{"id":"2da5f490-8944-4627-aa8e-81dedf35df4d","arxiv_id":"2505.21015","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"New classes of upper moments and down-Fisher measures satisfy a battery of sharp informational inequalities whose optimal constants and minimizers are transported from the classical Stam, Cramér-Rao, and moment-entropy inequalities.","lead":"The paper defines two new families of informational measures, upper moments and down-Fisher measures, built by applying classical moment and Fisher-information functionals to transformed probability densities, and proves a series of sharp inequalities relating them to moments, entropies, and Fisher information.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"All new inequalities inherit their validity from the unproved Lemma 2.1 and the companion-paper Stam inequality [38, Thm 5.1]; a hidden restriction or sign/regularity failure in those imported identities would invalidate the claimed constants and minimizers.","rationale":"The paper's central mechanism is a substitution-and-rename argument: known inequalities are applied to up- or down-transformed densities, and Lemma 2.1 converts the resulting quantities into the new functionals. The reader's weakest assumption identifies exactly the point I consider most load-bearing: Lemma 2.1 and the tri-parametric Stam inequality of the companion preprint are imported without proof or complete hypotheses. I examined the algebraic route of Theorem 4.1 and found it internally coherent once Lemma 2.1 is granted; the apparent exponent discrepancy in Theorem 3.1 is more plausibly an omitted monotone-power step than a fatal error. The genuinely fragile part is that the identities are stated for very general densities and parameter ranges, while the paper itself acknowledges open regularity questions and a concrete failure of iterated down-transformations. A single independent verification of the identities on a heavy-tailed density with negative p would resolve whether this is an exposition gap or a substantive defect. Since I found no concrete false step that would require rejection, the conditional verdict is appropriate, and my read does not change it.","tokens_in":28375,"tokens_out":43732,"duration_ms":430547,"concrete_test":"Independently verify Lemma 2.1 directly from Definitions 2.1 and 2.2 for a heavy-tailed density in the advertised extended regime: take f(x)=x^{-2} on (1, infinity), alpha=1, p=-1/4. Direct computation gives D_1[f](s)=1/(2 s^{1/2}) on (0,1), so sigma_{-1/4}[D_1[f]] = (1/2 * integral_0^1 s^{-3/4} ds)^{-4} = 1/16; the right-hand side N_{3/4}[f]^{-1} = (integral_1^infinity x^{-3/2} dx)^{-4} = 1/16. If the printed identity does not reproduce this equality, the central inequalities fail. Then repeat the check for alpha=3/2 and p=-1/3, and separately verify Eq. (4.12) and Eq. (4.22) with the [38] Stam constant on this same family; any mismatch shows the imported identities do not cover the extended parameter range.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4 is a substitution argument. Eq. (4.12) rewrites N_lambda[U_alpha[f]] through Lemma 2.1; Theorems 4.3 and 4.4 use both identities of Lemma 2.1 together with the classical Cramer-Rao and with the tri-parametric Stam inequality imported from [38, Thm 5.1]; Theorem 4.6 applies the same machinery to a double down-transformation. None of these load-bearing identities is proved in the present manuscript: Lemma 2.1 is cited to the companion preprint, and its exact hypotheses are not stated. The extended regime advertised by the paper--negative p, sub-1 exponents, densities with divergent edges, and iterated down transformations--is precisely where those identities are least secure. The paper itself flags the problem: after Theorem 4.3, the regularity classes under which U_alpha0[f] is absolutely continuous or of bounded variation are left as an open problem; after Theorem 4.2, a counterexample shows D_beta[D_alpha[f]] need not exist; and sharpness of the iterated inequalities is only conjectured. If Lemma 2.1 or [38, Thm 5.1] fails on any of these ranges, Theorems 4.1, 4.3, 4.4, and 4.6 inherit the failure, and the claimed optimal constants and minimizers are not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces two new families of informational functionals: (p,α)-upper moments M_{p,α}[f] and their deviations m_{p,α}[f], including higher-order versions, and (p,q,λ)-down-Fisher measures φ_{p,q,λ}[f]. These are obtained by applying classical moments, entropy powers, and Fisher-type information measures to the up/down transformed densities of the companion preprint [38]. The central contribution is a set of informational inequalities: Theorem 4.1 (upper-moment–moment), Theorem 4.2 (iterated upper-moment inequalities), Theorem 4.3 (upper-moment–entropy), Theorem 4.4 (down-Fisher–Fisher), Theorem 4.6 (modified Stam-like inequalities), plus corollaries bounding classical moment–entropy, Stam, and Cramér–Rao products. The proofs are substitution arguments: known