{"id":"a24c4e8e-4665-4201-a846-0ca459eb0d85","arxiv_id":"2508.13913","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Boundedness of (I+L)^{-beta/2} exp(i tau L^{gamma/2}) and L^{i tau} on weak Hardy spaces WH_{X,L}(X) with growth (1+|tau|)^{n(1/s0-r/2)} under Davies-Gaffney or Gaussian kernel conditions.","lead":"This paper proves growth estimates for Schrödinger groups and imaginary power operators acting on weak Hardy spaces built from a general class of Banach function spaces on metric measure spaces. It extends known sharp estimates to a broad framework that includes weighted, variable-exponent, Orlicz and mixed-norm spaces.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.8's stated β-range includes the endpoint, but the proof requires r>1 there; the k-sum in the I2 estimate diverges, so the endpoint estimate is unsupported.","rationale":"The reader's verdict is already CONDITIONAL, and its rationale mentions the endpoint proof gap. However, the reader's stated weakest_assumption is the use of the concurrent atomic decomposition [29]. My stress-test identifies a different, more immediately load-bearing concern: the proof of Theorem 1.8 fails precisely at the endpoint of the stated β-range, because the r ∈ (0,1) needed for the high-i weak-type estimate cannot satisfy r > 1. This is not a question of external dependence; it is visible inside the submitted argument. The same mechanism appears at the boundary of Theorem 1.15 but is harmless there because the hypothesis is strict. The reliance on [29] is also significant, but it is a citation to a published concurrent paper and could be verified externally; the endpoint divergence is a concrete correctness gap in the present text. For these reasons I agree with the CONDITIONAL verdict: the main ideas are plausible for the generic case, but the manuscript should be revised to either prove the endpoint case or state the theorem with a strict inequality in β. No change to the reader's overall verdict is needed.","tokens_in":40976,"tokens_out":6764,"duration_ms":70363,"concrete_test":"Recompute the k-summation in the I2 estimate of Theorem 1.8 with β set exactly to γn(1/s0 − 1/2) and any r ∈ (0,1). The relevant factor is 2^{−k s0[βr/γ − (1/s0 − r/2)n]} = 2^{k s0 n(1−r)}, which grows with k; hence the proof's bound diverges. This analytically settles that the submitted proof does not cover the endpoint. To decide whether the endpoint estimate itself is true, either supply a direct endpoint argument (e.g., with r = 1 plus additional i-decay), or restrict Theorem 1.8 to β > γn(1/s0 − 1/2).","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.8 asserts the Schrödinger-group bound for all β ≥ γn(1/s0 − 1/2). In the proof, the high-level term I2 is handled by choosing r ∈ (0,1) such that β > γn(1/(r s0) − 1/2) (see the line after (3.16)). This is possible only when β > γn(1/s0 − 1/2); at the endpoint β = γn(1/s0 − 1/2), the required inequality becomes r > 1. The obstruction is explicit in the k-summation after (3.17): the factor is 2^{−k s0[βr/γ − (1/s0 − r/2)n]}. At the endpoint this equals 2^{k s0 n(1−r)}, which is exponentially growing in k, so the sum over k ∈ Z_+ diverges and the bound (3.18) cannot be obtained by the given argument. The same calculation shows that the proof of Theorem 1.15 would fail at the boundary α = n(1/s0 − 1/2), though there the stated strict inequality α > ... avoids the issue. Because the endpoint β is explicitly included in Theorem 1.8, the abstract, and the derived strong-Hardy results, the proof as written does not establish the full stated claim. Separately, the proof begins from the atomic decomposition imported from [29, Theorems 3.4 and 3.5] via (3.2)–(3.3); if [29] has unstated restrictions beyond Assumptions 1.3 and 1.4, the gap would be wider. But the endpoint failure is internal to this manuscript and is the most load-bearing concern.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops weak Hardy spaces WH_{X,L} associated with a ball quasi-Banach function space X and a nonnegative self-adjoint operator L satisfying Davies-Gaffney estimates.  Theorem 1.6 states atomic and molecular decompositions for these spaces, but the proof is not included; the authors refer to the concurrent paper [29].  