{"id":"0e34f15e-2085-4435-9d76-273bd43e7606","arxiv_id":"1908.08831","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Spherical Fourier multipliers satisfying a product Marcinkiewicz condition are Lp-bounded on products of rank-one noncompact symmetric spaces, for p not equal to 2.","lead":"On products of two rank-one hyperbolic-type symmetric spaces, the authors prove a Marcinkiewicz-style multiplier theorem for the spherical Fourier transform: frequency multipliers with product-type derivative bounds yield bounded operators on Lp. A generalist might read it because it fills a specific gap in harmonic analysis where earlier Hörmander-type theorems failed at the coordinate walls.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Corollary 3.4(ii) is the load-bearing transference step; its proof is omitted and its hypotheses (non-semisimple A1×G2, K2-bi-invariant extension, norm preservation) are unchecked.","rationale":"The paper is a serious and mostly careful proof of a new multiplier theorem. I found no outright contradiction in the analytic estimates, and the broad strategy (local expansion, wall terms, infinity terms) is coherent. The reader's verdict of CONDITIONAL is appropriate. The single most load-bearing weak point is Corollary 3.4(ii), exactly as the reader identified. It is used repeatedly for the B1 and B2 estimates, and it is dispatched with 'straightforward to check' and a citation. The non-semisimple group A1×G2 is not obviously covered by the standard Coifman-Weiss transference theorem, the K2-bi-invariant extension needs a well-definedness argument, and the norm inequality is not proven. Because the rest of the argument depends on this reduction, the central claim is not fully verified until this corollary is supplied. The paper itself flags several omitted details, but none is as load-bearing as this one. The m≥3 extension is also omitted, but it is not part of Theorem 1.1's main two-factor statement. I therefore agree with the reader's weakest assumption and see no reason to change the conditional verdict.","tokens_in":51862,"tokens_out":21884,"duration_ms":221775,"concrete_test":"Write out the missing proof of Corollary 3.4(ii) by applying [CW, Theorem 8.7] to G=A1×G2, K={e1}×K2, A=A1×A2: define the K-bi-invariant extension F explicitly, verify its well-definedness from the Weyl invariance of κ_{v1}, and identify the exact constant and the factor (δ2 vs D2^{1/p}) in the measure change. As a minimal computational check, take G1=G2=SL(2,R), p=3/2, and κ_{v1}(a1,a2)=f(a1)g(a2) with smooth compactly supported f,g; compute both sides using the standard Herz majorization formula for K-bi-invariant convolutors on SL(2,R) and the classical Marcinkiewicz theorem on R2. If the inequality fails, or requires an additional spherical-function factor, the corollary as stated is false and the proof of Theorem 1.1 must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Corollary 3.4(ii) is the pivotal reduction that takes Cv^p bounds from A1×A2 to A1×G2; it is used in Proposition 6.2(iii) (estimate (6.12)) and in Proposition 7.4(iii) to control the B1 and B2 terms. The proof is one sentence: 'straightforward to check' plus a citation to [CW, Theorem 8.7]. The statement requires three things that are never shown: (a) the direct product A1×G2, with A1≅R, satisfies the exact hypotheses of the quoted transference theorem (it is not semisimple, and the 'Cartan decomposition' ({e1}×K2)(A1×A2)({e1}×K2) must be checked against the theorem's formulation); (b) the {e1}×K2-bi-invariant extension of κ_{v1} from A1×A2 to A1×G2 is well-defined, which requires κ_{v1}(a1,·) to be constant on K2 double cosets in A2 (Weyl invariance), and this is not verified; (c) the extension is norm-preserving, i.e., ||F||_{Cv^p(A1×G2)} ≤ C||f||_{Cv^p(A1×A2)} with C independent of v1. If (a)-(c) fail, the estimates for κ1A2 and κ11 collapse, so Theorem 1.1 has no foundation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a Marcinkiewicz-type multiplier theorem for the spherical Fourier transform on a product X1 times X2 of two rank-one noncompact symmetric spaces. Under separate Marcinkiewicz-type conditions in each variable, with derivative orders N1 > (n1+3)/2 and N2 > (n2+3)/2, any G-invariant operator with such a multiplier is shown to be bounded on Lp(X1 times X2) for p in (1, infinity) other than 2. The proof splits the kernel into a local part B0, a wall part B1, and an infinity part B2, applying local and Harish-Chandra expansions of spherical functions and reducing the estimates to convolutor bounds on abelian groups and semidirect products, with a transference principle from Coifman-Weiss and a previous paper by two of the authors used at a key reduction step.","tokens_in":52240,"tokens_out":9409,"duration_ms":87588,"significance":"If the proof is completed, the result is a substantial extension of Marcinkiewicz multiplier theory to products of rank-one noncompact symmetric spaces, covering operators such as products of imaginary powers of Laplacians that are outside the scope of Hörmander-type theorems. The paper contains a useful comparison with Ionescu's higher-rank condition in Remark 4.2, showing that the two conditions are independent near the walls, and the B0/B1/B2 decomposition is natural and well organized. The analytic estimates in Propositions 5.1, 5.3, 6.2, and 7.4 are extensive and for the most part carefully written. However, the paper is not fully self-contained because a load-bearing transference step is asserted rather than proved, and several supporting lemmas have omitted proofs.","major_comments":[{"comment":"This corollary is the only step that moves Cv^p bounds from the abelian group A1 times A2 to the non-abelian group A1 times G2, and it is used directly at equation (6.12) in Proposition 6.2(iii) to control the tau^p,3 term for the kernel kappa_{1A2}. The proof is one sentence: the group A1 times G2 is said to admit the Cartan decomposition ({e1} times K2)(A1 times A2)({e1} times K2), and [CW, Theorem 8.7] is applied to the {e1} times K2-bi-invariant extension of kappa_{v1}. The manuscript never verifies that (a) the quoted transference theorem applies to the non-semisimple group A1 times G2, (b) the {e1} times K2-bi-invariant extension of kappa_{v1} is well-defined, which requires checking Weyl invariance on A2, or (c) the extension is norm-preserving with a constant independent of v1. Without (6.12), the estimate for kappa_{1A2} collapses and the boundedness of B1 is not established. Please provide a complete proof of Corollary 3.4(ii), or state precisely the transference theorem used and verify all of its hypotheses.","section":"Section 3.1, Corollary 3.4(ii)"},{"comment":"The estimate in Lemma 5.2(iii) is stated without proof ('The proof of (iii) follows the lines of the proof of (ii). We omit the details.'), but it is needed in Proposition 5.3(iii) for the case where n1 is even and n2 is odd. Specifically, it provides the Marcinkiewicz condition for the function d_{v1} H that is required to conclude kappa_{1,1} belongs to Cv^p(X). The mixed-parity case is not literally the same as (ii), because after the change of variables only one angular integration remains and the Weyl-invariance argument must be checked separately. The omitted details should be supplied.","section":"Section 5, Lemma 5.2(iii)"},{"comment":"The proof assumes that the multiplier mB is pre-multiplied by ~h_eps(lambda) = exp(-eps(lambda1^2 + lambda2^2)) and states that this is no loss of generality because the final bounds depend on ||mB||_{M(Tp;N)} and the norms of mB ~h_eps converge to that value. No proof of this convergence or of the passage to the limit as eps tends to 0 is given. Since Theorem 1.1 is stated for all multipliers in M(Tp;N) and not only for rapidly decreasing ones, this limiting argument must be supplied, or the theorem must be restricted to the regularized class and a separate argument given for the general case.","section":"Section 4, Outline after (4.8)"}],"minor_comments":[{"comment":"In both definitions the displayed norm for the 'at infinity' class is written with the same symbol as the non-infinity class: Eq. (4.2) should be ||m||_{M_infty(Tp;N)} and Eq. (4.7) should be ||m||_{M_infty(a*;N)}.","section":"Definition 4.1, Eq. (4.2); Definition 4.3, Eq. (4.7)"},{"comment":"The sentence 'We denote by 3 the two dimensional vector (3, 3)' is confusing; use a distinct multi-index notation, for example