{"id":"859db9c1-2c23-43f1-9215-f1e0920ddab4","arxiv_id":"2508.21269","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The distance from f to a subspace V of the inhomogeneous Lipschitz space Λ_s is equivalent to a critical index measuring where the fractional heat semigroup derivative ∂^r_t T_{α,t} f exceeds ε t^{s-rα}.","lead":"This paper proves that the distance from a function in a Lipschitz space to a large family of subspaces can be read off from how fast certain derivatives of a fractional heat flow decay. The result gives a semigroup-based measuring stick for approximation error, extending earlier work that only handled the Poisson semigroup or finite difference operators.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lower bound rests on Lemma 9.2 from unpublished companion [12]; if (9.3) fails, Proposition 6.3 and Theorems 3.5/3.12 collapse, so standalone verification is needed.","rationale":"The reader's weakest-assumption analysis correctly isolates Lemma 9.2 as the most load-bearing unverified ingredient. The estimate is quoted from an unpublished companion paper with no available arXiv number, and no proof is supplied in the present manuscript. Its role is not peripheral: Lemma 9.2 is the only bridge between the fractional-heat-semigroup bad sets D_{α,r,j} and the finite-difference sets S_{2ℓ,j}, and this bridge is needed for the lower bound in Theorem 3.5 and Theorem 3.12. Without it, the semigroup characterization of distances is not established, even if the theorems are true. I see no fatal mathematical flaw in the parts that are proved; the kernel estimates in Section 5, the hyperbolic-neighborhood arguments, and the lattice framework are coherent. The unpublished-reference issue is exactly the kind of missing support that warrants a conditional verdict: the claim should be accepted only once the companion lemma is supplied or independently verified. I considered whether the apparent mismatch between the regularity assumption L>αr+1 in Theorems 3.5/3.12 and the weaker L≥r−1 in the applications is more serious, but Remark 3.3 explicitly allows fixing a wavelet system with sufficiently large regularity, so the applications can be read as choosing L large enough to meet the abstract theorem's condition; this is a presentation issue rather than a collapse of the argument. The abstract's omission of the αr>s+3 condition and the ε^0 term is also real but secondary. I therefore agree with the reader's CONDITIONAL verdict and see no need to change it.","tokens_in":49068,"tokens_out":8624,"duration_ms":87499,"concrete_test":"Independently re-derive Lemma 9.1 (equivalently, the estimate (9.1)) without invoking Lemma 9.2's pointwise bound (9.3), using only the multiplier m_ℓ properties in Lemma 9.3 and the Fourier representation (9.6)–(9.9). If the derivation cannot be completed, or if it requires strengthening Lemma 9.3 beyond what is stated, obtain the companion manuscript [12] and verify Lemma 5.11 directly. A minimal check: test (9.3) for ℓ=1, n=1, and f(x)=|x|^s near x=0; if the right-hand side fails to bound the left-hand side as t→0, the lower-bound argument is invalid.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of the lower bound in both main theorems depends on Proposition 6.3, which says D_{α,r,j}(s,f,ε) ⊂ ⋃_{i=j+1}^{j+8ℓ} (S_{2ℓ,i}(s,f,cε))_R. Proposition 6.3 is deduced from Lemma 9.1, and the step from Lemma 9.1 to the inclusion uses (9.2). Lemma 9.1 in turn is proved only via Lemma 9.2, the ball-average estimate |f(x)-B_{ℓ,t}f(x)| ≤ C sup_{t/(8ℓ)≤y≤t/2} sup_{|x'-x|≤4ℓt} Δ^{2ℓ}f(x',y). This estimate is quoted from [12, Lemma 5.11], an unpublished companion paper whose arXiv identifier is given as '???'