The main results are Theorem 1.8, a sharp time-growth estimate for the Schrödinger group (I+L)^{-β/2}e^{iτL^{γ/2}} on WH_{X,L}, and Theorem 1.15, an analogous bound for imaginary powers L^{iτ} under Gaussian kernel assumptions.  Section 5 applies these results to Orlicz, variable Lebesgue, weighted Lebesgue, and mixed-norm Lebesgue spaces.  The proofs combine the imported atomic decomposition with functional calculus estimates from [4] and [3] and with norm estimates (3.6)-(3.7) and (4.6)-(4.7).","tokens_in":41381,"tokens_out":9671,"duration_ms":104398,"significance":"If the stated estimates are fully established, the paper would provide a substantial extension of Schrödinger-group and imaginary-power estimates from the known L^p/classical Hardy-space setting to weak Hardy spaces built on very general ball quasi-Banach function spaces.  The framework is broad and the applications to weighted, mixed-norm, Orlicz, and variable Lebesgue spaces are potentially useful.  The paper is written in a clear style, and, conditional on the quoted atomic decomposition, the local estimates (3.6)-(3.7) and (4.6)-(4.7) are presented in detail.  No fitted parameters or circular argument appear.  However, the main theorems as stated are not fully supported by the proofs: the parameter r in Theorem 1.8 and the endpoint cases in both main theorems are not covered by the estimates that are actually derived.","major_comments":[{"comment":"The theorem allows β ≥ γn(1/s0 − 1/2) and every r∈(0,1].  In estimating I2 the proof asserts: “Since β>γn(1/s0 − 1/2), it follows that there exists r∈(0,1) such that β>γn(1/(r s0) − 1/2).”  The subsequent k-summation requires r > n/s0 / (β/γ + n/2).  At the endpoint β = γn(1/s0 − 1/2) this threshold equals 1, so no such r<1 exists and the k-sum diverges.  Even when β is strictly above the endpoint, the proof produces the estimate (3.18) for the auxiliary r chosen by the author, not for the arbitrary r appearing in the theorem statement; in particular r=1 is excluded by construction.  Thus inequality (1.5) is not proved for the full parameter range stated in the abstract and in Theorem 1.8.","section":"§3, proof of Theorem 1.8, Eqs. (3.17)-(3.18)"},{"comment":"The same issue occurs in the II2 estimate.  The proof chooses r∈(n/s0/(α+n/2),1), while the theorem states r∈[n/s0/(α+n/2),1].  At the lower endpoint the k-exponent s0[rα − (1/s0 − r/2)n] vanishes, so the sum over k diverges and the stated bound (1.9) is not obtained.  The endpoint r=1 is also not covered by the open interval used in the proof.  The theorem’s parameter range therefore needs to be narrowed, or a separate argument for the missing endpoints must be supplied.","section":"§4, proof of Theorem 1.15, Eqs. (4.13)-(4.14)"},{"comment":"These strong-Hardy-space results are stated without proof, with the remark that the proofs are similar.  The exponent in Theorem 1.9 is n(1/s0 − 1/2), which corresponds exactly to the r=1 endpoint of Theorem 1.8.  If Theorem 1.9 is intended to follow from Theorem 1.8 with r=1, then the missing endpoint argument for Theorem 1.8 must be provided.  If it is proved by a separate route, that route should be written out, especially because the endpoint β=γn(1/s0−1/2) is included in the statement.","section":"Theorems 1.9 and 1.10"},{"comment":"The atomic and molecular decomposition is not proved in this paper; the reader is referred to [29, Theorems 3.4 and 3.5].  This is bibliographically acceptable provided the hypotheses match exactly.  Since both (3.3) and (4.3) are imported from that result, the authors should explicitly state that Assumptions 1.3 and 1.4 (with the same range of s0 and q0) are precisely the assumptions of [29, Theorems 3.4 and 3.5], with no additional restriction on the ball quasi-Banach space X.  Any unstated condition in [29] would propagate to Theorems 1.8 and 1.15.","section":"Theorem 1.6 and Remark 1.7"}],"minor_comments":[{"comment":"The exponent n(1/s0 − r/2) contains s0, but s0 is not introduced in the statements of these theorems.  The statements should either assume s0 explicitly or quantify it as “there exists s0∈(0,min{1,p−}) such that…”.","section":"Section 5, Theorems 5.4-5.7 and 5.11-5.14"},{"comment":"“the some way” should be “the same way”.","section":"Definition 2.9"},{"comment":"The final sentence says “This completes the proof of Theorem 5.10”; it should refer to Theorem 5.16.","section":"Proof of Theorem 5.16"},{"comment":"“supporeted” should be “supported”.","section":"Acknowledgements"},{"comment":"The symbol r is used both for the fixed parameter in Theorem 1.8 and for the auxiliary exponent chosen in the proof of I2.  