{"id":"a4a87389-1544-414e-bcb5-67619b6cb68a","arxiv_id":"2608.09297","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"A g-calculus pseudo-Tsallis distribution is introduced, its Ruppeiner curvature is claimed to depend only on alpha at leading order, and it fits WTI oil price deviations better than several benchmarks.","lead":"The paper builds deformed statistical distributions by conjugating exponentials and logarithms through a power function, generalizing Tsallis statistics. It claims the new model fits daily oil-price deviation data better than standard benchmarks and that its thermodynamic curvature depends only on the new deformation parameter at leading order.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equation (28) is unnormalized and does not reduce to Eq. (11) at q=1, so the PDF-level construction and the leading-curvature claim in Section VI are unsupported.","rationale":"The paper's intended contribution is a two-level pseudo-Tsallis family whose PDF-level member Eq. (28) is claimed to reduce at q=1 to Eq. (11) and to serve as input for the thermodynamic-geometry calculation. This is the load-bearing link between the algebraic construction and the central curvature result. The reader's objection is correct and I verified it directly: at q=1, Eq. (28) gives lambda^{1/alpha} e^{-lambda x^alpha/alpha}, whose integral is alpha^{(1-alpha)/alpha} Gamma(1/alpha), equal to 1 only for alpha=1; Eq. (11) has an additional Gamma(1+1/alpha) factor and exponent -lambda x^alpha, so the two densities are different even before normalization. The first-order density used in Section VI, Eq. (33), is the expansion of the unnormalized expression, not of Eq. (11): its alpha-term x(1-ln x) matches the alpha-derivative of the q->1 branch, whereas Eq. (19) from Eq. (11) contains 1-gamma_E - x ln x. Thus the PDF-level curvature calculation and the claim that alpha alone controls leading-order curvature are not supported. I do not see a similar defect in the CDF-level construction (Eqs. (29), (30)) or in the empirical likelihood (Eq. (32)), which use the CDF level; those parts could survive repair. But because the abstract and Section VI assert the both-level geometric result, the manuscript needs a corrected derivation before the central claim can be accepted. No machine-checked proof or reproducible code is present to offset this. I therefore agree with the reader's rejection and recommend no change to the verdict.","tokens_in":17128,"tokens_out":10125,"duration_ms":85672,"concrete_test":"Normalize Eq. (28) by its exact integral I(lambda,q,alpha) (numerically verify I(1,1,2)=sqrt(pi/2)), define f_norm=f/I, and re-derive the first-order expansion in (alpha-1) and (q-1) from f_norm. Recompute N_alpha, U_alpha, R_alpha, and R_q in Section VI, and independently check whether the q=1 branch of f_norm equals Eq. (11). If Appendix B changes or R_q becomes nonzero, the central geometric claim fails; if it does not, the normalization defect is not load-bearing for the curvature result.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation (28), the PDF-level pseudo-Tsallis density, is not a normalized probability density and does not reduce to the paper's own Eq. (11) at q=1. For the q=1 branch, f(x)=(lambda e^{-lambda x^alpha})^{1/alpha}; its integral over x>0 is alpha^{(1-alpha)/alpha} Gamma(1/alpha), which equals 1 only for alpha=1. For example, alpha=2 and lambda=1 gives integral sqrt(pi/2), not 1. Eq. (11) contains the normalizer Gamma(1+1/alpha) and the exponent -lambda x^alpha rather than -lambda x^alpha/alpha, so the two densities disagree even as functions. Section VI expands this unnormalized object: Eq. (33) is the Taylor expansion of the q->1 branch of Eq. (28), not of Eq. (11), and its alpha-term x(1-ln x) differs from the correct expansion of Eq. (11), Eq. (19), which contains 1-gamma_E - x ln x. Consequently, N_alpha, U_alpha, R_alpha, and the conclusion R_q=0 in Section VI inherit the invalid density. The CDF-level model (Eqs. (29)-(30)) is normalized by construction and is what the empirical AIC/BIC analysis uses, so those parts may be salvageable; however, the paper's stated two-level construction and the PDF-level curvature claim are unsupported as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a pseudo-analytic formalism based on g-calculus, in which standard algebraic operations are conjugated through a monotone generator g(x)=x^alpha, and uses it to define pseudo-exponential and pseudo-Tsallis distributions at both the probability-density and cumulative-distribution levels. The authors then compute Ruppeiner