{"id":"624070c5-8e4b-4784-b7ff-a2c28ea81eda","arxiv_id":"2501.13695","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A survey with new examples and a new Popoviciu-type inequality for strongly superadditive functions, derived from weak majorization.","lead":"Mathematicians study functions that preserve or reverse the usual behavior of addition. This paper catalogs the stricter strongly subadditive and strongly superadditive versions of this property, adds new examples, and proves a new inequality family.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Corollary 7 is false as stated (N=1, A=C=0, B=10, p=1), so the advertised determinant application is not a consequence of Theorem 8.","rationale":"The reader's weakest assumption (monotonicity in Theorem 8) is legitimate but not a demonstrated flaw: the proof needs exactly that assumption to get the first partial-sum inequality, and the two-term weak majorization step then works. A more decisive issue is that Corollary 7, a displayed application of Theorem 8, is false as stated, with a one-dimensional counterexample. This does not discredit Theorem 8 itself, but it means the paper's advertised determinant consequence must be corrected or removed before the paper is used as a reference. The reader's CONDITIONAL verdict is therefore still appropriate, but for a different reason than the one emphasized in the reader's weakest_assumption.","tokens_in":14110,"tokens_out":11142,"duration_ms":97385,"concrete_test":"Check Corollary 7 with N=1, A=C=0, B=10, p=1. If the displayed inequality fails, re-derive Corollary 7 from the symmetrized inequality in Theorem 8 by substituting F=(det)^p and correcting the formula to ((det A)^p+(det B)^p+(det C)^p)/3 + det(A+B+C)^p ≥ (2/3)[(det(A+B))^p+(det(B+C))^p+(det(A+C))^p].","verdict_should_be":"UNCHANGED","load_bearing_attack":"Corollary 7 in §5 is not a valid consequence of Theorem 8 and is false as written. For N=1, Sym+(1,R) is [0,∞) and det(X)=X. Take p=1, A=0, B=10, C=0. The claimed inequality becomes det(A)^p + det(C)^p + det(A+B+C)^p = 0+0+10 = 10, while (2/3)[det(A+B)^p + det(B+C)^p + det(A+C)^p] = (2/3)(10+10+0) = 40/3, so 10 ≥ 40/3 is false. The symmetrization in Theorem 8 actually gives (F(A)+F(B)+F(C))/3 + F(A+B+C) ≥ (2/3)(F(A+B)+F(B+C)+F(A+C)); multiplying by 3 yields F(A)+F(B)+F(C)+3F(A+B+C) ≥ 2(...). Corollary 7 has neither F(B) nor the factor 3 on F(A+B+C), so the displayed determinant inequality is not the Popoviciu-type consequence described. This is a concrete error in the paper's headline application, independent of the monotonicity assumption.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates strongly subadditive and strongly superadditive functions on convex cones. It gives one-variable characterizations (Theorems 1 and 2), Hessian characterizations in several variables (Theorem 4), a connection to submodularity, and constructions from complete monotone functions. Its main new result is Theorem 8, which uses the Tomić–Weyl weak majorization theorem to generate Popoviciu-type inequalities from monotone strongly superadditive functions. The paper also states a number of examples and applications, including determinant inequalities.","tokens_in":14372,"tokens_out":16440,"duration_ms":124601,"significance":"The core theoretical contribution, Theorem 8, appears correct and provides a clean mechanism for generating two- and three-term functional inequalities from monotone strongly superadditive functions. The paper also collects a useful set of examples, several with proofs, such as Corollaries 1, 2, and 4. However, the manuscript contains two significant errors in its advertised applications: Corollary 7 is numerically false, and Corollary 3 is inconsistent with the paper's own definition of strong subadditivity and with Remark 3. Because these are load-bearing claims, the manuscript requires major revision before it can be considered for publication.","major_comments":[{"comment":"Corollary 7 is false as stated. For