N0 = (3,3), or write the condition componentwise as N1 > (n1+3)/2 and N2 > (n2+3)/2.","section":"Section 4, Theorem restatement"},{"comment":"The symbol delta2 is used in Corollary 3.4(ii) but is not defined there; it should be defined explicitly (as the density in Cartan coordinates on A2, introduced in Section 2.3) at the point of first use.","section":"Section 3.1, Corollary 3.4"},{"comment":"The proof of Lemma 7.1(iv) is omitted with the remark that it is similar to (iii) with roles interchanged. Since the statement is used in Lemma 7.2 to prove that phi^p_{11} lies in Cv^q(A), a short proof should be included for completeness.","section":"Section 7, Lemma 7.1(iv)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the main result is significant if the proof is completed. The central risk is the unproved Coifman-Weiss transference step in Corollary 3.4(ii), which is the hinge of the B1 analysis; I would ask the authors to give a full proof rather than a citation. The reliance on [CMW], which involves two of the same authors, is not a problem per se because that is a published independent theorem, but the specific transference application needs to be checked carefully. The remaining omissions are technical and likely repairable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know about 1908.08831: it proves the first Marcinkiewicz-type multiplier theorem for products of two rank-one noncompact symmetric spaces, and the product condition (4.1) is new and not subsumed by Ionescu's earlier condition. The result is real and not in the literature. The proof follows Ionescu's transference strategy, splitting the kernel into B0, B1, B2; the estimates in Propositions 5.1, 5.3, 6.2 and 7.4 are detailed and mostly written out. The paper handles operators like L1^{it}(L1+L2)^{iu}L2^{iv} that escape Hormander-type theorems, and Remark 4.2 correctly shows (4.1) and (4.3) are independent. I found no internal contradiction.\n\nSoft spots: the paper has several 'details omitted' or 'straightforward to check' passages. The one that matters is Corollary 3.4(ii), the transference from A1xA2 to A1xG2 used in the B1 and B2 estimates. The proof is one sentence citing Coifman-Weiss. I checked the specific worries in the stress-test note: the K2-bi-invariance of kappa_{v1} does follow from kappa being K-bi-invariant, and A1xG2 does admit the Cartan decomposition ({e1}xK2)(A1xA2)({e1}xK2) because A1 is abelian and G2=K2A2K2. So the 'extension is well-defined' objection does not land. What is missing is an explicit verification that [CW, Theorem 8.7] covers this non-semisimple product group, and that the constant is independent of v1. That is a real gap in presentation, not a fatal one; it should be supplied before acceptance. Other omissions are smaller: parts of Lemma 5.2(iii), Lemma 7.1(iv), and the even/odd parity cases in Proposition 5.3 are left as 'similar'; the m>=3 extension is stated as 'straightforward though lengthy' with no proof. The epsilon-regularization is also asserted without proof; for the M-norm it is plausible but deserves a check.\n\nThe central claim holds up: I agree with the reader's conditional verdict, with MODERATE confidence because I cannot machine-check the analytic estimates and the omitted transference proof is load-bearing. This is a paper for harmonic analysts working on symmetric spaces and multiplier theorems. It deserves a serious referee. Recommendation: send to peer review; ask the authors to verify the transference hypotheses explicitly and either prove or demote the m>=3 statement.","headline":"First real Marcinkiewicz theorem for products of rank-one symmetric spaces; proof is solid but with a load-bearing transference step left as 'straightforward'.","tokens_in":52767,"tokens_out":8750,"would_cite":true,"duration_ms":82422,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["43A85","43A32"],"pacs":[],"model":"deepseek-v4-flash","headline":"A product Marcinkiewicz multiplier theorem holds on products of rank-one noncompact symmetric spaces.","keywords":["Marcinkiewicz multiplier","symmetric space","spherical Fourier transform","Lp boundedness","transference","rank one","product space","noncompact type"],"falsifier":"Compute, for a