moment-entropy, Stam, and Cramér–Rao inequalities are applied to up/down transformed densities, and the transformation identities of Lemma 2.1 are used to rewrite the resulting terms. Applications to the Hausdorff moment problem and MaxEnt-type reconstruction are also discussed.","tokens_in":28514,"tokens_out":6614,"duration_ms":67599,"significance":"If the results are correct in the stated range, the paper gives sharp quantitative inequalities with explicit constants and minimizers for a broad family of densities, including heavy-tailed and boundary-divergent ones, and it connects these to generalized Beta minimizers. The substitution method is transparent and parameter-free: no constants are fitted and no new inequality is assumed, so the derivations are reproducible from the cited classical results and identities. The appendix providing explicit minimizer formulas is a useful service. The main caveats are that the load-bearing transformation identities are imported from a companion preprint without proof, the regularity hypotheses for several theorems are either too weak or left as open problems, and some advertised optimality is explicitly conjectural.","major_comments":[{"comment":"The identities (2.28)–(2.29), together with the α=2 variants, are used in every proof of Section 4 (for example in Eq. (4.12), Theorem 4.3, and Theorem 4.4), but they are stated without proof and without explicit hypotheses beyond 'let f be a probability density'. Since the paper advertises validity for negative p, sub-1 exponents, and densities that diverge at the edge of a compact support, the exact regularity and parameter ranges under which these identities hold must be stated; otherwise the new inequalities inherit every failure of the imported identities. The proof in [38, Section 3.1] should either be reproduced in an appendix or the companion preprint should be made available with the precise hypotheses verified for the extended ranges.","section":"Section 2.2, Lemma 2.1"},{"comment":"The theorem is stated for 'any probability density function', but in the mirrored case the proof applies the mirrored moment-entropy inequality (2.12), which in Section 2 is stated only for continuously differentiable densities. The up-transformed density f↑_α need not be continuously differentiable for an arbitrary f, so the theorem as stated lacks the regularity hypothesis needed for the application of (2.12). This affects the validity of the result in the mirrored range (4.5), one of the paper's advertised extensions.","section":"Theorem 4.1, mirrored range (4.5)"},{"comment":"The abstract and introduction announce 'optimal constants and minimizers' for the new inequalities, but the paragraph after Theorem 4.2 explicitly states that the iterated minimizer D_{α1}[D_{α0}[g_{p,λ}]] is well defined only under the restriction α0 > 2 - λ/p* - 1/p, and that sharpness in the remaining cases is only conjectured. Thus Theorem 4.2 and the iterated part of Theorem 4.3 (for α1 < 2) are not fully proved as sharp inequalities. The statements and abstract should either be restricted to the proved cases or clearly mark the optimality as conjectural.","section":"Section 4.1, after Theorem 4.2"},{"comment":"The statement says 'for any probability density function f such that D_{2-λ}[f] is absolutely continuous', but the down transformation is only defined for decreasing densities (Definition 2.1), and for p<1 the tri-parametric Stam inequality (2.16) with condition (2.17) requires f to be continuously differentiable with f' < 0. The theorem omits the monotonicity and differentiability conditions on f, so condition (4.21) alone does not guarantee that D_{2-λ}[f] is well defined. The statement should include the hypotheses that are actually needed for the imported Stam inequality.","section":"Theorem 4.4"},{"comment":"The open problem stated immediately after Theorem 4.3 acknowledges that the class of densities f for which U_{α0}[f] is absolutely continuous (or of bounded variation) is not characterized. Since Theorem 4.3 is presented as applying 'for any probability density f' under a condition that is left uncharacterized, the practical scope of the theorem is unclear; at minimum, the statement should separate the algebraic validity of the inequality from the currently open question of which densities satisfy the required regularity of U_{α0}[f].","section":"Theorem 4.3, open problem"}],"minor_comments":[{"comment":"There is a typographical error in the abstract and repeated in the introduction: 'some of the the most important informational inequalities' should read 'some of the most important informational inequalities'.","section":"Abstract and Introduction"},{"comment":"The notation \\widetilde{g}_{p,λ} is defined twice with the same two cases (λ>0 and λ<0); this is harmless but could be simplified to a single definition.","section":"Eq. (2.13)"},{"comment":"The phrase 'we raise the previous inequality to the power α1 − 2' is potentially confusing because when α1 − 2 < 0 the