N=1, Sym+(1,R)=[0,∞) and det(X)=X. Taking p=1, A=0, B=10, and C=0, the claimed inequality becomes 0+0+10 ≥ (2/3)(10+10+0)=40/3, which is false. Moreover, this inequality does not follow from the symmetrized Popoviciu inequality in Theorem 8: applying that theorem with Φ=det and f(t)=t^p yields (det A)^p+(det B)^p+(det C)^p+3(det(A+B+C))^p ≥ 2[(det(A+B))^p+(det(B+C))^p+(det(A+C))^p], which contains an extra (det B)^p term and a factor 3 on the final determinant term. The corollary should be corrected to match the actual consequence of Theorem 8, or withdrawn.","section":"§5, Corollary 7"},{"comment":"Corollary 3 is inconsistent with the paper's definition of strong subadditivity. In Section 1, a function is called strongly subadditive only if it also satisfies the subadditivity inequality (1.1). For N=1 and A1=1, Φ(x)=log x is not subadditive (e.g., Φ(1+1)=log 2 > 0 = Φ(1)+Φ(1)), yet Corollary 3 asserts that Φ is strongly subadditive. This contradiction is explicitly acknowledged in Remark 3, which states that log det is neither subadditive nor superadditive but satisfies the 2-monotone inequality (1.2). Corollary 3 should be reworded to assert only (1.2), or the definition of strong subadditivity should be revised consistently throughout the paper.","section":"§2, Corollary 3"}],"minor_comments":[{"comment":"The statement of Theorem 7 is ambiguous: the phrase \"while when b1 ≤ ... ≤ bn, then the conclusion works in the reverse direction\" is imprecise regarding the required ordering of the a's and the direction of the inequality. The proof uses only the first clause, so the statement could be clarified by stating the standard weak majorization theorem in full.","section":"§5, Theorem 7"},{"comment":"There is an extra parenthesis in the line \"Φ is differentiable on Sym++(N,R))\", which should read \"Sym++(N,R)\".","section":"§2, Proof of Corollary 2"},{"comment":"Theorem 3 states that Φ is defined on the interior cone E++ but then refers to Φ(0); the paper should clarify that Φ is defined on the closure of the positive cone or that Φ(0) is interpreted as the value at the origin when the origin is in the domain.","section":"§2, Theorem 3"},{"comment":"The sentence \"The same works for the function Φβ\" is vague about the precise set of exponents for which the claimed strong superadditivity holds; a more explicit statement would improve readability.","section":"§4"}],"recommendation":"major_revision","confidential_remarks":"The paper's central theorem (Theorem 8) is sound, but the headline applications (Corollaries 3 and 7) are incorrect as written. These are fixable: Corollary 3 can be restated as a 2-monotonicity statement, and Corollary 7 can be corrected to the actual consequence of Theorem 8. The errors appear to be overstatements rather than deep flaws, but they must be fixed before publication. The referee sees no reason to doubt the author's good faith."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things. First, Theorem 8 is real: the Popoviciu-type inequality for monotone strongly superadditive functions is new, and the proof via Tomić–Weyl weak majorization is correct. The comonotonic log-sum-exp observation in Remark 4 is also new and looks right. Second, the paper contains two concrete false statements. Corollary 3 claims log det is strongly subadditive, which is false under the paper's own definition because log det is not subadditive (Remark 3 admits this). Corollary 7, the advertised determinant application of Theorem 8, is false as written: for N=1, A=C=0, B=10, p=1, it asserts 10 ≥ 40/3. The stress-test note is right. The symmetrized Theorem 8 gives F(A)+F(B)+F(C)+3F(A+B+C) ≥ 2(F(A+B)+F(B+C)+F(A+C)), not the displayed inequality. So the headline application is not a consequence of Theorem 8.