{"id":"a8342780-d1c0-4e3a-b288-87d38adf64ee","arxiv_id":"1908.06313","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Boundedness of a class of Hardy-type integral operators on rearrangement-invariant spaces is equivalent to checking only non-increasing functions, with only a constant-factor loss, on the entire positive half-line.","lead":"Mathematicians usually check hard inequalities only on the simplest test functions, decreasing ones, and a reduction principle says this is enough. This paper proves such a reduction principle for a broad class of integral operators on the whole positive half-line, where the known version only worked on finite intervals.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: Theorem 1.1's proof is internally consistent; the cited Lemma 3.3 and the averaging-operator step in Theorem 3.9 do not show a gap for m=1.","rationale":"The paper's main theorem is a reduction principle for the kernel H_I f(t)=∫_t^∞ f(s)/I(s) ds. The proof is a duality argument: boundedness on the cone of non-increasing functions is dualized to an estimate on R_I g^*, and the only nontrivial norm comparison is Theorem 3.9. I traced the chain: the down-associate norm identity, the rearrangement invariance, the mutual associativity of H_I and R_I, and the use of the averaging operator A in (3.14) are consistent. For m=1 the essential-decreasing condition is automatic for non-decreasing I, so Theorem 1.1 follows from Theorem 3.10. The reader's conditional verdict is driven by dense exposition and mechanical typos; I do not see a mathematical gap. A fresh check of the averaging-operator norm bound (3.15) would guard the one external theorem used without proof, but nothing in the present text indicates that the bound fails.","tokens_in":12847,"tokens_out":45359,"duration_ms":464746,"concrete_test":"Verify that the operator A in (3.14) satisfies the Bennett--Sharpley averaging-operator hypothesis used at (3.15): for a single interval E=(c,d) and a non-increasing step function h, compute the submajorization ∫_0^t (A h)^*(s) ds ≤ ∫_0^t h^*(s) ds for all t, and confirm ‖A h‖_X ≤ ‖h‖_X for X=L^2 and X=L^∞. If this fails for any t, the constant 4 in Theorem 1.1 is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no load-bearing objection to Theorem 1.1. The two steps on which the argument most visibly leans are the one-sided estimate (3.4)/(3.5) in Lemma 3.3 and the averaging-operator norm bound (3.15) in Theorem 3.9. For m=1, (3.4) follows from monotonicity of I and non-increasingness of f^*, with constant 2; (3.15) is justified because A(g) is the conditional expectation of g^* with respect to a partition whose atoms are the intervals (c_k,d_k) and the singletons outside E, so it preserves constants and integrals and is a doubly stochastic/averaging operator. I do not find a circular step or a missing hypothesis in the m=1 case. The denser Definition 3.6 and higher-order Theorem 3.10 are more restrictive but not needed for Theorem 1.1 and are not shown to be inconsistent.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a reduction principle for a class of integral operators of the form H_I^m f(t) = ∫_t^∞ f(s)/I(s) (∫_s^t 1/I(r) dr)^{m-1} ds, with I non-decreasing and m ∈ N. The headline result (Theorem 1.1) is the case m=1: for rearrangement-invariant Banach function norms X and Y on (0,∞), boundedness of this operator on all non-negative f is equivalent to boundedness on the cone of non-increasing non-negative f, with the full constant at most 4 times the restricted one. The proof follows the strategy of Cianchi–Pick–Slavíková, using the Hardy–Littlewood inequality, down-associate norms, and an averaging-operator argument. A more general version (Theorem 3.10) is proved for all m under an additional 'essentially decreasing in t' condition on Φ_I^m(t,s) introduced in Definition 3.6.","tokens_in":13042,"tokens_out":17217,"duration_ms":162023,"significance":"If the proof is correct, the paper gives a useful extension of the reduction principle from finite-measure settings to the infinite-measure space (0,∞), which is relevant to potential operators and Riesz-type kernels. The main theorem is clean and the constant 4 for m=1 is explicit. The paper is largely self-contained and builds on classical external results; the adaptation to infinite measure is nontrivial, as the author notes. The higher-order formulation is conditional on a somewhat opaque hypothesis, but the m=1 result, which is the advertised headline, does not depend on that condition. The manuscript would be a solid contribution to the function-space literature once the technical issues below are addressed.","major_comments":[{"comment":"The operator A defined in (3.14) is not linear, because A(g) is expressed through g*, and the rearrangement map g ↦ g* is