{"id":"d6af1381-4750-4608-a117-7dc96094e0cc","arxiv_id":"1908.01865","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper states new distribution characterizations based on identical distribution or independence of linear forms with random coefficients, but does not fully prove all of them.","lead":"This paper gives new characterizations of two-point, uniform, and logistic-like random distributions using linear forms with random coefficients. The results are plausible but several proofs are omitted and one proof contains a flawed completeness step, so the paper needs revision.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.1, the squared hyperbolic secant characterization, is stated with no proof; that central claim is unverified as written.","rationale":"The reader's weakest_assumption is the completeness of the metric space F in Section 3. The completeness claim is indeed false, as a Cauchy sequence in that metric can converge to a function of the form t^3 φ(t) with φ merely continuous. However, the contraction mapping theorem is not actually needed for the conclusion: a contraction on any metric space has at most one fixed point, and both K(t) and cos(t) are elements of F that satisfy K=BK and cos=Bcos. The inequality d(Bg1,Bg2)≤1/2 d(g1,g2) then forces K=cos. So the completeness error is an exposable flaw but not the load-bearing gap. The genuinely load-bearing gap is Theorem 4.1, whose proof is explicitly omitted; this is a central characterization and cannot be checked from the manuscript. There is also an internal inconsistency in the statement of Theorem 1.2: for even n the stated support and probabilities do not match equation (1.5) and do not sum to 1, and for n=2 the stated P[ε=2]=1 would make the identity hold for Cauchy-type distributions rather than characterize two-point laws. These issues make the paper not acceptable as written, but the intended results appear plausible and repairable, so a conditional acceptance contingent on supplying the missing proof and correcting Theorem 1.2 is more appropriate than an outright rejection.","tokens_in":3939,"tokens_out":25373,"duration_ms":229665,"concrete_test":"Reconstruct the proof of Theorem 4.1: from the distributional identity write the characteristic-function equation 1/2(f(t)+f(t/2))f^2(t/2)=f(t)f^2(t/4); set K(t)=f(t)/f(t/2) and reduce to K^2(t/2)(K(t)+1)=2K(t). Solve this equation under the stated assumptions on f (symmetric characteristic function, f(0)=1, f'(0)=0, f''(0)<0, finite third moment) and verify the only solution is K(t)=sech(t/2), so f(t)=t/sinh(t) up to scale. If the solution is not unique or the reduction fails, the theorem is unsupported.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claims include a new characterization of the squared hyperbolic secant distribution, but Theorem 4.1 is the only result supplying that characterization and its proof is explicitly omitted ('The proof of Theorem 4.1 is very similar to that of the previous Theorem 3.1 and, therefore, is omitted'). This is load-bearing: Theorem 4.1 asserts an iff statement, and no functional equation, contraction argument, or uniqueness proof is given from which a reader could check that the identity in distribution between L1=ξX1+1/2(X2+X3) and L2=X1+1/4(X2+X3) forces f(t)=t/sinh(t) up to scale. It also leaves open whether other characteristic functions could satisfy the same identity under the stated finite-third-moment assumption. The reader's cited weakness, the false completeness claim for the metric space F in Section 3, is real but not fatal: the contraction inequality d(Bg1,Bg2)≤1/2 d(g1,g2) alone gives uniqueness, and since both K(t) and cos(t) are fixed points in F, K=cos(t) follows without invoking completeness. Thus the most load-bearing unresolved point is the missing proof of Theorem 4.1, not the Section 3 completeness assertion.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes characterizations of three distribution families—two-point, uniform, and squared hyperbolic secant—through identities or independences of linear forms with random coefficients. Theorem 1.1 characterizes the two-point distribution by the identity εX1 =d 1/2(X1+X2), and Theorem 1.2 gives an n-variable analogue; Theorem 2.1 characterizes the two-point distribution by independence of two linear forms, reducing to Theorem 1.1. Theorem 3.1 claims that L1 = ξX1 + 1/4(X2+X3) and L2 = 1/2(X1+X2+X3) are identically distributed if and only if X is uniform on a symmetric interval, with a proof based on a contraction argument for K(t)=f(t)/f(t/2). Theorem 4.1 makes an analogous claim for the squared hyperbolic secant distribution, but its proof is omitted.","tokens_in":4254,"tokens_out":12398,"duration_ms":109736,"significance":"If the characterizations hold, they add to a small literature on distributional identities with random coefficients and extend the intensively monotone operator method beyond the Gaussian and hyperbolic secant cases. The paper's clean reduction in Theorem 2.1 and the product formula in Section 3 are attractive. However, the presentation is not self-contained: the main tool is a citation to the author's book [2], Theorem 4.1 is unproved, and the proof of Theorem 3.1 contains a false completeness assertion. The results are plausible, but as written the paper does not rigorously establish the claimed characterizations.","major_comments":[{"comment":"Theorem 4.1 is the sole characterization of the squared hyperbolic secant distribution, yet its proof is omitted with only the sentence \"The proof of Theorem 4.1 is very similar to that of the previous Theorem 3.1 and, therefore, is omitted.