\n\nWhat the paper does well: it is a readable survey of strongly sub/superadditive functions, with a clean catalogue of examples (trace powers, determinant, Shannon entropy, complete monotone functions) and several correct characterizations (Theorems 1–4, Corollaries 1–2, 4–6). The proofs are short and mostly self-contained. Theorem 8, once separated from Corollary 7, is a legitimate extension of the majorization program and could be useful.\n\nSoft spots beyond the false statements: Theorem 3 has a domain ambiguity—it states Φ on E++ but uses Φ(0) and strong additivity on the closed cone. The monotonicity assumption in Theorem 8 is real and not discussed; without it the weak majorization step fails. The paper also leans heavily on the author's own textbook [20], though the cited background claims do not seem circular.\n\nWho this is for: anyone working on functional inequalities, matrix analysis, or entropy-like functions will find Theorem 8 and the catalogue valuable. But the paper is not ready as a reference in its current form. The false corollaries need correcting, and the determinant application should be either fixed or withdrawn. I would send it to review because the core idea is sound and new, but I would expect a revision.\n\nRecommendation: reject from desk; send to referee with a clear request to check Corollaries 3 and 7 and the domain issue in Theorem 3.","headline":"Theorem 8 is a genuine and correct new inequality, but the paper as posted has two concrete internal errors—Corollary 3 and Corollary 7—that make it unsafe as a reference until fixed.","tokens_in":14884,"tokens_out":2359,"would_cite":false,"duration_ms":20667,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["26A51","39B62","46B20","26D15"],"pacs":[],"model":"deepseek-v4-flash","headline":"A monotone strongly superadditive function on a convex cone, composed with any nondecreasing convex function, yields a two-term inequality that symmetrizes to a Popoviciu-type inequality.","keywords":["strongly subadditive functions","strongly superadditive functions","convex cones","weak majorization","Popoviciu inequality","completely monotone functions","determinant inequality","von Neumann entropy"],"falsifier":"Check whether monotonicity is really needed: let $C=\\mathbb{R}_+^2$, $\\Phi(x_1,x_2)=x_1^2+x_2^2-2x_1-2x_2$, $f(t)=\\max(t,0)$, $x=(1+\\sqrt{3/2},0)$, $y=(0,1)$, and $z=(0,0)$. The Hessian of $\\Phi$ has nonnegative entries and $\\Phi(0)=0$, so $\\Phi$ is strongly superadditive, yet $\\Phi$ is not monotone. The claimed inequality becomes $0\\ge 1/2$, which is false.","tokens_in":2102,"feed_emoji":"➗","tokens_out":4078,"duration_ms":170362,"temperature":0.7,"pith_summary":"Strongly subadditive and superadditive functions, whose second-order differences have a definite sign on every triple in a convex cone, are far less studied than ordinary subadditive and superadditive functions, even though the class includes von Neumann entropy, the determinant on positive definite matrices, and trace powers. The paper maps this territory by characterizing these functions in one variable, in several variables via Hessian signs, and through their connection to submodularity and completely monotone functions. Its main new result, Theorem 8, says that a monotone strongly superadditive function, pushed through any nondecreasing convex function, gives a three-variable inequality of Popoviciu type. If correct, this turns a wide stock of known examples into generators of explicit inequalities in matrix analysis, probability, and lattice theory.","feed_headline":"Monotone superadditive maps yield new inequalities","feed_subtitle":"Determinants, traces, and Lp norms become generators of Popoviciu-type three-variable inequalities.","key_machinery":"The load-bearing machinery is the weak majorization theorem of Tomi\\'c and Weyl: for a nondecreasing convex $f$, when a decreasing sequence of numbers has partial sums bounded above by another sequence, the sum of $f$ over the first sequence is bounded above by the sum over the second. Theorem 8 feeds this theorem the four numbers $\\Phi(y+z)$, $\\Phi(x+z)$, $\\Phi(x+y+z)$, and $\\Phi(z)$; strong superadditivity supplies the