thermodynamic curvature for these families by perturbing around the classical limit (alpha=1, q=1), and claim that to leading order the scalar curvature is controlled entirely by the generator deformation parameter alpha, with the nonextensivity parameter q entering only at higher order. The paper closes with an empirical application to absolute deviations of daily WTI crude-oil prices from their 100-day moving average, using AIC and BIC to argue that the proposed pseudo-Tsallis distribution fits better than standard benchmarks.","tokens_in":17490,"tokens_out":4668,"duration_ms":44953,"significance":"If the construction and curvature calculation were valid, the paper would offer a unified g-calculus route to deformed statistics and an interesting geometric hierarchy in which alpha, rather than q, encodes the leading effective interaction. The CDF-level family in Eqs. (29)-(30) is normalized by construction and the empirical comparison is transparent, so parts of the framework may be salvageable. However, the PDF-level pseudo-Tsallis density used for the central geometric claim is not a valid normalized probability density, and the perturbative expansion in Section VI is not obtained from the normalized pseudo-exponential density of Eq. (11). Because the main geometric statement rests on this invalid object, the paper's central contribution is not supported as written.","major_comments":[{"comment":"The q=1 branch of the pseudo-Tsallis PDF in Eq. (28), f(x)=(lambda e^{-lambda x^alpha})^{1/alpha}, is not normalized for alpha != 1: its integral over x>0 is alpha^{(1-alpha)/alpha} Gamma(1/alpha), which equals 1 only for alpha=1. For example, alpha=2 and lambda=1 give integral sqrt(pi/2), not 1. Moreover, this branch does not reduce to the pseudo-exponential density in Eq. (11), which contains the normalizer Gamma(1+1/alpha) and the exponent -lambda x^alpha; the two functions disagree even up to normalization. Consequently, the PDF-level pseudo-Tsallis construction in Section V.A, and every PDF-level thermodynamic quantity built on it in Section VI, are not supported.","section":"Section V.A, Eq. (28)"},{"comment":"Equation (33) is presented as the first-order expansion of the pseudo-Tsallis PDF around alpha=1 and q=1, but it is actually the expansion of the unnormalized q=1 branch of Eq. (28), not of the normalized density in Eq. (11). The alpha-dependent term x(1-ln x) differs from the correct first-order expansion of Eq. (11) given in Eq. (19), which contains 1-gamma_E - x ln x. The expressions for N_alpha, U_alpha, R_alpha in Appendix B and the assertion R_q=0 in Eq. (37) are therefore derived from an invalid density and do not establish the paper's central claim that the scalar curvature is governed solely by alpha at leading order.","section":"Section VI, Eq. (33)"},{"comment":"The identities displayed in Eq. (27) are incorrect: the standard relations are exp_q(ln_q(x)) = x and ln_q(exp_q(x)) = x, not 1. As written, Eq. (27) contradicts the correct reciprocal relations given in Eqs. (8)-(9) and weakens the claimed consistency of the deformed logarithm-exponential pair on which the pseudo-Tsallis construction relies.","section":"Section V, Eq. (27)"},{"comment":"The empirical support for the 'superior description' claim is an in-sample maximum-likelihood comparison using AIC and BIC. The PTE model has three free parameters (lambda, q, alpha) while several benchmarks have two, and no out-of-sample or predictive validation is provided. The abstract's wording 'superior description' is therefore stronger than the evidence in Table I supports; the table demonstrates only that the three-parameter family achieves a better in-sample information-criterion score on this particular dataset.","section":"Section VIII, Table I"}],"minor_comments":[{"comment":"The pseudo-exponential application is referenced only to [45] and no results are reported in the present paper; the data, fitted parameters, and comparison table should be included if the claim is to be assessed.","section":"Section VII"},{"comment":"The table has several labelling inconsistencies: the 'q-Tsallis exponential' row has no alpha entry, the 'T QG' row is described with parameters (mu, beta, q) while the column headers show alpha, lambda, q, and the 'N(alpha, lambda^2)' row does not explain whether the fitted entries 4.1527 and 5.7462 are location and scale parameters.","section":"Table I"},{"comment":"The statement that the two expansions 'exhibit a high degree of similarity' and that the choice between PDF- and CDF-level constructions is 'inconsequential' is not quantified; a direct numerical