nontrivial $K_2$-bi-invariant kernel $\\kappa_{v_1}$ on $A_1\\times G_2$, both sides of the inequality in Corollary 3.4(ii), using the modular weight $D_1^{1/p}\\delta_2$ on $A_1\\times A_2$; if the left-hand $\\mathrm{Cv}^p(A_1\\times G_2)$ norm exceeds the right-hand $\\mathrm{Cv}^p(A_1\\times A_2)$ norm times the claimed constant, the transference step fails.","tokens_in":51664,"feed_emoji":"📐","tokens_out":8470,"duration_ms":80505,"temperature":0.7,"pith_summary":"The paper proves that a Marcinkiewicz-type condition on the spherical Fourier transform—one that controls mixed derivatives by products of the one-variable weights $\\Theta_p^{(1)}(\\lambda_1)\\Theta_p^{(2)}(\\lambda_2)$ on the tube $T_p^{(1)}\\times T_p^{(2)}$—is enough to make a $G$-invariant operator bounded on $L^p(X_1\\times X_2)$ for $p\\in(1,\\infty)\\setminus\\{2\\}$, provided the differentiation orders exceed $(n_j+3)/2$. This fills a gap: Euclidean-type multiplier theorems on the product cannot treat multipliers that are singular along the coordinate walls, such as those associated with operators mixing the two Laplace–Beltrami operators. The proof splits the convolution kernel into local, wall, and infinity pieces, then controls each piece by spherical-function expansions and by transference to the Iwasawa groups. If correct, the result provides the first Marcinkiewicz multiplier theorem for spherical transforms on noncompact symmetric spaces, and it extends to products of any finite number of rank-one factors.","feed_headline":"Product Marcinkiewicz multipliers are Lp-bounded on symmetric spaces","feed_subtitle":"The theorem covers multipliers singular along coordinate walls, beyond Euclidean-type results for products.","key_machinery":"The central object is the product Marcinkiewicz norm $\\|m\\|_{M(T_p;N)}$ on the tube $T_p=T_p^{(1)}\\times T_p^{(2)}$, defined by the weights $\\Theta_p^{(j)}(\\lambda_j)=\\min(|\\lambda_j-i\\delta(p)\\rho_j|,|\\lambda_j+i\\delta(p)\\rho_j|)$; this is the object that converts derivative estimates into $L^p$ boundedness. The carrying mechanism is a three-piece kernel decomposition: a local piece near the identity uses Bessel-function expansions, a wall piece uses one asymptotic expansion and one local expansion, and the infinity piece uses two asymptotic expansions. Each piece is then handled by a semidirect-product transference theorem that bounds convolution operators on $N_1A_1\\times G_2$ (or $N_1A_1\\times N_2A_2$) in terms of easier convolution norms on the abelian subgroups $A_1\\times A_2$.","core_discovery":"On the product $X=X_1\\times X_2$ of two rank-one noncompact symmetric spaces, the authors establish Theorem 1.1: for $p\\in(1,\\infty)\\setminus\\{2\\}$ and $N_j>(n_j+3)/2$, every $G$-invariant operator $B$ whose spherical multiplier $m_B$ is holomorphic on $T_p^{(1)}\\times T_p^{(2)}$ and satisfies $|\\partial_{\\lambda_1}^{j_1}\\partial_{\\lambda_2}^{j_2}m_B(\\lambda_1,\\lambda_2)|\\le C\\,\\Theta_p^{(1)}(\\lambda_1)^{-j_1}\\Theta_p^{(2)}(\\lambda_2)^{-j_2}$ extends to a bounded operator on $L^p(X)$, with operator norm controlled by the multiplier norm. The point is that the two derivative weights factor: each factor only sees the distance from the corresponding one-variable tube boundary, so the multiplier may be singular along an entire coordinate wall. The proof decomposes the kernel into $B_0+B_1+B_2$, representing contributions near the origin, near one wall, and away from both walls, and analyses each with the local and asymptotic expansions of spherical functions together with a transference theorem that moves estimates from solvable Iwasawa groups to the symmetric space.","pith_inferences":["A natural test is to extend the theorem to higher-rank symmetric spaces by using products of one-variable wall distances; the independence example suggests that the correct formulation should weight each root direction separately rather than by a single distance to the polyhedron.","The unproved transference step in Corollary 3.4(ii) should be checked before relying on the wall estimates; writing out the proof or finding a counterexample would settle whether the current route is valid.","The result implies that symmetric-space analogues of