total-sum comparison, and monotonicity supplies the ordering needed to apply weak majorization. Around this, the paper uses second-order difference operators $\\Delta_x\\Delta_y\\Phi(z)$ to characterize strong subadditivity and superadditivity, and uses the Laplace-transform representation of completely monotone functions to generate examples that satisfy the hypotheses.","core_discovery":"The paper's central assertion is Theorem 8: if $C$ is a convex cone, $\\Phi:C\\to[0,\\infty)$ is strongly superadditive, meaning $\\Phi(x+y+z)+\\Phi(z)\\ge \\Phi(x+z)+\\Phi(y+z)$ for all $x,y,z\\in C$, and $\\Phi$ is monotone in the cone order, then for every nondecreasing convex function $f$ and all $x,y,z\\in C$, $f(\\Phi(x+y+z))+f(\\Phi(z))\\ge f(\\Phi(x+z))+f(\\Phi(y+z))$. Symmetrizing this two-term inequality yields a Popoviciu-type inequality. The author presents this as a new functional-inequality principle, with the determinant, trace functions, $L^p$ norms, and shifted completely monotone functions as instances.","pith_inferences":["Editor's inference: The theorem can be read as an inequality machine: each monotone strongly superadditive $\\Phi$ and each nondecreasing convex $f$ on its range gives valid two- and three-term inequalities, so choosing $f(t)=e^t$, $f(t)=t^p$, or $f(t)=\\max(t,0)$ expands the paper's determinant and trace corollaries into families not listed there.","Editor's inference: In the one-dimensional case, convexity plus $\\Phi(0)\\le 0$ automatically makes $\\Phi$ strongly superadditive and renders the monotonicity condition redundant, so the genuinely new content of Theorem 8 lies in cones of dimension two or higher, where checking monotonicity via the differential (as in Theorem 3) is the practical entry point.","Editor's inference: The combination of strong superadditivity and monotonicity behaves like a discrete convexity in the cone order, and the paper's link to submodularity suggests that Popoviciu-type inequalities may also hold for monotone submodular set functions or Choquet integrals, where monotonicity is often built into the model."],"forward_implications":["Corollary 7: for all $A,B,C\\in\\mathrm{Sym}_+(N,\\mathbb{R})$ and $p\\ge 1$, $(\\det A)^p+(\\det C)^p+\\det(A+B+C)^p \\ge \\frac{2}{3}\\left[(\\det(A+B))^p+(\\det(B+C))^p+(\\det(A+C))^p\\right]$.","For $p\\in[1,2]$, $\\mathrm{trace}(A^p)$ is strongly superadditive on positive semidefinite matrices, and for $p\\in[0,1]$ it is strongly subadditive, so Theorem 8 produces matching functional inequalities for both ranges.","For $p>1$, the function $f\\mapsto \\|f\\|_p^p$ is strongly superadditive on the positive cone of $L^p(\\mathbb{R})$, giving Banach-lattice inequalities.","The log-sum-exp function is not strongly superadditive in general but is comonotonic strongly superadditive, so the theorem's inequalities hold for pairwise comonotonic triples, as noted in Remark 5.","Completely monotone functions, shifted and centered at their origin value, become strongly superadditive; applying Theorem 8 to them yields explicit inequalities for functions such as $(\\det(I+A))^{-\\beta}-1$."],"supporting_citations":[{"why":"Supplies the weak-majorization theorem (Theorem 6.1.4) that drives the proof of Theorem 8, along with background on convex functions.","marker":"[20]"},{"why":"The classical Lieb-Ruskai proof of strong subadditivity of von Neumann entropy, which motivates the entire class.","marker":"[15]"},{"why":"Source of the fact, used in Corollary 2, that the determinant is strongly superadditive on positive semidefinite matrices.","marker":"[28]"},{"why":"Provides complete-monotonicity criteria and integral representations used in Section 4 to produce strongly superadditive functions.","marker":"[25]"},{"why":"The Bernstein-Hausdorff-Widder-Choquet representation theorem, used to show completely monotone functions are 2-monotone increasing.","marker":"[7]"},{"why":"Names the Popoviciu-type inequality that Theorem 8's symmetrization