not linear. Therefore A cannot be an averaging operator in the sense of [1, Chapter 2, Theorem 4.8], and the norm bound (3.15) does not follow from the cited theorem as stated. This is load-bearing for the proof of (3.13). The argument can be repaired locally: define a genuine averaging operator T by T(h)=h χ_{(0,∞)\\E} + Σ_k (avg_{ (c_k,d_k) } h) χ_{(c_k,d_k)}, and then write A(g)=T(g*). Since T is averaging and X is rearrangement-invariant, ‖T(g*)‖_X ≤ ‖g*‖_X = ‖g‖_X, so (3.15) holds. Please rewrite the passage accordingly and adjust the surrounding notation.","section":"§3, Theorem 3.9, Eq. (3.14)–(3.15)"},{"comment":"In the displayed chain of equalities and inequalities, the factor 1/(c_k − d_k) should be 1/(d_k − c_k). As written, the second equality introduces a negative term, so the subsequent inequality is invalid. With the denominator corrected, the estimate follows by (3.14) and (3.5). This is a localized error, but it must be corrected because it occurs in the main computation of the theorem.","section":"§3, Theorem 3.9, displayed chain after (3.15)"}],"minor_comments":[{"comment":"The formula G^m_I f(t) = sup_{s≥t} R^m_I f^*(t) contains a typo: the argument of R^m_I f^* should be s, not t.","section":"§3, proof of Theorem 3.9, after (3.18)"},{"comment":"The word 'euqimeasurable' is misspelled; it should be 'equimeasurable'.","section":"§2.1, Definition 2.3"},{"comment":"In the first sentence of the proof, 'there there exists' should read 'there exists'.","section":"§3, Lemma 3.8, proof"},{"comment":"The 'essentially decreasing in t' condition is quite involved; adding a non-example or a brief discussion of how it fails for natural choices such as I(t)=1 for m>1 would help the reader calibrate its scope.","section":"§3, Definition 3.6"}],"recommendation":"major_revision","confidential_remarks":"The central claim (Theorem 1.1) appears sound, and the proof strategy is credible. The two major comments are localized and fixable; they concern the proof of Theorem 3.9 rather than the overall architecture. The published version of this article may already contain corrections; my report is based on the submitted manuscript as provided."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper delivers what it promises. It extends the Cianchi–Pick–Slavíková reduction principle from finite intervals to the whole half-line for a class of Hardy-type operators, and the m=1 case is in good shape. The higher-order generalization is real but rests on a new condition that I would want to see stress-tested before relying on it.\n\nThe main theorem (Theorem 1.1) says that for non-decreasing I, boundedness of the operator f ↦ ∫_t^∞ f(s)/I(s) ds on all nonnegative f is equivalent to boundedness on non-increasing f, with constant multiplied by 4. That is a useful tool for spaces over R^n and similar infinite-measure settings. The proof for m=1 is complete as far as I can tell: Lemma 3.3 uses monotonicity of I and rearrangement properties, and the averaging operator in Theorem 3.9 is a genuine conditional expectation over the partition, so the norm bound (3.15) is justified. I checked the stress-test note against the text and it lands: no circularity, no missing hypothesis for m=1.\n\nThe genuine novelty is in handling infinite measure. The author introduces Definition 3.6, a condition that makes the kernel Φ essentially decreasing in t, and Lemma 3.8 shows that the set E where the rearrangement and the supremum differ splits into bounded intervals. That is a real structural step, not a cosmetic rewrite of [3]. For m=1 the condition is automatic, so Theorem 1.1 does not depend on it.\n\nSoft spots: Definition 3.6 is hard to parse, and the paper does not give an example where it fails for m>1 while the reduction principle might still hold. That makes the scope of Theorem 3.10 less clear than it could be. There are also several typos in displayed formulas — one in the proof of Theorem 3.9 has c_k−d_k where it should be d_k−c_k — but they are clearly mechanical and do not affect the argument.\n\nWho this is for: people working on rearrangement-invariant spaces, Sobolev embeddings, or boundedness of Hardy-type operators. It deserves a serious referee: the m=1 result is well-supported and useful, and the higher-order case is worth careful scrutiny even if the exposition needs work. I would engage with it and send it out.","headline":"Genuinely extends the finite-interval reduction principle to infinite measure with a solid m=1 proof; the higher-order generalization rests on a dense new condition that deserves a careful look.","tokens_in":13543,"tokens_out":2502,"would_cite":true,"duration_ms":23044,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46E30","26D10"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a class of kernel-type integral