inequality direction changes; the subsequent sentence does address this, but rephrasing would improve clarity.","section":"Section 4.1, Theorem 4.2 proof"},{"comment":"There is a stray punctuation artifact in the sentence defining p after Eq. (4.45): '2 + \\tilde{p} − β, .' should be cleaned up.","section":"Theorem 4.6 proof"},{"comment":"The displayed formulas for g⋆_{p,α,q} in cases (a)–(d) mix the parameters q and λ; aligning the notation explicitly with Eq. (4.2) would help readers check the minimizer expressions.","section":"Appendix"}],"recommendation":"major_revision","confidential_remarks":"The paper is heavily dependent on the companion preprint [38] for both Lemma 2.1 and the tri-parametric Stam inequality. The editor may wish to verify the status of [38] and consider asking for a self-contained statement or proof of the key identities. The abstract's blanket claim of optimal constants and minimizers also exceeds what the paper proves, since several sharpness statements are marked as conjectures. These issues are fixable within the manuscript's scope, hence my recommendation of major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper defines upper moments M_{p,alpha} as p-moments of up-transformed densities and down-Fisher measures as Fisher information of down-transformed densities, then proves a family of inequalities (upper-moment-moment, upper-moment-entropy, down-Fisher-Fisher, modified Stam) by substituting the transformation identities of Lemma 2.1 into the classical moment-entropy, Stam, and Cramér-Rao inequalities. The functionals and the inequality statements are new; the constants and minimizers are explicit and computed from known stretched Gaussians. Spot checks of Theorems 4.1 and 4.3 confirm the substitution-and-rename pattern is mechanically valid in the classical parameter range. The paper is honest about some gaps: after Theorem 4.3 it flags the regularity classes needed for absolute continuity of U_alpha0[f] as an open problem, and after Theorem 4.2 it gives a counterexample where a second down transformation fails and concedes that sharpness of iterated inequalities is only conjectured. That is real epistemic modesty.\n\nThe soft spots are in proportion. First, the abstract and introduction claim optimal constants and minimizers without qualification, while the body only establishes sharpness for first-order inequalities; the iterated ones are conjectural. Second, Theorem 3.1's statement and proof disagree on exponents of the down-Fisher measures; likely a typo but needs fixing. Third, the derivation chain leans heavily on the companion preprint [38] for Lemma 2.1 and the tri-parametric Stam inequality, and the hypotheses under which those hold on the extended parameter ranges (negative p, sub-1 exponents, boundary-divergent densities) are not fully stated here. The paper acknowledges some of this, but a referee cannot fully verify the central claims without the companion manuscript in hand.\n\nNone of this condemns the mechanism. The new functionals are natural, the inequalities are parameter-free, and the method is transparent. The paper is for researchers working on informational inequalities, generalized Fisher information, and moment problems; it is a solid contribution if the companion results hold. It deserves a serious referee, and the editor should ensure the referee has access to [38]. The open problems should be stated as such rather than buried under the abstract's unqualified claim of optimal constants and minimizers.\n\nRecommendation: send to peer review. It is legitimate conditional work, not a desk reject.","headline":"New functionals and inequality statements built by substituting known inequalities through up/down transformations; sound in the classical range, but sharpness and regularity claims outrun what is proved, and the load-bearing identities sit in an unreviewed companion preprint.","tokens_in":29283,"tokens_out":1818,"would_cite":true,"duration_ms":18595,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A17","60E15","26D15","62B10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper introduces upper moments and down-Fisher measures—informational functionals built from the up/down transforms—and claims they satisfy sharp inequalities with explicit constants and minimizers, extending moment-entropy, Stam, and…","keywords":["upper moments","down-Fisher measures","up/down transformations","informational inequalities","Rényi entropy","Fisher information","moment-entropy inequality","Hausdorff moment problem"],"falsifier":"Take a decreasing heavy-tailed density such as $f(x)=C(1+|x|)^{-\\eta}$ with $1<\\eta<2$, choose parameters in the mirrored range (4.5), and compute both sides of inequality (4.3) directly; if $(m_{p^*,\\alpha}[f]/\\sigma_q[f])^{\\Theta_1(p,\\alpha,q)}$ ever falls below the claimed constant $\\kappa^{(-1)}_{p,\\alpha,q}$, then the transformation identity fails on that range and the theorem's claim