{"id":"48452730-04c2-47fb-835f-fdc988f7e447","arxiv_id":"2509.18003","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Wave operators for fractional Schrödinger operators are proven bounded on Lp for all 1≤p≤∞ under decay or smallness conditions on the potential, yielding dispersive and Strichartz estimates for the perturbed flow.","lead":"This paper proves that wave operators tied to fractional versions of the Schrödinger equation remain bounded on every Lp space, given potentials that are small or decay fast enough. The result extends earlier work on integer-order operators and yields new time-decay and Strichartz estimates for the perturbed fractional equation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.2's proof depends on unproved resolvent bounds (Props 3.3, 3.4) imported from placeholder preprint [15]; if those bounds fail, the Lp conclusion is unsupported.","rationale":"The reader correctly identifies condition (iv) as load-bearing, but condition (iv) is an explicit part of Theorem 1.2's hypothesis, not a hidden assumption. The more acute concern for the proof's correctness is that the paper's central proof depends on Propositions 3.3 and 3.4, which are stated without proof and attributed to an unpublished, incompletely cited preprint [15]. Reference [15] is literally a placeholder title ('dispersive estimates for fractional Schrodinger or something'), which prevents independent verification. Even the low-energy Lemma 3.2 is only sketched with pointers to earlier papers. The self-contained portions — the Born series bounds (Section 2) and the low-energy kernel estimates modulo Proposition 3.3 — appear substantial, so the paper is not obviously wrong, but the main theorem's proof is gated on unverifiable imported results. This does not change the reader's CONDITIONAL verdict: it reinforces it. Therefore NO verdict change is recommended.","tokens_in":26152,"tokens_out":9346,"duration_ms":82790,"concrete_test":"Obtain the complete text of [15] and verify Proposition 3.4 directly: for a non-integer α>1 (e.g., α=3/2, n=5) with |V(x)|≲<x>^{-β}, β>1, prove or disprove the bound ||<x>^{-1/2-}R_V(λ^{2α})<y>^{-1/2-}||_{L2→L2} ≲ λ^{1-2α} for λ≥1, using stationary phase asymptotics of the fractional resolvent kernel. Also check the derivative bounds (19)-(20) in Proposition 3.3 for λr≲1. If the high-energy bound fails, compute the true λ-growth; if it is worse than λ^{1-2α}, the kernel estimate (31) is not admissible and the Lp conclusion for large potentials is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central Lp boundedness claim for non-small potentials (Theorem 1.2) rests on three technical estimates: Proposition 3.3 (free resolvent kernel representation and derivative bounds), Proposition 3.4 (high-energy limiting absorption bound ||<x>^{-1/2-} R_V(λ^{2α}) <y>^{-1/2-}||_{L2→L2} ≲ λ^{1-2α}), and Lemma 3.2 (low-energy bounds on Γ_ℓ(λ)). Propositions 3.3 and 3.4 are stated without proof and attributed to the authors' own preprint [15]; reference [15] appears as 'dispersive estimates for fractional Schrodinger or something, preprint' — a placeholder, not a verifiable citation. Lemma 3.2 is only sketched, referencing [16,17]. If Proposition 3.4's bound is false — for example, if the non-integer α stationary phase produces a stronger λ-growth — then the pointwise domination (31) in Proposition 5.1 fails and the high-energy tail integral need not be admissible. Thus, even assuming the explicit no-positive-eigenvalues hypothesis (iv), the proof is not self-contained and the conclusion is only as secure as the unpublished [15]. The reader's embedded-eigenvalue concern is real but is an explicit hypothesis; the deeper issue is that the proof of the theorem under that hypothesis is incomplete here.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the L^p-continuity of wave operators for fractional Schrödinger operators H=(-Δ)^α+V(x) with α>1 non-integer and n>2α. The main results are Theorem 1.1 (small potentials) and Theorem 1.2 (large decaying potentials with a spectral assumption), both asserting that the wave operators extend to bounded operators on L^p for all 1≤p≤∞. The proof uses the stationary representation of the wave operator, a Born-series expansion, and a low/high-energy decomposition. The low-energy analysis relies on the free resolvent kernel representation (Proposition 3.3) and bounds on the operators Γ_ℓ(λ) (Lemma 3.2); the high-energy analysis relies on a limiting absorption bound (Proposition 3.4) and a pointwise kernel domination (Proposition 5.1). Corollaries give dispersive and Strichartz estimates for the perturbed semigroup.","tokens_in":26455,"tokens_out":3401,"duration_ms":32134,"significance":"If all technical inputs are valid, the result is a significant extension of the integer-order wave-operator L^p theory of [16,17] to non-integer α, and it would provide the first such L^p statements for non-local fractional Schrödinger operators. The announced dispersive and Strichartz corollaries are natural and potentially useful. The paper is clearly written and the overall strategy is a coherent adaptation of prior work. However, the present manuscript is not self-contained: several load-bearing estimates are either stated without proof and attributed to the authors' own preprint [15], which appears only as a placeholder reference, or are only sketched with pointers to [16,17]. The central claims are therefore conditional on unpublished or non-verifiable ingredients, and the lack of detailed proofs for those estimates is the main obstacle to acceptance.","major_comments":[{"comment":"Propositions 3.3 and 3.4 are the key resolvent estimates used throughout the low- and high-energy arguments. Proposition 3.3 supplies the kernel representation (18) and the derivative bounds (19)–(20); Proposition 3.4 supplies the high-energy limiting absorption bound ∥⟨x⟩^{-1/2-}R_V(λ^{2α})⟨y⟩^{-1/2-}∥_{L^2→L^2}≲λ^{1-2α}. Both are stated without proof and attributed to the authors' preprint [15], whose entry in the bibliography reads \"dispersive estimates for fractional Schrodinger