. The published source [10, Theorem 1] is cited only for the second part (9.4), the uniform Λ_s estimate, not for the pointwise estimate (9.3). If (9.3) requires extra hypotheses (e.g., higher regularity of f, a larger spatial neighborhood, or a different range of y) or is simply false, then Lemma 9.1 fails, Proposition 6.3 fails, and the chain leading to (6.4) and to the analogous lower bound in Theorem 3.12 collapses. The statement of the main theorem might still be true, but it is not supported by the proof in this manuscript. A secondary issue is that the abstract omits the condition αr > s+3 and the ε^0_X term present in Theorems 3.5/3.12, making the advertised claim look stronger than what is proved; this is presentation-level but worth correcting.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper characterizes the distance in the inhomogeneous Lipschitz space Λ_s from a function f to a subspace V = Λ^s_X (defined via Daubechies wavelets and a quasi-normed lattice X) by the critical index of the fractional heat semigroup 'bad' sets, together with a coarse-scale wavelet term. The two main general results are Theorem 3.5, under Assumption I (a doubling condition on X), and Theorem 3.12, under Assumption II (a Carleson-type measure condition). Applications are given to Sobolev, Besov, Triebel–Lizorkin, Besov-type, and Triebel–Lizorkin-type spaces. The proof is structured as: kernel estimates and a semigroup characterization of Λ_s (Theorem 5.6), three propositions comparing semigroup bad sets with finite-difference bad sets (Propositions 6.1–6.3), and a reduction to the authors' earlier distance characterizations in [11] via Lemmas 4.6–4.8.","tokens_in":49482,"tokens_out":8330,"duration_ms":82022,"significance":"If fully substantiated, this is a significant unified result: it extends the Garnett–Jones and Saksman–Soler i Gibert distance characterizations to all s∈(0,∞) and all fractional semigroup orders α∈(0,∞), for a broad class of subspaces. The paper is transparent in structure and contains self-contained proofs of the fractional heat kernel estimates (Theorem 5.1), the semigroup characterization of Λ_s (Theorem 5.6), and the derivative estimates (Theorem 5.7). The main theorems are explicitly reduced to three technical propositions, whose proofs are detailed conditional on one quoted estimate. The manuscript is not machine-checked, but the logical dependencies are clearly delineated.","major_comments":[{"comment":"The pointwise ball-average estimate |f(x)−B_{ℓ,t}f(x)| ≤ C sup_{t/(8ℓ)≤y≤t/2} sup_{|x′−x|≤4ℓt} Δ^{2ℓ}f(x′,y) is quoted from the authors' companion paper [12, Lemma 5.11], whose arXiv identifier is listed as '???' in the references. This estimate is the sole source for the pointwise bound used in Lemma 9.1; Lemma 9.1 is in turn the basis of Proposition 6.3, which is used in the lower-bound proof of Theorem 3.5 (Eq. (6.4)) and Theorem 3.12 (Eq. (10.4)). The published reference [10] is cited only for the uniform Λ_s estimate (9.4), not for (9.3). As written, the lower-bound chain in the main theorems is conditional on an unverified external statement. Please provide a proof, a precise available reference, or explicit hypotheses under which (9.3) is valid and verify that they are satisfied by f∈Λ_s.","section":"§9, Lemma 9.2 and Eq. (9.3)"}],"minor_comments":[{"comment":"The abstract and the introductory summary state the distance equivalence as ε_{α,r,s,ν}(f)∼dist(f,V)_{Λ_s}, but Theorems 3.5 and 3.12 require αr>s+3 and characterize the distance by ε_X f + ε^0_X f (or their ν-analogues), with an additional coarse-scale wavelet term. The abstract should be adjusted so that the advertised claim matches the proved statement.","section":"Abstract and §2 vs. Theorems 3.5/3.12"},{"comment":"The closure criterion for B^s_{∞,q} is stated using the finite-difference sets S_{r,j}(s,f,ε). This contradicts Section 2 and Theorem 11.5(ii), which use the semigroup bad sets D_{α,r,j}(s,f,ε). This appears to be a typo: S_{r,j} should be D_{α,r,j}.","section":"Theorem 11.6(ii)"},{"comment":"In the proof of Theorem 3.12, the references to 'Definition 3.10 (iii)' in the neighborhood-finiteness equivalence should be to Definition 3.10(iv). Also, the sentence 'This proves the upper estimate (10.3)' should read 'lower estimate'.","section":"§10"},{"comment":"Reference [12] is listed with 'arXiv: ???'