collapses.","tokens_in":1971,"feed_emoji":"📐","tokens_out":5872,"duration_ms":103851,"temperature":0.7,"pith_summary":"The paper introduces two new families of informational functionals, upper moments and down-Fisher measures, formed by applying classical functionals such as p-moments and Fisher information to densities transformed by the 'up' and 'down' operators introduced in the authors' companion work. It claims that, for these new functionals, the classical moment-entropy, Stam, and Cramér-Rao inequalities admit sharp extensions with explicit optimal constants and minimizers, including in parameter ranges where the underlying density has heavy tails or diverges at the edge of its support. The upshot would be quantitative upper bounds on moment-entropy, Stam, and Cramér-Rao products for a substantially wider class of densities than the stretched Gaussians that minimize the classical versions. The paper also shows that generalized Beta densities play the extremal role for upper moments with fixed moment, and that maximizing upper moments of higher order under fixed lower-order upper-moments extends the MaxEnt approach to the Hausdorff moment problem.","feed_headline":"Sharp moment-entropy and Stam bounds reach heavy-tailed densities","feed_subtitle":"Upper moments and down-Fisher measures extend classical inequalities beyond the stretched-Gaussian setting.","key_machinery":"The machinery is the mutually inverse pair of transformations, up $U_\\alpha$ and down $D_\\alpha$, taken from the authors' preceding work. The down transform $D_\\alpha[f](s) = f^\\alpha(x(s))|f'(x(s))|^{-1}$ rewrites a decreasing density in terms of its own derivative, while the up transform is its inverse; together they map heavy-tailed densities to better-behaved ones and vice versa. Lemma 2.1 supplies the load-bearing identities expressing moments, Rényi entropy power, and Fisher information of transformed densities in terms of the original density's functionals. Each new theorem is obtained by substituting these identities into the classical moment-entropy, Stam, and Cramér-Rao inequalities, or into the companion tri-parametric Stam inequality, and then identifying the optimal constant with the classical constant and the minimizer with the transformed classical minimizer.","core_discovery":"The central claim is that the up and down transformations act as a dictionary: every classical informational inequality applied to a transformed density becomes a new sharp inequality for the original density, with constants inherited from the classical result and minimizers obtained by pulling the classical minimizer back through the transformation. Concretely, Theorem 4.1 states that for $p$, $q$, $\\alpha$ in the classical range (4.4) or the mirrored range (4.5), $(m_{p^*,\\alpha}[f]/\\sigma_q[f])^{\\Theta_1(p,\\alpha,q)} \\ge \\kappa^{(-1)}_{p,\\alpha,q}$, with optimal constant and minimizer $D_\\alpha[g_{p,\\lambda}]$, the down-transform of a stretched Gaussian; Theorems 4.3, 4.4, and 4.6 give analogous sharp upper-moment–entropy, down-Fisher–Fisher, and modified Stam inequalities. The paper argues that these inequalities reveal the same structural relationship between upper moments and moments as exists between moments and Rényi entropy power, and between entropy power and generalized Fisher information, with the down-Fisher measure completing the chain.","pith_inferences":["Because each proof is a substitution of the transformation identities into a classical inequality, the same recipe would produce additional sharp inequalities for any future inequality whose minimizer is a stretched Gaussian; this generalization is not stated in the paper.","The authors' conjecture that only decreasing exponential and power densities survive infinitely many down transforms could be probed numerically by seeking other decreasing densities whose down transform remains decreasing and integrable, which would enlarge the class where the higher-order upper-moment inequalities (4.15) are sharp.","A natural extension to higher dimensions or manifolds is left implicit; if the up/down dictionary admits a multidimensional analogue, the same construction would define new complexity measures suitable for quantum position-momentum uncertainty products.","Testing the sharpness of the constants on a family of generalized Beta densities with varying edge exponents would offer a concrete numerical check of the optimality claims without needing to verify the full regularity conditions."],"forward_implications":["The moment-entropy, Stam, and Cramér-Rao products become bounded above by explicit functions for densities with heavy tails or divergences at the edge of their support, not only for the classical stretched-Gaussian minimizers.","The upper-moment–moment inequality (4.3) is saturated by the down-transformed stretched Gaussian $D_\\alpha[g_{p,\\lambda}]$, a generalized Beta density, placing generalized Beta laws in the role that stretched Gaussians play for ordinary moments.","When $\\Theta_1(p,\\alpha,q^*)>0$, inequality (4.3) yields sharp bounds