operators on $(0,\\infty)$, boundedness between rearrangement-invariant spaces is equivalent to boundedness on non-increasing functions, with a universal constant $4$ (or $2^{m+1}$ for iterated kernels).","keywords":["reduction principle","Hardy-Littlewood inequality","kernel-type operators","rearrangement-invariant spaces","non-increasing rearrangement","down-associate norm","Lorentz spaces"],"falsifier":"A concrete falsifying calculation would be to compute the ratio of the optimal constants in the full inequality (1.1) and the restricted monotone-cone inequality (1.2) for, say, $I(t)=t^\\alpha$ with $\\alpha\\ge 1$ and $X=Y=L^p$; the theorem predicts the ratio is at most $4$ for $m=1$, so any ratio exceeding $4$ would settle the claim as false. For $m>1$, the corresponding test is a weight violating the essential-decreasing condition where restricted boundedness holds but full boundedness fails.","tokens_in":12641,"feed_emoji":"📉","tokens_out":9835,"duration_ms":84951,"temperature":0.7,"pith_summary":"This paper establishes a reduction principle for a class of kernel-type integral operators on the half-line: an operator of the form $f\\mapsto \\int_t^\\infty f(s)/I(s)\\,ds$, with $I$ non-decreasing, is bounded between two rearrangement-invariant Banach function spaces if and only if it is bounded when tested on non-increasing non-negative functions only. The restricted constant controls the full constant with a factor of $4$, and for $m$-fold iterated versions the factor is $2^{m+1}$. This matters because the monotone cone is far easier to test, and because the result works on the infinite measure space $(0,\\infty)$, covering situations where the classical Hardy-Littlewood rearrangement inequality for integrations away from zero fails. The paper thereby extends a previously known finite-interval result to settings relevant for operators such as Laplace transforms and potential-type operators.","feed_headline":"Kernel-operator bounds reduce to checking decreasing functions","feed_subtitle":"Full boundedness follows from the monotone case with a universal factor 4, for a large class of kernels.","key_machinery":"The argument is carried by the mutually associate pair of operators $R_I^m f(t)=\\frac{1}{(m-1)!}\\frac{1}{I(t)}\\int_0^t f(s)\\left(\\int_s^t \\frac{1}{I(r)}\\,dr\\right)^{m-1}ds$ and $H_I^m f(t)=\\frac{1}{(m-1)!}\\int_t^\\infty \\frac{f(s)}{I(s)}\\left(\\int_s^t \\frac{1}{I(r)}\\,dr\\right)^{m-1}ds$, together with the upper envelope $G_I^m f(t)=\\sup_{s\\ge t} R_I^m f^*(s)$, where $f^*$ is the non-increasing rearrangement. The decisive Lemma 3.8 shows that for functions with finite-measure support and weights satisfying the essential-decreasing condition, the set $E=\\{t:R_I^m f^*(t)<G_I^m f(t)\\}$ is a disjoint union of bounded intervals on which $G_I^m f$ is constantly equal to $R_I^m f^*(d_k)$. Replacing $g^*$ on each gap by its average defines an averaging operator that is norm-decreasing on rearrangement-invariant spaces; combining this with the dyadic estimate $(d-c)R_I^m f^*(d)\\le 2^{m+1}\\int_c^d R_I^m f^*(t)\\,dt$ yields the key inequality $\\|G_I^m f\\|_{X'}\\le 2^{m+1}\\|R_I^m f^*\\|_{X'_d}$, where $\\|\\cdot\\|_{X'_d}$ is the down-associate norm, the supremum of $\\int |f|g^*$ over the unit ball of $X$. The equivalence for the $H$-operator then follows from duality and mutual associativity.","core_discovery":"The central claim is Theorem 1.1: for every non-decreasing $I:(0,\\infty)\\to(0,\\infty)$ and rearrangement-invariant Banach function norms $X,Y$ on $(0,\\infty)$, the inequality $\\|\\int_t^\\infty f(s)/I(s)\\,ds\\|_Y \\le C\\|f\\|_X$ holds for all non-negative $f$ if and only if it holds for all non-increasing non-negative $f$, and the best restricted constant $C'$ yields the full inequality with $C=4C'$. The same proof gives Theorem 3.10 for the iterated kernels $H_I^m$ with kernel $\\frac{1}{(m-1)!}\\frac{1}{I(s)}\\left(\\int_s^t \\frac{1}{I(r)}\\,dr\\right)^{m-1}$, under an 'essentially decreasing in $t$' condition on the kernel ratio, with constant $2^{m+1}C'$. The discovery is that the failure of the Hardy-Littlewood rearrangement inequality away from zero does not destroy the reduction principle for these kernels; the monotone-cone inequality is enough to recover full boundedness with an explicit, norm-independent loss.","pith_inferences":["A natural testable conjecture is that the 'essentially decreasing' condition is also necessary for the reduction to hold for $m>1$; the paper proves sufficiency only and gives no counterexample when the condition fails.","For concrete choices $X=L^p$, $Y=L^q$, the monotone-cone inequality is often equivalent to a simple one-dimensional integral condition on $I$; combined with the theorem this should yield explicit criteria for the full operator, which