comparison of the two R(z) curves would make the claim precise.","section":"Section IV, Fig. 1"},{"comment":"Reference [16] is malformed ('D VG, en.'), and the citation of the authors' own related work [30] as an established result should be clarified with its publication status and a fuller description of its content.","section":"References"}],"recommendation":"reject","confidential_remarks":"The rejection is driven by the invalid PDF-level normalization in Eq. (28) and the resulting unsupported curvature calculation in Section VI; these are load-bearing for the abstract's main geometric claim. I would encourage the authors to repair the PDF-level definition by including the proper normalizer, redo the perturbative expansion, and then resubmit. The CDF-level model and the empirical comparison are not affected by this particular error and could form the basis of a revised manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe stress-test note is right, and it lands on the main claim. Equation (28), the PDF-level pseudo-Tsallis density, is not normalized for alpha != 1, and its q=1 branch does not equal Equation (11). The integral example (alpha=2, lambda=1 giving sqrt(pi/2)) is incontrovertible. Since Sections V.A and VI build the thermodynamic geometry on that density, the headline result—alpha alone controls the leading-order curvature, q is silent—is unsupported as stated. Also, Eq. (27) states exp_q(ln_q(x)) = 1, which is simply wrong; it should be x. That kind of formal slip in the setup makes it hard to trust the rest of the layer without a careful pass.\n\nThat said, there is real content here. The CDF-level pseudo-Tsallis family (Eqs. 29-30) is normalized by construction and appears to be new. The perturbative expansions in the appendices are substantial, and the qualitative picture—alpha as a sign-controlling interaction parameter—is physically interesting if it survives re-derivation. The empirical application to WTI price deviations is an honest in-sample AIC/BIC comparison; it is model selection, not prediction, but it does show the CDF-level model is competitive with standard benchmarks.\n\nThe soft spots are all in the PDF-level construction and the way the paper generalizes from it. The \"both levels\" claim in the abstract requires the two constructions to be parallel, and they are not as written. The curvature decoupling might still hold for the CDF-level model, but the paper doesn't establish it because the derivation is contaminated by the unnormalized density. The empirical section uses the CDF-level model, so it can be salvaged independently.\n\nWho is this for? People working on deformed statistics, thermodynamic geometry, or econophysics who want a concrete two-parameter heavy-tailed family. They should read the CDF-level parts and ignore the PDF-level curvature until the authors fix the normalization and re-run the expansion.\n\nRecommendation: this deserves a serious referee, not a desk reject, because the CDF-level model and the empirical comparison are substantive and the flaw is fixable. But the referee should be told to check normalization first. If the authors cannot fix Eq. (28) or explain why the curvature result survives the correct normalization, the paper should not be accepted.","headline":"The central geometric claim rests on a density that is not normalized and does not reduce to the paper's own q=1 limit; the CDF-level model and empirical AIC comparison may survive a rewrite, but the abstract's two-level claim does not hold as written.","tokens_in":17975,"tokens_out":3345,"would_cite":false,"duration_ms":31035,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B03","82B30"],"pacs":["05.70.-a","05.20.-y","89.65.Gh"],"model":"deepseek-v4-flash","headline":"This paper argues that transporting exponentials through the generator $g(x)=x^\\alpha$ produces a two-parameter family of deformed distributions (pseudo-exponential and pseudo-Tsallis) in which, to first order around the classical limit…","keywords":["g-calculus","pseudo-analysis","deformed statistics","Tsallis statistics","thermodynamic geometry","Ruppeiner metric","pseudo-exponential distribution","WTI crude oil fluctuations"],"falsifier":"Evaluate the normalization integral of Eq. (28) numerically for a few parameter values, e.g. $\\alpha=2$, $q=1$, $\\lambda=1$; since the result is $\\sqrt{\\pi/2}$ rather than $1$, a normalized-density assumption fails, which would require revisiting the likelihood and curvature expressions built on that density.","tokens_in":16951,"feed_emoji":"📈","tokens_out":9814,"duration_ms":87128,"temperature":0.7,"pith_summary":"The