e^{-b2√(L_m/L_{m0})}) with b2 in [2.3,3.0]. They then perform a dynamical-system analysis with variables x,y,z,u and report stable late-time attractors for both models, concluding that the models can explain cosmic acceleration without a cosmological constant.","tokens_in":11876,"tokens_out":4752,"duration_ms":52064,"significance":"The observational reconstruction pipeline is a strength: the authors use public cosmic-chronometer, supernova, and BAO data, propagate errors through Monte Carlo draws, and explicitly state their assumed action and parameter ranges. If the dynamical-system analysis were correct, the paper would provide a data-driven route to viable f(R,L_m) models. However, the central mathematical step is invalid for the models under study: the quantity b=d ln f_R/d ln R vanishes identically because f_R=1/2, so the autonomous equations and all reported critical points do not follow. In addition, the Friedmann constraint makes the reconstructed models inconsistent as real cosmological solutions. These are load-bearing failures, not presentation issues, and they invalidate the paper's principal claim.","major_comments":[{"comment":"The paper closes the dynamical system using ∂R/(6H^3)=xz/b with b=d ln f_R/d ln R=f_RR R/f_R. For both models f(R,L_m)=R/2+α f(L_m), so f_R=1/2 and f_RR=0, hence b=0 and Eq. (26) is undefined. Consequently the autonomous systems in Eqs. (29) and (31), and the critical points B and P2 derived from them, are not consequences of the stated actions. This is the load-bearing step on which the claimed late-time attractors rest.","section":"Section V, Eq. (26)"},{"comment":"With f_R=1/2 and f(R,L_m)=R/2+f(L_m), the first Friedmann equation reduces to 3H^2=2 f_{L_m} ρ - f(L_m) when L_m=ρ. For the power-law model f=α L^{b1}, this gives 3H^2=(2b1-1)α L^{b1}. With the paper's positive α=(1-Ω_{m0})/[(1+2b1)(6H_0^2)^{1-b1}] and b1 in [0.018,0.025], the right-hand side is negative for positive L_m, so no real Friedmann solution exists. The exponential model has the same structural inconsistency: at L_m=L_{m0}, 3H^2=α ρ_0 [e^{-b2}(b2+1)-1], which is negative for b2 in [2.3,3.0]. The reconstructed models are therefore not solutions of the field equations they are claimed to describe.","section":"Section II, Eq. (11) and Section IV"},{"comment":"The reconstruction anchors f(z=0) by imposing ΛCDM with f_{L_m}=0 at z=0, and the two functional forms with their parameter ranges are selected after inspecting the reconstructed 1σ band. This makes the subsequent 'viability' claim partly circular: the models are post-hoc fits to a band that already contains ΛCDM, rather than independent predictions. The abstract's statement that the models explain acceleration 'without explicitly invoking a cosmological constant' is therefore overstated, since a ΛCDM-like initial condition is built into the reconstruction.","section":"Section IV, Eq. (17) and Fig. 2"}],"minor_comments":[{"comment":"The term (g_{\\mu\\nu}\\nabla^\\mu\\nabla^\\mu - \\nabla_\\mu\\nabla_\\nu)f_R appears to contain a typo: the second operator should be the d'Alembertian, not a covariant derivative with a repeated index.","section":"Section II, Eq. (4)"},{"comment":"The exponential-model exponent is denoted b2 in Section IV and in the abstract but sometimes as β in the text and in Section V; the notation should be unified.","section":"Throughout"},{"comment":"The table caption mixes Cosmic Chronometer and BAO data but does not clearly separate the two subsets; labeling each block would improve reproducibility.","section":"Table I"},{"comment":"The evolution equation for z is model-independent and identical in both systems, so the late-time critical point z=2 reflects only the kinematic identity R=6(\\dot H+2H^2); it does not test the reconstructed f(L_m) forms.","section":"Section V, Eqs. (29) and (31)"}],"recommendation":"reject","confidential_remarks":"The paper's main result cannot be salvaged without replacing the models or the dynamical-system derivation: b=0 is not a technical slip but a direct consequence of the assumed action, and the Friedmann equation yields negative H^2 for the quoted parameter ranges. A revision would require a fundamentally different f(R,L_m) form with f_RR≠0 and a new reconstruction and stability analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the GP reconstruction of f(L_m) in f(R,L_m) gravity is a reasonable extension of the Cai et al. pipeline, and seeing the ΛCDM curve sit inside the 1σ band is a good sanity check. But the dynamical stability analysis is invalid, and the two proposed models don't actually satisfy the field equations with the parameters quoted.