. Apart from the mathematical dependence discussed above, the reference itself is incomplete and must be replaced with a proper identifier before submission.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main theorems are plausible and the reduction to Propositions 6.1–6.3 is clear, but the current manuscript is not self-contained at a load-bearing point: Eq. (9.3) is taken from an unpublished companion paper with no available identifier. I would like to see the companion [12] posted and the reference completed, and a short verification (or proof sketch) of (9.3) included. The dependence on [11] is also substantial, but [11] is available (arXiv:2505.16116) and is cited precisely, which is acceptable. The abstract overstates the main results slightly; this is easily fixed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a genuine extension of the Garnett–Jones program: the authors use fractional heat semigroup derivatives to characterize distances in Lipschitz spaces for all α>0 and s>0, which goes well beyond Saksman–Soler i Gibert's Poisson-semigroup result and replaces the finite-difference machinery of their companion [11] with a semigroup approach. The general framework (Assumptions I/II, Daubechies wavelet lattices) is structurally sound, and the applications to Besov, Triebel–Lizorkin and type spaces follow cleanly. The kernel estimates in Section 5 and the K-functional proof of Theorem 5.6 are solid, and Proposition 6.1's hyperbolic-metric argument is neat. This is serious work from people who know the area.\n\nThe soft spot is exactly where the stress test lands. The lower bound in Theorems 3.5 and 3.12 depends on Proposition 6.3, which is derived from Lemma 9.1. Lemma 9.1's proof uses estimate (9.3), the pointwise ball-average bound, quoted from [12, Lemma 5.11] — an unpublished companion with the arXiv identifier literally listed as '???'. This is load-bearing. I could not verify (9.3) from the text; the published source [10] is cited only for the uniform Λ_s bound (9.4), not for (9.3). If (9.3) needs extra hypotheses (a larger neighborhood, a different range of y, or higher regularity), the proof of Proposition 6.3 fails, and with it the lower bound. The statement may still be true, but the paper as written does not support it.\n\nSecond, minor but real: the abstract promises a characterization for 'a broad class of subspaces' without mentioning the technical condition αr > s+3 or the extra ε^0_X term that appears in Theorems 3.5 and 3.12. So the advertised theorem is stronger than what is proved. This is presentation, but worth fixing.\n\nI don't share any worry about circularity: the definitions are not rigged. The equivalence really goes through the inclusions and the external finite-difference/wavelet characterizations.\n\nBottom line: this deserves a serious referee, but the referee must have the companion lemma in hand. I'd send it to peer review with a clear request: either prove Lemma 9.2 in the paper, or cite a publicly available version of [12], and align the abstract with the theorems. If that happens, acceptance is plausible. Right now, standalone verification is impossible. I'd cite this paper if I were working in this area, but I'd wait for the companion to appear.","headline":"A serious extension of Garnett–Jones via fractional heat semigroups, but the lower bound hangs on an unpublished companion lemma with a missing identifier; referee it, but require the lemma to be supplied.","tokens_in":50003,"tokens_out":2951,"would_cite":true,"duration_ms":30191,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46E35","26A16","35K08","42C40","42E35"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the distance from a Lipschitz function to a broad family of subspaces—Sobolev, Besov, Triebel–Lizorkin, and Besov-type spaces—is characterized, up to constants, by a critical threshold of the fractional heat semigroup","keywords":["fractional heat semigroup","distance