\" No functional equation, contraction argument, or uniqueness proof is given, so the iff claim is unverified. At minimum the authors must provide the characteristic-function equation analogous to (3.1), the definition of the appropriate K(t), and the uniqueness argument; a reader should be able to check that the identity forces f(t)=t/sinh(t) up to scale.","section":"Section 4, Theorem 4.1"},{"comment":"The assertion that F is a complete metric space under d(g1,g2)=sup |g1(t)-g2(t)|/|t^3| is false. A Cauchy sequence of C^3 symmetric functions in this metric need not converge to a C^3 limit (it may converge to a function with only C^1 or worse regularity), so the contraction mapping theorem cannot be applied as stated. Nevertheless, the contraction inequality d(Bg1,Bg2) ≤ 1/2 d(g1,g2) is sufficient to establish uniqueness of a fixed point in F because both K(t) and cos(t) are fixed points: if two fixed points existed, their d-distance would be at most half itself. Please replace the completeness-based argument with this direct uniqueness argument.","section":"Section 3, after Eq. (3.3)"},{"comment":"The support of εn is misstated relative to the probabilities in (1.4)-(1.5). For odd n, the probabilities assign mass to values n, n-2, ..., 1, but the text says εn takes values 0,2,...,n-1. For even n, the text says values 0,2,...,n-2 while also giving P[εn=n]; for n=3 the stated support and probabilities are already inconsistent. In addition, the proof is only the sentence that equations (1.4)-(1.5) can be solved similarly; a full reduction to Theorem 1.1 or to the intensively monotone operator method must be given. The theorem cannot be accepted in its current form.","section":"Theorem 1.2, lines after the statement"},{"comment":"The proof of Theorem 1.1 is a pointer to [2]: after defining the operator A, it says that it is easy to check that A is intensively monotone and that the family {cos(at)} is strongly E+-positive, and then appeals to Theorem 1.1.1 of [2]. Since this theorem is the foundation for Theorem 2.1 and the later sections, the authors should either state and verify the hypotheses of [2, Theorem 1.1.1] explicitly or give an elementary proof of (1.2) (e.g., by iterating the functional equation for characteristic functions).","section":"Theorem 1.1 proof"}],"minor_comments":[{"comment":"The abstract contains a typo: 'rel ated' should read 'related'.","section":"Abstract"},{"comment":"'each with probabilities tf rac12' is a LaTeX artifact; it should read 'each with probability 1/2'.","section":"Section 2"},{"comment":"The characteristic-function computation leading to (3.1) is not shown; adding one line would help the reader verify the identity.","section":"Section 3, before Eq. (3.1)"},{"comment":"The solution is written as 'g(t)=sin(At)/(At)', but the characteristic function is elsewhere denoted f; please use a consistent symbol.","section":"Section 3, after Eq. (3.1)"},{"comment":"The joint characteristic function is written with an unmatched closing parenthesis; the expression should be E[exp{i(s+t)εX1 + i(s-t)(1-ε)X2}] = 1/2(f(s+t)+f(s-t)).","section":"Section 2, display equation"},{"comment":"The condition |g(t)|≤1 is not justified. For a general characteristic function f with finite third moment, why is |K(t)|≤1 on [-T,T]? This follows from continuity near 0 for sufficiently small T, but the authors should state it.","section":"Section 3, definition of F"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on the author's own prior work [1,2] for the main method and for the proof of Theorem 1.1; this is not by itself a defect, but it makes the paper less self-contained. If Theorem 4.1 cannot be given a complete proof in revision, the paper should be rejected, since that theorem is a central claimed contribution. The reference [2] is from 1984 and may not be readily accessible; providing the arguments would improve the paper's suitability for a general statistics journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper is a mixed bag: it has at least one new characterization that looks right, but it also leaves its most novel theorem unproved and makes a false completeness claim that is embarrassingly easy to fix. The paper itself is only six pages, so the gaps stand out.\n\nWhat is actually new: the characterizations for the uniform distribution (Theorem 3.1) and the squared hyperbolic secant distribution (Theorem 4.1). The uniform one is the real contribution. The idea of writing K(t) = f(t)/f(t/2) and reducing the problem to a cosine equation is clever, and the contraction argument, once repaired, does the job. Theorem 2.1 is a clean reduction to Theorem 1.1, and Theorem 1.2 is a reasonable extension, though its proof is only sketched.