This makes the argument hard to follow and contributes to the confusion about which r appears in the final estimate.  Please use a different symbol, e.g. u, for the auxiliary exponent.","section":"Notation in §3"}],"recommendation":"major_revision","confidential_remarks":"The overlap with [29] is acknowledged by the authors and appears to be a standard concurrent-work situation.  The main concern is not the reliance on [29] but the fact that the central parameter ranges in Theorems 1.8 and 1.15 are not covered by the proofs as written.  This is fixable either by narrowing the statements or by supplying the missing endpoint arguments, but it must be addressed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the take: the paper proves Schrödinger group and imaginary power estimates on weak Hardy spaces WH_{X,L} for ball quasi-Banach spaces. That's genuinely new. But the proof of Theorem 1.8 has a real endpoint gap, and the paper outsources its atomic decomposition to a concurrent paper [29], so the current manuscript isn't self-contained.\n\nThe new content is the derivation of the estimates on the weak spaces. The setting is general — doubling spaces, Davies-Gaffney operators — and the results cover Orlicz, variable, weighted, and mixed-norm Lebesgue spaces. The technical machinery is handled carefully, and the paper gives credit where it's due. The proof is mostly a clean combination of known functional calculus estimates from [4] and [3] with the atomic decomposition. For β strictly greater than γn(1/s0 − 1/2), the argument works.\n\nThe soft spot: the theorem statement includes the endpoint β = γn(1/s0 − 1/2). The proof at (3.16)–(3.18) chooses r ∈ (0,1) so that β > γn(1/(r s0) − 1/2). At the endpoint that is impossible, and the k-sum then has the factor 2^{−k s0[βr/γ − (1/s0 − r/2)n]}, which becomes 2^{kn(1−r)} — exponentially divergent. So the argument simply does not reach the endpoint, even though the abstract and the strong-Hardy corollaries rely on it. Theorem 1.15 keeps a strict inequality, so that one is fine. The other soft spot is that the atomic/molecular characterization, Theorem 1.6, is not proved in this paper; the authors explicitly send you to [29]. If [29] has restrictions beyond Assumptions 1.3 and 1.4, the gap widens. The paper is honest about this, but the decisive decomposition is not part of this submission.\n\nAll in all, this is a competent paper that is very likely correct for β above the endpoint. The endpoint might be true, but the proof doesn't establish it. I'd send it to a serious referee, but the authors should either prove the endpoint or restrict the statement. Not a desk reject — a revise-and-resubmit.","headline":"New Schrödinger-group estimates on weak Hardy spaces for ball quasi-Banach spaces, but the endpoint of Theorem 1.8 is not proved.","tokens_in":41858,"tokens_out":3606,"would_cite":true,"duration_ms":36900,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B35","42B30","42B25"],"pacs":[],"model":"deepseek-v4-flash","headline":"Schrödinger groups and imaginary power operators are bounded, with explicit polynomial growth in time, on weak Hardy spaces modeled on ball quasi-Banach function spaces.","keywords":["Weak Hardy spaces","Ball quasi-Banach function spaces","Non-negative self-adjoint operators","Davies-Gaffney estimates","Schrödinger groups","Imaginary power operators","Atomic decompositions","Spaces of homogeneous type"],"falsifier":"The decisive test is estimate (3.6)/(4.6): pick L=−Δ on R^n, take X=L^1(R^n) with an atom a supported in the unit ball, and compute the L^2 norm of S_L((I+L)^{-β/2}e^{iτL^{γ/2}}a) on the annulus U_k((1+|τ|)B). If the decay in k is slower than 2^{-kβ/γ}, the local estimate fails and the polynomial bound on WH_{1,−Δ} would be false. Simpler: for L=−Δ and X=L^1, test directly whether the weak Hardy norm of the flow can grow faster than (1+|τ|)^{n/2} for a well-chosen f; the theorem predicts it cannot.","tokens_in":40863,"feed_emoji":"📐","tokens_out":6757,"duration_ms":66887,"temperature":0.7,"pith_summary":"This