reproduces.","marker":"[22]"},{"why":"Establishes strong subadditivity and superadditivity of $\\mathrm{trace}(A^p)$, one of the paper's main example families.","marker":"[4]"},{"why":"Proves the operator monotonicity of $(1+t^p)^{1/p}$, used to build a strongly superadditive trace function.","marker":"[12]"}],"fun_headline_variants":["Superadditive maps and convex functions yield new inequalities","Convex functions amplify superadditive maps into inequalities","New Popoviciu-type inequalities from superadditive maps","Monotone superadditive maps: convex transforms create inequalities","Superadditivity + convexity = Popoviciu-type inequalities"],"cache_read_input_tokens":17024,"weakest_assumption_plain":"The load-bearing premise is that $\\Phi$ is monotone in the cone order ($x\\le y$ implies $\\Phi(x)\\le\\Phi(y)$); without it, the entries fed into weak majorization cannot be ordered, and the claimed inequality can fail even for strongly superadditive functions.","fun_headline_variants_meta":{"raw":{"variants":["Superadditive maps and convex functions yield new inequalities","Convex functions amplify superadditive maps into inequalities","New Popoviciu-type inequalities from superadditive maps","Monotone superadditive maps: convex transforms create inequalities","Superadditivity + convexity = Popoviciu-type inequalities"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000621,"raw_usage":{"total_tokens":2762,"prompt_tokens":714,"completion_tokens":2048,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":330,"completion_tokens_details":{"reasoning_tokens":1965}},"tokens_in":330,"tokens_out":2048,"duration_ms":14036,"temperature":1.0,"reasoning_tokens":1965,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T15:42:13.191097+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check whether monotonicity is really needed: let $C=\\mathbb{R}_+^2$, $\\Phi(x_1,x_2)=x_1^2+x_2^2-2x_1-2x_2$, $f(t)=\\max(t,0)$, $x=(1+\\sqrt{3/2},0)$, $y=(0,1)$, and $z=(0,0)$. The Hessian of $\\Phi$ has nonnegative entries and $\\Phi(0)=0$, so $\\Phi$ is strongly superadditive, yet $\\Phi$ is not monotone. The claimed inequality becomes $0\\ge 1/2$, which is false.","supporting_citations":[{"cited_title":"A Contemporary Approach","cited_arxiv_id":null,"evidence_quote":"Supplies the weak-majorization theorem (Theorem 6.1.4) that drives the proof of Theorem 8, along with background on convex functions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The classical Lieb-Ruskai proof of strong subadditivity of von Neumann entropy, which motivates the entire class."},{"cited_title":"2nd ed., Springer, New York (2011) Department of Mathematics, University of Craiova, Craiova 2 00585, Romania Email address : constantin.p.niculescu@gmail.com","cited_arxiv_id":null,"evidence_quote":"Source of the fact, used in Corollary 2, that the determinant is strongly superadditive on positive semidefinite matrices."},{"cited_title":"Acta Math","cited_arxiv_id":null,"evidence_quote":"Provides complete-monotonicity criteria and integral representations used in Section 4 to produce strongly superadditive functions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The Bernstein-Hausdorff-Widder-Choquet representation theorem, used to show completely monotone functions are 2-monotone increasing."},{"cited_title":"Analele ¸ stiint ¸ifice Univ","cited_arxiv_id":null,"evidence_quote":"Names the Popoviciu-type inequality that Theorem 8's symmetrization reproduces."},{"cited_title":"In: International Conferen ce on Engineering Optimization (EngOpt 2008), Rio de Janeiro, June 1–5, 2008, Paper # 0615, P rinted and CDROM Pro- ceedings, ISBN 978-85-7650-152-7","cited_arxiv_id":null,"evidence_quote":"Establishes strong subadditivity and superadditivity of $\\mathrm{trace}(A^p)$, one of the paper's main example families."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the operator monotonicity of $(1+t^p)^{1/p}$, used to build a strongly superadditive trace function."}],"review_version":1}