in Lipschitz spaces","critical index","bad sets","Daubechies wavelets","Besov–Triebel–Lizorkin spaces","Carleson-type measures","quasi-normed lattices"],"falsifier":"Take a concrete f∈Λ_s, e.g. on R with s=1/2 and a truncated linear ramp, choose α=1, r=2, V=J^s(bmo), and compute both the critical index ε_{α,r,s,ν}(f) and the true distance inf_{g∈V}∥f-g∥_{Λ_s}; if their ratio exceeds the equivalence constants asserted in Theorem 11.12, the theorem is false. A sharper test is to check Lemma 9.2 directly: build a sequence of continuous functions with uniformly small Δ^{2ℓ}f but ball-average error growing in t, which would break the chain of inequalities in Proposition 6.3.","tokens_in":48968,"feed_emoji":"🔥","tokens_out":7823,"duration_ms":70040,"temperature":0.7,"pith_summary":"The paper's goal is to compute the distance from a function f in the inhomogeneous Lipschitz space Λ_s to a non-dense subspace V by looking only at the fractional heat semigroup e^{-t(-Δ)^{α/2}}. It shows that the distance is equivalent to the smallest ε for which the set of points where |∂_t^r e^{-t^α(-Δ)^{α/2}} f(x)| exceeds ε t^{s-rα} has finite size, measured by an admissible set function tailored to V. This gives one uniform formula covering intersections of Λ_s with Sobolev, Besov, Triebel–Lizorkin, and Besov-type spaces for every smoothness s>0 and every order α>0. The result extends earlier Poisson-semigroup and finite-difference characterizations, and it yields a description of the closure of each such subspace in Λ_s.","feed_headline":"Semigroup bad sets measure distance to subspaces","feed_subtitle":"A threshold on fractional-heat derivatives pins down Lipschitz distance to Sobolev, Besov, and Triebel–Lizorkin spaces.","key_machinery":"The central object is the bad set D_{α,r}(s,f,ε) = {(x,t)∈R^n×(0,1]: |∂_t^r (T_{α,t^α} f)(x)| > ε t^{s-rα}}, together with an admissible set function ν and the critical index ε_{α,r,s,ν}(f) = inf{ε: ν(D)<∞}; the distance to the subspace is shown to be comparable to this critical index. The technical engine is a chain of comparisons between these semigroup bad sets and finite-difference super-level sets S_{r1,j}, carried by higher-order ball average operators and the estimate |f(x)-B_{ℓ,t}f(x)| ≤ C sup Δ^{2ℓ} f(x',y) (quoted from the companion paper [12]), along with hyperbolic-metric stability of the bad sets.","core_discovery":"Under a doubling-type condition on the underlying quasi-normed lattice (Assumption I), Theorem 3.5 asserts that for every f∈Λ_s, d(f,Λ^s_X)_{Λ_s} ∼ ε^0_X f + ε_X f, where ε_X f is the critical index of the fractional-heat bad sets D_{α,r,j}(s,f,ε) and ε^0_X f tracks coarse-scale father-wavelet coefficients. Under the weaker Carleson-type measure condition (Assumption II), Theorem 3.12 gives the analogous equivalence with a Carleson-type measure ν. The paper's applications translate these general statements into explicit formulas for Besov, Triebel–Lizorkin, and Besov-type spaces, including endpoint cases with p=∞, and into criteria for membership in the Λ_s-closures of these spaces.","pith_inferences":["Editorial inference: the same bad-set/critical-index scheme should transfer to other semigroups with comparable pointwise kernel decay, e.g. symmetric α-stable processes, giving distance characterizations in their associated Hölder-type spaces.","Editorial inference: since ε_X f is defined through time-integrated bad sets, the result suggests a sampling algorithm that estimates Lipschitz distances by evaluating semigroup derivatives on dyadic time shells, without building wavelet coefficients.","Editorial inference: a quantitative version of the cited ball-average estimate would convert the equivalence into explicit approximation-error bounds, including the dependence of constants on dimension and regularity.","Editorial inference: the split into a coarse-scale term ε^0_X f and a multiscale term ε_X f suggests that approximation in Λ_s decomposes independently into large-scale (father wavelet) and