paper develops a pseudo-analytic framework in which ordinary statistical distributions are rebuilt by conjugating functions through a monotone generator, here $g(x)=x^\\alpha$. It constructs deformed exponentials at both the probability-density and cumulative-distribution levels, recovering ordinary exponentials at $\\alpha=1$ and Tsallis-type $q$-exponentials as a further layer. Its central geometric claim is that near the classical limit the thermodynamic scalar curvature is determined by $\\alpha$ alone, with $q$ entering only at higher order, so the generator deformation acts as the leading effective interaction. On the empirical side, the pseudo-Tsallis model is claimed to fit absolute deviations of WTI crude-oil prices from their 100-day moving average better than exponential, Weibull, gamma-exponential, normal, Laplace, and Tsallis-$q$-Gaussian benchmarks by AIC/BIC. A sympathetic reader would care because this offers a parameter-hierarchy route from deformed statistics to thermodynamics and a concrete econophysics application.","feed_headline":"One deformation parameter alone sets the curvature at leading order","feed_subtitle":"New pseudo-Tsallis family: alpha alone sets leading curvature, and it beats standard benchmarks on WTI prices.","key_machinery":"The engine is $g$-calculus: for a monotone generator $g$, every function $f$ is re-expressed as $f_g(x)=g^{-1}(f(g(x)))$, so the ordinary exponential becomes $\\exp_g(x)=g^{-1}(e^{g(x)})$ and the $q$-exponential becomes $\\exp_{g,q}(x)=g^{-1}(\\exp_q(g(x)))$. With $g(x)=x^\\alpha$, this yields pseudo-exponential and pseudo-Tsallis densities at the PDF level (Eq. 11 and Eq. 28) and at the CDF level (Eq. 13 and Eq. 30). The thermodynamic analysis then uses the Ruppeiner metric, built from the Hessian of $\\ln Z$ in the $(\\beta,\\gamma)$ plane, and expands the densities to first order in $(\\alpha-1)$ and $(q-1)$; the curvature formula evaluates $R$ from the metric components and their derivatives. The load-bearing step is that the $q$-corrections to $R$ vanish at this order, leaving only $\\alpha$.","core_discovery":"The paper's discovery, on its own terms, is a hierarchy in deformation effects: expanding the pseudo-Tsallis density around $(\\alpha,q)=(1,1)$ gives first-order corrections in both parameters, but the thermodynamic scalar curvature $R$ receives a first-order contribution only from $(\\alpha-1)$; the $q$-correction $R_q$ vanishes. Thus, to leading order, the sign and magnitude of the effective statistical interaction (positive $R$ for $\\alpha<1$, negative for $\\alpha>1$, zero at $\\alpha=1$) are controlled by the generator exponent $\\alpha$, while nonextensivity $q$ only appears at higher orders. This decoupling is argued to mirror earlier findings that nonextensive Maxwell–Boltzmann statistics alone leaves curvature zero. The same paper also claims that the pseudo-Tsallis family, fitted by maximum likelihood, outperforms standard competing distributions on the WTI data set, providing empirical support for the construction.","pith_inferences":["An extension the paper does not make is that its first-order equivalence between PDF- and CDF-level constructions suggests a broader invariance: any two deformation routes that agree at first order will look identical in curvature, so distinguishing them requires second-order terms or normalization constraints.","The normalization failure for Eq. (28) at $\\alpha\\neq 1$ means the empirical likelihood and the thermodynamic curvature inherit an un-normalized density; correcting the normalization would change the fitted parameters and the curvature coefficients, so the quantitative AIC/BIC comparison should be rechecked.","One testable extension is to apply the same $g$-calculus construction with other generators (e.g., $g(x)=\\ln(1+x)$ or $g(x)=x^\\beta$ with two exponents) to see whether the decoupling of the second parameter is special to $x^\\alpha$ or generic to conjugacy deformations.","Another extension is to compute the second-order curvature $R_{qq}$ and $R_{\\alpha q}$; the paper's hierarchy is only a leading-order statement, and higher-order couplings would determine whether $\\alpha$ remains the dominant interaction channel."],"forward_implications":["For the pseudo-exponential model, $\\alpha<1$ gives positive scalar curvature (effective attraction), $\\alpha>1$ negative curvature (effective repulsion), and $\\alpha=1$ recovers the zero-curvature Maxwell–Boltzmann gas.","The PDF-level and CDF-level pseudo-exponential constructions yield nearly identical first-order thermodynamics, so the two