paper proves time-growth estimates for Schrödinger flows associated with a broad class of differential operators. On a doubling metric measure space, for any non-negative self-adjoint operator L whose heat semigroup satisfies the Davies-Gaffney off-diagonal decay, it introduces a weak Hardy space WH_{X,L} built from a ball quasi-Banach function space X — a norm on measurable functions that only needs ball indicators to be finite. It shows that both the regularized Schrödinger group (I+L)^{-β/2}e^{iτL^{γ/2}} and, under stronger Gaussian kernel assumptions, the imaginary power operator L^{iτ} map WH_{X,L} to itself with bounds growing like (1+|τ|)^{n(1/s0−r/2)}. The same estimates hold for the strong Hardy space by embedding, and the framework covers weighted Lebesgue, Orlicz, variable Lebesgue, and mixed-norm Lebesgue spaces, where the results are claimed to be new even in Euclidean settings. The starting point is an atomic/molecular characterization of WH_{X,L}, whose proof the authors defer to an independent companion result.","feed_headline":"Schrödinger flows get polynomial bounds on weak Hardy spaces","feed_subtitle":"New atomic estimates give (1+|τ|) growth for fractional-power semigroups, across weighted, Orlicz and mixed-norm spaces.","key_machinery":"The load-bearing object is the atomic decomposition of WH_{X,L} (Theorem 1.6, proved in a companion paper): every f in the space is a sum of (X,M)L-atoms a_{i,j} supported, up to L^M smoothing, on balls B_{i,j}, with level-dependent coefficients λ_{i,j}=2^i||1_{B_{i,j}}||_X. The norm is controlled by sup_i ||(Σ_j [λ_{i,j}1_{B_{i,j}}/||1_{B_{i,j}}||_X]^{s0})^{1/s0}||_X. The operator is applied via the spectral functional calculus plus the identity I=(I−e^{-r_B^2 L})^M+P(r_B^2 L), splitting each transformed atom into a part with good off-diagonal decay and a part supported near the original ball. The main local estimates (3.6)–(3.7) and (4.6)–(4.7) give L2 decay 2^{-kβ/γ} or 2^{-kα} on the k-t","core_discovery":"The central claim is that, for 0<γ≠1, β≥γn(1/s0−1/2), r∈(0,1], and τ∈R, every f∈WH_{X,L} satisfies ||(I+L)^{-β/2}e^{iτL^{γ/2}}f||_{WH_{X,L}} ≤ C(1+|τ|)^{n(1/s0−r/2)}||f||_{WH_{X,L}}; and, if the space is Ahlfors n-regular, L has Gaussian heat-kernel upper bounds and the spectral multiplier kernels satisfy condition (1.8), then for α>n(1/s0−1/2), r∈(n/s0/(α+n/2),1], ||L^{iτ}f||_{WH_{X,L}} ≤ C(1+|τ|)^{n(1/s0−r/2)}||f||_{WH_{X,L}}. Here s0 is the exponent controlling the boundedness of the Hardy–Littlewood maximal operator on a convexified associate space. The proof reduces the operator action to atoms localized in balls B, estimates the Lusin area function of the transformed atom on annuli of","pith_inferences":["We read the proof as suggesting the same machinery works for any spectral multiplier whose derivatives obey the Hörmander-type conditions used in (1.8); the paper only states the two applications and does not draw this multiplier-theorem consequence.","The exponent n(1/s0−r/2) is not flagged by the authors as sharp; a natural test is to compare it with known sharp exponents for L=−Δ on R^n and X=L^r, which could indicate whether the r-choice is genuine or an artifact of the proof.","Because Theorem 1.6 is imported from [29], a prudent check is to verify that the ball quasi-Banach spaces in Section 5 meet the hypotheses of [29] exactly as stated; if any of those spaces requires an extra separability or Fatou condition, some of the advertised applications would need adjustment."],"forward_implications":["Strong Hardy spaces H_{X,L} inherit the same polynomial growth because H_{X,L} embeds continuously into WH_{X,L}; this is why the results are stated as new for strong Hardy spaces too.","The Riesz means I_{s,t}(L) associated with L^{γ/2} are uniformly bounded on WH_{X,L} for s≥γn(1/s0−1/2), by the same argument used in Corollary 1.12.","Taking X=L^r on a doubling space recovers and extends earlier Schrödinger-group estimates for r∈(0,1] under the Davies-Gaffney assumption, which is weaker than Gaussian upper bounds.","Instantiating X with weighted Lebesgue, Orlicz, variable Lebesgue, or mixed-norm Lebesgue spaces yields four families of new theorems on those spaces, including Euclidean R^n and Hermite operators."],"supporting_citations":[{"why":"Supplies the atomic and molecular