\n\nThe problem isn't subtle. Both models have f(R,L_m)=R/2+f(L_m), so f_R=1/2 and f_RR=0. The variable b=f_RR R/f_R in Eq. (26) is identically zero, making \\dot R/(6H^3)=xz/b undefined. The autonomous systems (29) and (31) do not follow. That alone kills the stability conclusion.\n\nThe stress-test note goes further, and it checks out: plugging the power-law model into the first Friedmann equation gives 3H^2=(2b_1-1)α L^{b_1}. With their positive α and b_1∈[0.018,0.025], the right-hand side is negative for positive L. The exponential model has the same structure: 3H^2=αL_0[(1+b_2 s)e^{-b_2 s}-1], negative for all s>0. So the claimed stable late-time attractors are not points in the solution space.\n\nWhat the paper does do well is the reconstruction itself: it is a clean data-driven exercise, and the authors are transparent that the functional forms are borrowed from f(T) work. But the parameter ranges are selected after seeing the reconstructed band, so the consistency check is circular. The initial condition f(z=0)=6H_0^2(Ω_m0-1) also assumes ΛCDM with f_Lm=0, which the proposed models do not respect.\n\nRecommendation: don't send this to a referee as is. The error is decisive. To salvage the idea, the authors would need a non-trivial f(R) sector with b≠0 and a normalization that yields a positive right-hand side in the Friedmann equation. As it stands, it's a reject.","headline":"A clean reconstruction pipeline undermined by a division by zero and by models that violate the Friedmann equation.","tokens_in":12357,"tokens_out":9142,"would_cite":false,"duration_ms":86158,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["83D05","83F05"],"pacs":["04.50.Kd","98.80.-k"],"model":"deepseek-v4-flash","headline":"This paper reconstructs the matter part of f(R, matter-Lagrangian) gravity from Hubble data and reports that the two resulting models both end in a stable, accelerating de Sitter phase without a cosmological constant.","keywords":["f(R,L_m) gravity","Gaussian process reconstruction","late-time cosmic acceleration","dynamical system analysis","matter Lagrangian","Hubble parameter data","power-law model","square-root exponential model"],"falsifier":"One concrete check is to derive the autonomous equations directly from the exact Friedmann equations for $f(R,\\mathcal{L}_m)=R/2+\\alpha\\mathcal{L}_m^{b_1}$ and for the square-root-exponential form, without using the $b=d\\ln f_R/d\\ln R$ step, and then compute the eigenvalues of the late-time critical point; if that point is not $(z,u)=(2,0)$ with all eigenvalues negative, the paper's stability claim fails.","tokens_in":11182,"feed_emoji":"🌌","tokens_out":15265,"duration_ms":131605,"temperature":0.7,"pith_summary":"This paper tries to establish that the function $f(\\mathcal{L}_m)$ in $f(R,\\mathcal{L}_m)$ gravity can be reconstructed directly from Hubble-expansion data, without fixing its functional form in advance, and that the reconstructed forms lead to viable cosmic evolution. The authors apply a non-parametric Gaussian-process reconstruction to cosmic-chronometer, supernova, and BAO data, producing two models compatible with the 1-$\\sigma$ region: a power law $f_1(\\mathcal{L}_m)=\\alpha\\mathcal{L}_m^{b_1}$ with $b_1\\in[0.018,0.025]$, and a square-root-exponential form $f_2(\\mathcal{L}_m)=\\alpha\\mathcal{L}_{m0}(1-e^{-b_2\\sqrt{\\mathcal{L}_m/\\mathcal{L}_{m0}}})$ with $b_2\\in[2.3,3.0]$. A dynamical-system analysis of the reconstructed models finds a stable late-time critical point in both cases, with deceleration parameter $q=-1$ and effective equation of state $w_{\\rm eff}=-1$. If correct, this gives a path to late-time cosmic acceleration that does not require introducing a cosmological constant by hand.","feed_headline":"Hubble data pick two gravity models that stably accelerate","feed_subtitle":"A data-driven reconstruction of modified gravity ends in a stable accelerating phase without a cosmological constant","key_machinery":"The argument runs through a two-stage pipeline. In the reconstruction stage, the Friedmann equation is turned into an iterative finite-difference recurrence for $f(z)$ in terms of the Gaussian-process-reconstructed $H(z)$ and $H'(z)$ using a squared-exponential covariance kernel, so $f(\\mathcal{L}_m)$ is read off from the data rather than chosen from an ansatz. In the stability stage, the cosmological equations are written as an autonomous system in the dimensionless variables $x=\\dot{f}_R/(H f_R)$, $y=f/(6H^2 f_R)$, $z=R/(6H^2)$, and $u=f_{\\mathcal{L}_m}\\rho/(3H^2 f_R)$, and the critical points are classified by the eigenvalues of the Jacobian matrix. The central object is the de Sitter point $(z,u)=(2,0)$: it appears for both reconstructed models with all negative eigenvalues, giving $q=-1$ and $w_{\\rm eff}=-1$.","core_discovery":"The central claim is that for a separable action $f(R,\\mathcal{L}_m)=R/2+f(\\mathcal{L}_m)$ with $\\mathcal{L}_m=\\rho$ in a pressureless universe, Gaussian-process reconstruction of $H(z)$ and $H'(z)$ from cosmic-chronometer, supernova, and BAO data fixes $f(\\mathcal{L}_m)$ up to a narrow parameter band, and two members of that band are late-time attractors. The paper reports that the $\\Lambda$CDM line lies inside the $1\\sigma$ reconstructed region, while the mean reconstructed curve is better described by the quadratic $f(\\mathcal{L}_m)=-2\\Lambda+\\alpha\\mathcal{L}_m+\\zeta\\mathcal{L}_m^2$ with $\\alpha\\approx -0.08582\\pm0.00345$ and $\\zeta\\approx(-1.512584\\pm0.563717)\\times10^{-6}$. Both the power-law model with $b_1\\in[0.018,0.025]$ and the square-root-exponential model with $b_2\\in[2.3,3.0]$ reach the critical point $(z,u)=(2,0)$, where $q=-1$ and $w_{\\rm eff}=-1$; the authors read this as a stable de Sitter attractor that explains late-time acceleration without a cosmological constant.","pith_inferences":["A natural next test is to redo the stability analysis for the exact $f(R,\\mathcal{L}_m)=R/2+f(\\mathcal{L}_m)$ equations without the intermediate variable $b=d\\ln f_R/d\\ln R$; since both reconstructed models have $f_R=1/2$, that intermediate variable vanishes, so the derivation needs a separate route to reach the same conclusion.","If the matter Lagrangian is instead identified with $-\\rho$ rather than $\\rho$, the reconstructed $f(\\mathcal{L}_m)$ and its stability points would likely shift; comparing the two conventions would test how much of the result is convention-dependent.","The same reconstruction could be applied to the first-order perturbation equations to predict growth-rate observables, connecting the background attractor result to large-scale structure.","As future Hubble data extend beyond the current redshift reach, the $1\\sigma$ band that defined $b_1$ and $b_2$ should narrow, making the paper's intervals the falsifiable output of the pipeline."],"forward_implications":["Both reconstructed models have a stable late-time attractor with $q=-1$ and $w_{\\rm eff}=-1$, so they reproduce cosmic acceleration without a cosmological constant.","The reconstructed ranges $b_1\\in[0.018,0.025]$ and $b_2\\in[2.3,3.0]$ become concrete observational predictions that future Hubble measurements at higher redshift will support or exclude.","The mean reconstructed curve favors a nonzero quadratic correction to $\\Lambda$CDM, with $\\alpha\\approx -0.08582$ and $\\zeta\\approx -1.51\\times10^{-6}$, giving a specific target for alternative models.","The Gaussian-process pipeline replaces arbitrary parametrizations of modified gravity with a data-driven selection of $f(\\mathcal{L}_m)$, which can be applied as new $H(z)$ data accumulate."],"supporting_citations":[{"why":"Defines