\n\nWhere the soft spots are: Theorem 4.1 is stated with no proof at all. The author says it is 'very similar' to Theorem 3.1, but it is not identical—the coefficients in the linear forms differ, and the resulting functional equation is not the same. That is a load-bearing hole. A reader cannot verify the characterization without redoing the analysis from scratch. Theorem 1.2 also gets a one-line proof sketch, which is thin but less serious because the pattern is clear. The completeness claim for the metric space F in Section 3 is false as stated: a Cauchy sequence in that metric need not converge to a three-times differentiable function. But the stress-test note is right—the contraction inequality d(Bg1,Bg2) ≤ 1/2 d(g1,g2) gives uniqueness directly, since both K and cos are fixed points in F. So that flaw is minor and easily patched. The reliance on [2] for the intensive monotone operator method is fine; that is a known technique, and the citation is appropriate.\n\nOverall, the paper is worth engaging with. The uniform characterization is likely correct and interesting to specialists. The squared hyperbolic secant result is plausible but unverified. The writing is clear and the author is honestly extending his own prior work, not overclaiming beyond what the method can support.\n\nRecommendation: send it to peer review, but require full proofs for Theorems 1.2 and 4.1, and a correction to the completeness statement. A serious referee can sort out the gaps without too much trouble. I would not cite it as a source for either characterization until the proofs appear.","headline":"A short paper with a genuinely new uniform-distribution characterization and a plausible but unproven squared-hyperbolic-secant analogue; the proof gaps are real but mostly fixable.","tokens_in":4626,"tokens_out":2552,"would_cite":false,"duration_ms":26864,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["62E10","60E10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that simple identities among random-coefficient linear forms force exactly the two-point, uniform, and squared hyperbolic secant distributions.","keywords":["characterization of distributions","two-point distribution","uniform distribution","squared hyperbolic secant distribution","linear forms with random coefficients","independence of linear forms","identical distribution of linear forms","intensively monotone operators"],"falsifier":"Search numerically for a symmetric non-degenerate distribution with finite absolute third moment whose characteristic function solves $\\tfrac12(f(t)+f(t/2))f^2(t/4)=f^3(t/2)$ but is not $\\sin(At)/(At)$; any such solution would disprove Theorem 3.1. The search can be run by evaluating the equation on mixtures of uniform or other symmetric distributions and checking whether the two linear forms $L_1$ and $L_2$ are identically distributed.","tokens_in":3766,"feed_emoji":"🎲","tokens_out":11402,"duration_ms":105882,"temperature":0.7,"pith_summary":"The paper sets out to show that a few very simple distributional identities involving linear forms with random coefficients are rigid: each identity forces the underlying random variable to belong to exactly one of three families. The first family, in Theorems 1.1, 1.2 and 2.1, is the two-point distribution taking $\\pm a$ with equal probability; the second, in Theorem 3.1, is the uniform distribution on a symmetric interval; the third, in Theorem 4.1, is the squared hyperbolic secant distribution. The proofs translate each distributional identity into a functional equation for the characteristic function and then solve that equation uniquely. The characterizations are meant to be exact fingerprints: observing the identity in data is enough to name the law, up to a scale parameter.","feed_headline":"Two linear-form identities pin down three classical laws","feed_subtitle":"Equalities linking random-coefficient sums single out three exact distributions.","key_machinery":"The argument is carried by converting a distributional identity into a functional equation for the characteristic function $f$, then proving uniqueness of its solution. The two-point theorems reduce to the equation $\\tfrac12(g(t)+1)=g^2(t/2)$, whose solution $g(t)=\\cos(at)$ is selected by the method of intensively monotone operators. The uniform theorem introduces the ratio $K(t)=f(t)/f(t/2)$, which obeys the same equation $K(t)=2K^2(t/2)-1$; uniqueness is obtained by making the operator $Bg=2g^2(t/2)-1$ a contraction on the space $F$ of smooth symmetric functions with $g(0)=1$, $g''(0)=-1$, using the distance $d(g_1,g_2)=\\sup_t |g_1(t)-g_2(t)|/|t^3|$. The fixed point is $K(t)=\\cos(t)$, and iterating $f(t)=\\cos(at)f(t/2)$, $f(t)=\\cos(at)\\cos(at/2)f(t/4)$, and so on gives $f(t)=\\sin(2at)/(2at)$, the characteristic function of a uniform variable.","core_discovery":"On the paper's own terms, the discovery is that distributional equality or independence of particular linear statistics with random coefficients completely determines the law of the underlying variable. For two independent copies, the equality $\\varepsilon X_1 \\stackrel{d}{=} \\tfrac12(X_1+X_2)$, with $\\varepsilon$ taking values $0$ and $1$ with equal probability, holds if and only if $X$ takes only two values $\\pm a$ with equal probability; the same conclusion follows if $\\varepsilon X_1+(1-\\varepsilon)X_2$ and $\\varepsilon X_1-(1-\\varepsilon)X_2$ are independent. For three copies, the identity $\\xi X_1+\\tfrac14(X_2+X_3)\\stackrel{d}{=}\\tfrac12(X_1+X_2+X_3)$, with $\\xi$ taking values $\\tfrac12$ and $1$ with equal probability, forces $X$ to be uniform on $(-A,A)$, and rebalancing the coefficients to $\\xi X_1+\\tfrac12(X_2+X_3)\\stackrel{d}{=}X_1+\\tfrac14(X_2+X_3)$ forces the squared hyperbolic secant law with characteristic function $t/\\sinh t$, up to scale. Theorem 1.2 extends the two-point characterization to $n$ independent summands with a specially chosen random coefficient $\\varepsilon_n$. The proof of Theorem 4.1 is omitted, with the paper noting that it is very similar to the uniform case.","pith_inferences":["Changing the coefficients in the linear forms is a natural experiment: the ratio method $K(t)=f(t)/f(t/2)$ may characterize a whole family of laws whose characteristic functions are infinite products of cosines or hyperbolic functions, of which uniform and squared hyperbolic secant are two examples.","The independence version in Theorem 2.1 can be read as an identifiability result for source separation: if two random-coefficient mixtures of a symmetric source are independent, the source must be two-point, so the theorem gives a testable criterion for discrete sources.","A concrete next step is to test whether the uniform characterization survives with only a finite second moment; if it does, the essential condition is the distributional identity itself rather than the third-moment regularity used in the proof."],"forward_implications":["The two-point characterizations give a distributional test: if a symmetric variable satisfies either the equality in law or the independence condition, it is exactly a two-point variable, up to scale.","The uniform theorem characterizes uniformity with no density assumption, using only symmetry, a finite third absolute moment, and equality in distribution of two random-coefficient linear forms.","The squared hyperbolic secant characterization extends the same random-coefficient technique to a continuous heavy-tailed law, so the method is not limited to compactly supported distributions.","Together with the earlier hyperbolic secant characterization, these results suggest that random-coefficient linear-form identities form a unified way to single out classical distributions across discrete, compact, and unbounded cases."],"supporting_citations":[{"why":"Supplies the preceding characterization of the hyperbolic secant distribution by identical distribution and independence of linear forms with random coefficients, the pattern the present paper adapts.","marker":"[1]"},{"why":"Provides the method of intensively monotone operators and the strong E+-positivity theorem used to conclude that cos(at) is the only solution of the functional equations in Theorems 1.1 and 1.2.","marker":"[2]"}],"fun_headline_variants":["Random-coefficient sums single out three exact distributions","Three laws from two random-coefficient identities","Linear forms and independence pin down three laws","Two identities, three laws: two-point, uniform, SHSC","Random coefficients pin down two-point, uniform, and SHSC"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The uniform-distribution proof depends on the claim that the set of functions used in the contraction argument has no gaps under the distance $d(g_1,g_2)=\\sup_t |g_1(t)-g_2(t)|/|t^3|$; the paper states this completeness without proof, and if that claim fails the argument that $K(t)=\\cos(t)$ is the only fixed point no longer follows.","fun_headline_variants_meta":{"raw":{"variants":["Random-coefficient sums single out three exact distributions","Three laws from two random-coefficient identities","Linear forms and independence pin down three laws","Two identities, three laws: two-point, uniform, SHSC","Random coefficients pin down two-point, uniform, and SHSC"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001356,"raw_usage":{"total_tokens":5470,"prompt_tokens":878,"completion_tokens":4592,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":494,"completion_tokens_details":{"reasoning_tokens":4514}},"tokens_in":494,"tokens_out":4592,"duration_ms":35700,"temperature":1.0,"reasoning_tokens":4514,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:02:28.066863+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search numerically for a symmetric non-degenerate distribution with finite absolute third moment whose characteristic function solves $\\tfrac12(f(t)+f(t/2))f^2(t/4)=f^3(t/2)$ but is not $\\sin(At)/(At)$; any such solution would disprove Theorem 3.1. The search can be run by evaluating the equation on mixtures of uniform or other symmetric distributions and checking whether the two linear forms $L_1$ and $L_2$ are identically distributed.","supporting_citations":[{"cited_title":"Linear Statistics with Random Coefficients and Characterization of Hyperbolic Secant Distribution","cited_arxiv_id":"1905.09910","evidence_quote":"Supplies the preceding characterization of the hyperbolic secant distribution by identical distribution and independence of linear forms with random coefficients, the pattern the present paper adapts."},{"cited_title":"Kakosyan, L.B","cited_arxiv_id":null,"evidence_quote":"Provides the method of intensively monotone operators and the strong E+-positivity theorem used to conclude that cos(at) is the only solution of the functional equations in Theorems 1.1 and 1.2."}],"review_version":1}