operators with multipliers such as $(|\\lambda_1|^2+|\\lambda_2|^2)^{iu}|\\lambda_2|^{iv}$ are $L^p$ bounded, which joint spectral multiplier theorems based on sectors do not deliver."],"forward_implications":["Joint spectral multipliers of the pair $(L_1,L_2)$ that satisfy the product Marcinkiewicz estimates are $L^p(X)$ bounded; in particular the theorem covers operators whose multipliers are singular on coordinate walls, a case outside Euclidean-type multiplier theorems on the product.","By interpolation with the trivial $L^2$ case, the $L^p$ bound extends to every $r$ with $|1/r-1/2|\\le|1/p-1/2|$.","The statement extends to products $X_1\\times\\cdots\\times X_m$ of $m\\ge3$ rank-one noncompact symmetric spaces; the authors indicate that the modifications are straightforward but omit them.","The comparison in Remark 4.2 shows that the product Marcinkiewicz condition is independent of the higher-rank condition used previously, so the theorem genuinely enlarges the class of admissible multipliers near the walls."],"supporting_citations":[{"why":"Provides the rank-one multiplier theorem and the kernel decomposition whose two-variable analogue drives Sections 6 and 7.","marker":"[I1]"},{"why":"Supplies the local and asymptotic expansions for spherical functions used to split the kernel into $B_0$, $B_1$ and $B_2$.","marker":"[ST]"},{"why":"Gives the semidirect-product transference principle cited as Theorem 3.3, which moves convolution bounds from the product group back to simpler subgroups.","marker":"[CMW]"},{"why":"Supplies the transference theorem invoked in Corollary 3.4(ii) to pass from estimates on $A_1\\times A_2$ to estimates on $A_1\\times G_2$.","marker":"[CW]"},{"why":"Establishes that spherical multipliers of $L^p$ operators continue analytically to the tube $T_p$, fixing the domain on which the Marcinkiewicz conditions are imposed.","marker":"[CS]"},{"why":"Provides the classical Euclidean multiplier theorem whose product analogue motivates the main result.","marker":"[Ho]"},{"why":"Gives the Figà-Talamanca–Herz algebra facts used to multiply convolution kernels by cutoff functions.","marker":"[Co]"}],"fun_headline_variants":["Lp-bounded multipliers on product symmetric spaces, singular along walls","Coordinate-wall singular multipliers on symmetric space products are Lp-bounded","Marcinkiewicz multipliers on product symmetric spaces: singular along coordinate walls","Product symmetric spaces: Lp multipliers with factor weight singularities","Lp-boundedness of product-space multipliers with wall singularities"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on an asserted transference step that moves norm estimates from the flat subgroup $A_1\\times A_2$ to the larger group $A_1\\times G_2$ without increasing them; the step is stated as straightforward but not proved, and the estimates for two of the three kernel pieces collapse if it is wrong.","fun_headline_variants_meta":{"raw":{"variants":["Lp-bounded multipliers on product symmetric spaces, singular along walls","Coordinate-wall singular multipliers on symmetric space products are Lp-bounded","Marcinkiewicz multipliers on product symmetric spaces: singular along coordinate walls","Product symmetric spaces: Lp multipliers with factor weight singularities","Lp-boundedness of product-space multipliers with wall singularities"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000749,"raw_usage":{"total_tokens":3270,"prompt_tokens":814,"completion_tokens":2456,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":430,"completion_tokens_details":{"reasoning_tokens":2366}},"tokens_in":430,"tokens_out":2456,"duration_ms":18211,"temperature":1.0,"reasoning_tokens":2366,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:28:46.256295+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a nontrivial $K_2$-bi-invariant kernel $\\kappa_{v_1}$ on $A_1\\times G_2$, both sides of the inequality in Corollary 3.4(ii), using the modular weight $D_1^{1/p}\\delta_2$ on $A_1\\times A_2$; if the left-hand $\\mathrm{Cv}^p(A_1\\times G_2)$ norm exceeds the right-hand $\\mathrm{Cv}^p(A_1\\times A_2)$ norm times the claimed constant, the transference step fails.","supporting_citations":[],"review_version":1}