the $f(R,\\mathcal{L}_m)$ action and supplies the field equations and Friedmann equations on which the reconstruction is built.","marker":"[17]"},{"why":"Provides the Gaussian-process regression method used to reconstruct $H(z)$ and $H'(z)$.","marker":"[23]"},{"why":"Supplies the power-law and square-root-exponential $f(\\mathcal{L}_m)$ forms that are tested within the reconstructed 1-sigma region.","marker":"[29]"},{"why":"Supplies the cosmic-chronometer Hubble data used in the reconstruction.","marker":"[30]"},{"why":"Supplies the supernova $E(z)$ data and correlation matrix used alongside the chronometer data.","marker":"[31]"},{"why":"Supplies the radial BAO Hubble data points included in the combined reconstruction.","marker":"[33]"},{"why":"Supplies the present matter-density parameter used in the initial condition for reconstructing $f(z)$.","marker":"[35]"},{"why":"Supplies the autonomous-system formalism used to derive the stability equations and classify critical points.","marker":"[36, 37]"}],"fun_headline_variants":["Stable late-time acceleration from reconstructed modified gravity","Two modified gravity models are stable late-time attractors","Data-driven gravity models stably accelerate without Lambda","Gravity model reconstruction passes stability test, no Lambda"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The stability analysis depends on treating the curvature part of the action as a variable function of $R$ so that the ratio $b=d\\ln f_R/d\\ln R$ is defined and nonzero, but in the two reconstructed models the geometric part is exactly $R/2$, making $f_R=1/2$ and $b=0$, so the key equation used to close the autonomous system is undefined for those models.","fun_headline_variants_meta":{"raw":{"variants":["Stable late-time acceleration from reconstructed modified gravity","Two modified gravity models are stable late-time attractors","Data-driven gravity models stably accelerate without Lambda","Gravity model reconstruction passes stability test, no Lambda"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001472,"raw_usage":{"total_tokens":6007,"prompt_tokens":1123,"completion_tokens":4884,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":739,"completion_tokens_details":{"reasoning_tokens":4823}},"tokens_in":739,"tokens_out":4884,"duration_ms":34719,"temperature":1.0,"reasoning_tokens":4823,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:47:11.983998+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete check is to derive the autonomous equations directly from the exact Friedmann equations for $f(R,\\mathcal{L}_m)=R/2+\\alpha\\mathcal{L}_m^{b_1}$ and for the square-root-exponential form, without using the $b=d\\ln f_R/d\\ln R$ step, and then compute the eigenvalues of the late-time critical point; if that point is not $(z,u)=(2,0)$ with all eigenvalues negative, the paper's stability claim fails.","supporting_citations":[{"cited_title":"Harko, F.S.N","cited_arxiv_id":null,"evidence_quote":"Defines the $f(R,\\mathcal{L}_m)$ action and supplies the field equations and Friedmann equations on which the reconstruction is built."},{"cited_title":"Seikel, C","cited_arxiv_id":null,"evidence_quote":"Provides the Gaussian-process regression method used to reconstruct $H(z)$ and $H'(z)$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the power-law and square-root-exponential $f(\\mathcal{L}_m)$ forms that are tested within the reconstructed 1-sigma region."},{"cited_title":"Moresco, L","cited_arxiv_id":null,"evidence_quote":"Supplies the cosmic-chronometer Hubble data used in the reconstruction."},{"cited_title":"Riess, S.A","cited_arxiv_id":null,"evidence_quote":"Supplies the supernova $E(z)$ data and correlation matrix used alongside the chronometer data."},{"cited_title":"Zhang and J.-Q","cited_arxiv_id":null,"evidence_quote":"Supplies the radial BAO Hubble data points included in the combined reconstruction."},{"cited_title":"Aghanim, Y","cited_arxiv_id":null,"evidence_quote":"Supplies the present matter-density parameter used in the initial condition for reconstructing $f(z)$."}],"review_version":1}