routes to generalization are observationally equivalent at leading order.","In the pseudo-Tsallis model, $q$ does not affect curvature at first order, meaning purely nonextensive deformation alone does not generate leading-order thermodynamic interactions in this construction.","The pseudo-Tsallis fit to WTI absolute deviations has the lowest AIC and BIC among the distributions compared, so the family is claimed to be a practically useful heavy-tailed model for high-frequency financial fluctuations.","If the hierarchy persists at higher orders, the nonextensivity parameter $q$ would act as a perturbation on a background interaction fixed by $\\alpha$."],"supporting_citations":[{"why":"Supplies the Tsallis q-logarithm and q-exponential definitions and the nonextensive statistical framework that the pseudo-Tsallis construction layers onto.","marker":"[1]"},{"why":"Introduces g-calculus, the pseudo-operation framework through monotone generators on which the entire construction rests.","marker":"[33]"},{"why":"Provides the previous pseudo-exponential distribution whose normalization and PDF-level form the paper adopts and compares against.","marker":"[45]"},{"why":"Establishes the Ruppeiner thermodynamic-geometry framework used to define the scalar curvature as a measure of statistical interactions.","marker":"[46]"},{"why":"Gives the metric components as the Hessian of the log partition function, the starting point for computing the curvature.","marker":"[48]"},{"why":"Provides the canonical representation and the $(\\beta,\\gamma)$ conventions used to write the metric components in the paper.","marker":"[49]"},{"why":"Shows that nonextensive Maxwell–Boltzmann statistics leaves thermodynamic curvature zero over a range of q, lending consistency to the paper's claim that q decouples at leading order.","marker":"[60]"}],"fun_headline_variants":["Curvature's leading term: alpha only, q stays silent","Thermodynamic curvature decouples q: alpha alone leads","One parameter rules curvature: alpha beats benchmarks","Pseudo-distributions: alpha drives curvature, tops WTI fits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the pseudo-Tsallis density in Eq. (28) is a normalized probability density for every $\\alpha>0$; a direct check for $(\\alpha,q,\\lambda)=(2,1,1)$ gives $\\int_0^\\infty f(x)\\,dx=\\sqrt{\\pi/2}\\neq 1$, and the $q=1$ branch of Eq. (28) does not reproduce the normalized density in Eq. (11).","fun_headline_variants_meta":{"raw":{"variants":["Curvature's leading term: alpha only, q stays silent","Thermodynamic curvature decouples q: alpha alone leads","One parameter rules curvature: alpha beats benchmarks","Pseudo-distributions: alpha drives curvature, tops WTI fits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000454,"raw_usage":{"total_tokens":2286,"prompt_tokens":956,"completion_tokens":1330,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":572,"completion_tokens_details":{"reasoning_tokens":1260}},"tokens_in":572,"tokens_out":1330,"duration_ms":10075,"temperature":1.0,"reasoning_tokens":1260,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:56:39.273906+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the normalization integral of Eq. (28) numerically for a few parameter values, e.g. $\\alpha=2$, $q=1$, $\\lambda=1$; since the result is $\\sqrt{\\pi/2}$ rather than $1$, a normalized-density assumption fails, which would require revisiting the likelihood and curvature expressions built on that density.","supporting_citations":[{"cited_title":"Pap, International Journal of Approximate Reasoning 47, 368 (2008)","cited_arxiv_id":null,"evidence_quote":"Introduces g-calculus, the pseudo-operation framework through monotone generators on which the entire construction rests."},{"cited_title":"Mehri-Dehnavi, H","cited_arxiv_id":null,"evidence_quote":"Provides the previous pseudo-exponential distribution whose normalization and PDF-level form the paper adopts and compares against."},{"cited_title":"Janyszek and R","cited_arxiv_id":null,"evidence_quote":"Gives the metric components as the Hessian of the log partition function, the starting point for computing the curvature."},{"cited_title":"Mirza and H","cited_arxiv_id":null,"evidence_quote":"Provides the canonical representation and the $(\\beta,\\gamma)$ conventions used to write the metric components in the paper."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that nonextensive Maxwell–Boltzmann statistics leaves thermodynamic curvature zero over a range of q, lending consistency to the paper's claim that q decouples at leading order."}],"review_version":1}