small-scale (semigroup bad set) contributions."],"forward_implications":["The Λ_s distance to every subspace covered by the framework is determined by a single semigroup-derivative threshold, with constants depending only on framework parameters.","Membership in the closure of Λ^s_X is equivalent to finiteness of the bad-set measure at every ε>0 plus vanishing coarse wavelet coefficients.","The earlier Poisson-semigroup theorem for J^s(bmo) is extended from s∈(0,1] and α=1 to all s>0 and all α>0, including the nonlocal cases α≠1,2 where the Laplace equation is unavailable.","Endpoint spaces such as F^s_{∞,q} and B^s_{∞,q}, which fail the doubling condition, are handled by the Carleson-type measure formulation.","For Besov and Triebel–Lizorkin spaces the distance formula becomes an explicit expression involving the measure of the semigroup bad sets at dyadic time scales."],"supporting_citations":[{"why":"supplies the ball-average error estimate (9.3) that carries the lower-bound half of the main theorems","marker":"[12, Lemma 5.11]"},{"why":"establishes the finite-difference distance characterization used as the starting point for the upper bound under Assumption I","marker":"[11, Theorem 2.4]"},{"why":"supplies the finite-difference distance characterization used for the Assumption II case","marker":"[11, Theorem 2.5]"},{"why":"provides the ball-average approximation estimates and Fourier multiplier properties of the operators B_{ℓ,t}","marker":"[10, Theorem 1]"},{"why":"gives the K-functional equivalence for holomorphic semigroups used to characterize Λ_s by semigroup derivatives","marker":"[13, Theorem 5.1]"},{"why":"constructs the Daubechies wavelet system in which the subspaces Λ^s_X are defined","marker":"[34, Sections 3.8 and 3.9]"},{"why":"supplies the wavelet characterization identifying Besov spaces with Daubechies s-Lipschitz X-based spaces","marker":"[55, Proposition 1.11]"},{"why":"provides the Poisson-semigroup predecessor theorem that the present result generalizes to all α>0","marker":"[39, Theorem 4]"}],"fun_headline_variants":["Semigroup bad sets reveal distance to subspaces","Fractional heat pinpoints Lipschitz distance to subspaces","Bad-set critical index equals subspace distance","Heat semigroup measures distance to Sobolev, Besov spaces"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is the ball-average estimate quoted from the authors' unpublished companion paper [12]: for every continuous bounded f, the pointwise error |f(x)-B_{ℓ,t}f(x)| is controlled by a supremum of 2ℓ-th order finite differences on a neighboring space-time window; if this estimate fails or needs extra hypotheses, the proof of the lower bound for both main theorems collapses.","fun_headline_variants_meta":{"raw":{"variants":["Semigroup bad sets reveal distance to subspaces","Fractional heat pinpoints Lipschitz distance to subspaces","Bad-set critical index equals subspace distance","Heat semigroup measures distance to Sobolev, Besov spaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00077,"raw_usage":{"total_tokens":3441,"prompt_tokens":1132,"completion_tokens":2309,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":876,"completion_tokens_details":{"reasoning_tokens":2245}},"tokens_in":876,"tokens_out":2309,"duration_ms":16659,"temperature":1.0,"reasoning_tokens":2245,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T14:25:56.637597+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a concrete f∈Λ_s, e.g. on R with s=1/2 and a truncated linear ramp, choose α=1, r=2, V=J^s(bmo), and compute both the critical index ε_{α,r,s,ν}(f) and the true distance inf_{g∈V}∥f-g∥_{Λ_s}; if their ratio exceeds the equivalence constants asserted in Theorem 11.12, the theorem is false. A sharper test is to check Lemma 9.2 directly: build a sequence of continuous functions with uniformly small Δ^{2ℓ}f but ball-average error growing in t, which would break the chain of inequalities in Proposition 6.3.","supporting_citations":[],"review_version":1}