of the form $\\sigma_{q^*}[f] \\le A\\, m_{p^*,\\alpha}[f]$, which in turn give upper bounds for classical moment-entropy and Cramér-Rao products through (4.13) and (4.14).","The down-Fisher–Fisher inequality (4.22) and its consequences (4.30)–(4.32) produce upper bounds for Cramér-Rao and Stam products in terms of the down-Fisher measure, including cases with negative parameters in the mirrored domain.","Maximizing Rényi entropy under fixed moments is equivalent, via the down transformation, to maximizing a certain upper-moment (or logarithmic-moment) functional under fixed upper-moments, extending the MaxEnt principle to a wider class of reconstruction problems."],"supporting_citations":[{"why":"Companion preprint that supplies the up/down transformations, the Lemma 2.1 identities (Eqs. (2.28)-(2.29)), and the tri-parametric Stam inequality on which the new proofs are built.","marker":"[38]"},{"why":"Provides the classical Cramér-Rao and moment-entropy inequalities for Rényi entropy and generalized Fisher information, with stretched-Gaussian minimizers used as the starting point.","marker":"[27]"},{"why":"Establishes the moment-entropy inequalities whose transformed versions underlie Theorems 4.1 and 4.6.","marker":"[45]"},{"why":"Defines the generalized Fisher information and the q-Gaussian minimizers used in the Stam-type inequalities.","marker":"[44]"},{"why":"Supplies the generalized bi-parametric Stam inequality used in the down-Fisher–Fisher proof.","marker":"[36]"},{"why":"Provides generalized Cramér-Rao inequalities and characterizations of q-Gaussian distributions supporting the extended parameter ranges.","marker":"[28]"}],"fun_headline_variants":["Upper moments and down-Fisher measures sharpen classical bounds","New functionals extend sharp inequalities beyond stretched Gaussians","Dictionary trick yields optimal constants for new info inequalities","Generalized Beta solves new sharp moment and Stam bounds","Heavy-tailed densities hit sharp bounds via transformed functionals"],"cache_read_input_tokens":31104,"weakest_assumption_plain":"The load-bearing premise is that the identities connecting the up and down transforms to moments, entropies, and Fisher information continue to hold on the extended parameter ranges and for heavy-tailed or edge-divergent densities, and the paper takes those identities from its companion work without fully proving the required regularity conditions.","fun_headline_variants_meta":{"raw":{"variants":["Upper moments and down-Fisher measures sharpen classical bounds","New functionals extend sharp inequalities beyond stretched Gaussians","Dictionary trick yields optimal constants for new info inequalities","Generalized Beta solves new sharp moment and Stam bounds","Heavy-tailed densities hit sharp bounds via transformed functionals"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000217,"raw_usage":{"total_tokens":1439,"prompt_tokens":949,"completion_tokens":490,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":565,"completion_tokens_details":{"reasoning_tokens":414}},"tokens_in":565,"tokens_out":490,"duration_ms":5397,"temperature":1.0,"reasoning_tokens":414,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:44:40.282794+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a decreasing heavy-tailed density such as $f(x)=C(1+|x|)^{-\\eta}$ with $1<\\eta<2$, choose parameters in the mirrored range (4.5), and compute both sides of inequality (4.3) directly; if $(m_{p^*,\\alpha}[f]/\\sigma_q[f])^{\\Theta_1(p,\\alpha,q)}$ ever falls below the claimed constant $\\kappa^{(-1)}_{p,\\alpha,q}$, then the transformation identity fails on that range and the theorem's claim collapses.","supporting_citations":[{"cited_title":"Zozor, D","cited_arxiv_id":null,"evidence_quote":"Supplies the generalized bi-parametric Stam inequality used in the down-Fisher–Fisher proof."},{"cited_title":"A new pair of transformations and applications to generalized informational inequalities and Hausdorff moment problem","cited_arxiv_id":"2503.13686","evidence_quote":"Companion preprint that supplies the up/down transformations, the Lemma 2.1 identities (Eqs. (2.28)-(2.29)), and the tri-parametric Stam inequality on which the new proofs are built."},{"cited_title":"Lutwak, D","cited_arxiv_id":null,"evidence_quote":"Provides the classical Cramér-Rao and moment-entropy inequalities for Rényi entropy and generalized Fisher information, with stretched-Gaussian minimizers used as the starting point."},{"cited_title":"Lutwak, D","cited_arxiv_id":null,"evidence_quote":"Establishes the moment-entropy inequalities whose transformed versions underlie Theorems 4.1 and 4.6."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the generalized Fisher information and the q-Gaussian minimizers used in the Stam-type inequalities."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides generalized Cramér-Rao inequalities and characterizations of q-Gaussian distributions supporting the extended parameter ranges."}],"review_version":1}