characterization of WH_{X,L} that the proofs of Theorems 1.8 and 1.15 start from.","marker":"[29]"},{"why":"Provides the sharp Schrödinger-group estimates on Hardy spaces H^p_L and the functional-calculus off-diagonal estimates recycled in (3.6)–(3.7).","marker":"[4]"},{"why":"Gives the imaginary-power-operator theorem on Hardy spaces whose strategy is adapted in the proof of Theorem 1.15.","marker":"[3]"},{"why":"Introduces Hardy spaces associated with BQBF spaces and operators with Gaussian estimates, providing the baseline for the strong-space results and maximal-function characterizations.","marker":"[26]"},{"why":"Develops Hardy spaces H_{X,L} on doubling spaces with Davies-Gaffney or Gaussian estimates, including spectral-multiplier and Schrödinger-group results that the present paper extends to weak spaces.","marker":"[28]"},{"why":"Establishes weak Hardy spaces for BQBF spaces, including the weak-type Fefferman–Stein inequality and the continuous embedding H_{X,L}⊂WH_{X,L} used in Theorem 1.9.","marker":"[34]"}],"fun_headline_variants":["Polynomial bounds for Schrödinger flows on weak Hardy spaces","Imaginary power operators get polynomial bounds on weak Hardy spaces","Atomic decompositions tame Schrödinger groups on weak Hardy spaces","Polynomial growth for imaginary powers on weak Hardy spaces","New polynomial estimates for Schrödinger flows on weak Hardy spaces"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The estimates rest on the atomic and molecular characterization of WH_{X,L} stated as Theorem 1.6, whose proof is not given here but is referred to an independent companion result; any hidden extra condition on the ball quasi-Banach function space in that characterization would invalidate the main theorems.","fun_headline_variants_meta":{"raw":{"variants":["Polynomial bounds for Schrödinger flows on weak Hardy spaces","Imaginary power operators get polynomial bounds on weak Hardy spaces","Atomic decompositions tame Schrödinger groups on weak Hardy spaces","Polynomial growth for imaginary powers on weak Hardy spaces","New polynomial estimates for Schrödinger flows on weak Hardy spaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000537,"raw_usage":{"total_tokens":2651,"prompt_tokens":1217,"completion_tokens":1434,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":961,"completion_tokens_details":{"reasoning_tokens":1353}},"tokens_in":961,"tokens_out":1434,"duration_ms":10372,"temperature":1.0,"reasoning_tokens":1353,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T18:50:49.522376+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The decisive test is estimate (3.6)/(4.6): pick L=−Δ on R^n, take X=L^1(R^n) with an atom a supported in the unit ball, and compute the L^2 norm of S_L((I+L)^{-β/2}e^{iτL^{γ/2}}a) on the annulus U_k((1+|τ|)B). If the decay in k is slower than 2^{-kβ/γ}, the local estimate fails and the polynomial bound on WH_{1,−Δ} would be false. Simpler: for L=−Δ and X=L^1, test directly whether the weak Hardy norm of the flow can grow faster than (1+|τ|)^{n/2} for a well-chosen f; the theorem predicts it cannot.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the atomic and molecular characterization of WH_{X,L} that the proofs of Theorems 1.8 and 1.15 start from."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the sharp Schrödinger-group estimates on Hardy spaces H^p_L and the functional-calculus off-diagonal estimates recycled in (3.6)–(3.7)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the imaginary-power-operator theorem on Hardy spaces whose strategy is adapted in the proof of Theorem 1.15."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces Hardy spaces associated with BQBF spaces and operators with Gaussian estimates, providing the baseline for the strong-space results and maximal-function characterizations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Develops Hardy spaces H_{X,L} on doubling spaces with Davies-Gaffney or Gaussian estimates, including spectral-multiplier and Schrödinger-group results that the present paper extends to weak spaces."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes weak Hardy spaces for BQBF spaces, including the weak-type Fefferman–Stein inequality and the continuous embedding H_{X,L}⊂WH_{X,L} used in Theorem 1.9."}],"review_version":1}