could be checked numerically.","The constant $4$ for $m=1$ is likely not sharp; sharpening it would require replacing the averaging/truncation step in Lemma 3.8 or tracking the dyadic factor in the estimate it uses."],"forward_implications":["The boundedness of $H_I$ (and its iterates) on rearrangement-invariant spaces is completely determined by testing decreasing inputs; no additional condition on the non-decreasing $I$ is needed for $m=1$.","If the restricted inequality is verified with constant $C'$, the full inequality automatically holds with at most $4C'$ (or $2^{m+1}C'$ for $H_I^m$), so quantitative stability is built into the reduction.","The result applies on the infinite measure space $(0,\\infty)$, not only on finite intervals, so it is usable for operators studied over the whole Euclidean space, such as Riesz potentials and fractional maximal operators.","For $m>1$, the extra 'essentially decreasing' condition is satisfied, for example, by power weights $I(t)=t^\\alpha$ with $\\alpha\\ge 1$, giving a concrete family of higher-order kernels with the reduction property."],"supporting_citations":[{"why":"Supplies the background theory of rearrangements, rearrangement-invariant Banach function spaces, the Hardy-Littlewood inequality, resonance, and the averaging-operator theorem used in the main estimate.","marker":"[1]"},{"why":"Is the finite-interval reduction principle whose methods (Lemma 9.1, Proposition 9.3, Theorem 9.5) this paper adapts and extends to infinite measure spaces.","marker":"[3]"},{"why":"Provides the characterisation of boundedness of classical operators on Lorentz spaces used to exhibit the down-associate norms that illustrate the mechanism.","marker":"[4]"},{"why":"Summarizes the Lorentz-space embedding results used to compute the down-associate norm examples for $L^p$ spaces.","marker":"[2]"}],"fun_headline_variants":["Monotone test suffices for kernel operator bounds","Decreasing functions decide kernel operator norms","Restricted inequality implies full kernel operator bound","Monotone check gives full kernel bound","Kernel bound follows from monotone case only"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise for the headline result is simply that the weight $I$ is non-decreasing, since the proof uses this monotonicity to control the rearranged operator on dyadic intervals and to force the exceptional set $E$ into disjoint bounded intervals; for the higher-order version, the comparable premise is the 'essentially decreasing in $t$' condition on the kernel ratio, without which the interval-structure lemma no longer goes through.","fun_headline_variants_meta":{"raw":{"variants":["Monotone test suffices for kernel operator bounds","Decreasing functions decide kernel operator norms","Restricted inequality implies full kernel operator bound","Monotone check gives full kernel bound","Kernel bound follows from monotone case only"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000901,"raw_usage":{"total_tokens":3854,"prompt_tokens":896,"completion_tokens":2958,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":512,"completion_tokens_details":{"reasoning_tokens":2890}},"tokens_in":512,"tokens_out":2958,"duration_ms":18437,"temperature":1.0,"reasoning_tokens":2890,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:50:08.693430+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete falsifying calculation would be to compute the ratio of the optimal constants in the full inequality (1.1) and the restricted monotone-cone inequality (1.2) for, say, $I(t)=t^\\alpha$ with $\\alpha\\ge 1$ and $X=Y=L^p$; the theorem predicts the ratio is at most $4$ for $m=1$, so any ratio exceeding $4$ would settle the claim as false. For $m>1$, the corresponding test is a weight violating the essential-decreasing condition where restricted boundedness holds but full boundedness fails.","supporting_citations":[{"cited_title":"Bennett and R","cited_arxiv_id":null,"evidence_quote":"Supplies the background theory of rearrangements, rearrangement-invariant Banach function spaces, the Hardy-Littlewood inequality, resonance, and the averaging-operator theorem used in the main estimate."},{"cited_title":"Cianchi, L","cited_arxiv_id":null,"evidence_quote":"Is the finite-interval reduction principle whose methods (Lemma 9.1, Proposition 9.3, Theorem 9.5) this paper adapts and extends to infinite measure spaces."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the characterisation of boundedness of classical operators on Lorentz spaces used to exhibit the down-associate norms that illustrate the mechanism."},{"cited_title":"Carro, L","cited_arxiv_id":null,"evidence_quote":"Summarizes the Lorentz-space embedding results used to compute the down-associate norm examples for $L^p$ spaces."}],"review_version":1}