or something, preprint\". These estimates are load-bearing: the pointwise domination (31) in Proposition 5.1 and the admissibility arguments in Proposition 3.1 both depend on them. The manuscript cannot be verified until either complete proofs are included or [15] appears as a checkable reference with theorem numbers and proofs.","section":"Section 3, Propositions 3.3 and 3.4"},{"comment":"Lemma 3.2 is the main low-energy ingredient, but its proof is only a sketch. In particular, the claimed L^2-boundedness of the kernel (28) for β>n_*, the derivative bounds on [M_+(λ)]^{-1} in (29), and the large-ℓ decay estimate for A(λ,z_1,z_2) in (30) are asserted with references to [16,17] and a few sentences. These assertions are essential: Lemma 3.2 is what allows the low-energy tail to satisfy the hypotheses of Proposition 3.1, and without it the low-energy part of Theorem 1.2 is unsupported. The sketch may be a reasonable summary, but in a journal submission the full proof should appear, especially because the fractional case lacks the splitting identity (4) used in the integer-order arguments.","section":"Section 4, Lemma 3.2"},{"comment":"Proposition 5.1, which provides the pointwise bound (31) for the high-energy tail, is stated with the comment that the proof is \"a straightforward modification of the proof of Propositions 5.3 and 6.5 in [16]\". This is not adequate for the present setting. The integer-order argument uses the splitting identity (4), which the authors explicitly note is unavailable for fractional α; Proposition 3.4 is the only imported high-energy input. Since (31) is what combines with Lemma 5.2 to yield admissibility of the high-energy tail, a detailed proof of Proposition 5.1 is necessary. At minimum, the authors should spell out how the fractional resolvent bounds substitute for each step of the integer-order proof.","section":"Section 5, Proposition 5.1"},{"comment":"The theorem assumes away positive eigenvalues and threshold obstructions. This is an explicit hypothesis, so it is not an internal inconsistency. However, the introduction also states \"We leave the lack of embedded eigenvalues as an overarching assumption,\" which means the theorem is conditional on a spectral property that is not derived from the decay assumptions. Given the examples of Cuenin [10], this limitation should be stated as prominently in the abstract or theorem as it is in the body; the current formulation is acceptable mathematically but may overstate the class of potentials covered if readers overlook assumption (iv).","section":"Theorem 1.2, condition (iv)"}],"minor_comments":[{"comment":"Reference [15] is listed as \"dispersive estimates for fractional Schrodinger or something, preprint\". This is clearly a placeholder and must be replaced with a complete citation or the content must be included in the paper.","section":"References"},{"comment":"There are typographical inconsistencies in the text, e.g. \"Schr¨odinger\" vs. \"Schrödinger\", \"W A VE OPERATORS\" in headers, and some corrupted accents. The paper would benefit from a careful proofreading pass.","section":"Throughout"},{"comment":"The statement ends with \"provided that β > n\" without defining β in the proposition. Presumably β refers to the decay of V in the hypotheses of Lemma 3.2, but this should be made explicit.","section":"Proposition 3.1"},{"comment":"The statement of Lemma 2.4 says \"Morever\" for \"Moreover\", and the notation γ<0 with the case γ+j=-n may need a brief clarification of the logarithmic case; this is minor but worth correcting.","section":"Section 2, Lemma 2.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's central theorem is plausible and the strategy is a reasonable continuation of the authors' previous work, but the dependence on [15] is not merely cosmetic: Propositions 3.3 and 3.4 are the analytic core of both the low- and high-energy arguments. As it stands, the paper cannot be checked against the published record. I would encourage the editor to request that the authors either include full proofs of these propositions and of Lemma 3.2 and Proposition 5.1, or make [15] available as a preprint with a stable, complete citation. The self-citation pattern is understandable here, but the placeholder reference makes it impossible to assess novelty and correctness independently."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is the first paper to prove L^p boundedness of wave operators for fractional Schrödinger operators with non-integer α>1, and it does so by a legitimate adaptation of Yajima's scheme and the authors' integer-order work. The main new technical pieces in the paper itself are the Fourier multiplier bounds in Lemma 2.2, which are actually proved, and the low-energy Born series analysis in Section 3. If the theorem is true, the consequences — global L^1-to-L^∞ decay and Strichartz estimates for the perturbed equation in all n>2α — are real and useful.\n\nThe soft spots are not in the architecture but in what is imported. Theorem 1.1 omits the no-positive-eigenvalues assumption that Section 5 states is needed; either the smallness assumption rules out embedded eigenvalues (which is not shown) or the statement is missing a hypothesis. That is fixable but real. More seriously, the proof of Theorem 1.2 loads three weight-bearing estimates onto work not contained in this paper: Proposition 3.3 (free resolvent representations), Proposition 3.4 (high-energy limiting absorption bound), and to a lesser extent Lemma 3.2. The first two are stated without proof and attributed to the authors' own preprint [15], which appears in the bibliography as \"dispersive estimates for fractional Schrodinger or something, preprint.\" That is a placeholder, not a verifiable citation. If Proposition 3.4's bound has the wrong λ growth — say, the stationary phase for non-integer α gives an extra power — then the pointwise domination in Proposition 5.1 fails and the high-energy tail is not admissible. I have not verified every oscillatory integral line, so I cannot say the bound is false; I can say the paper's central conclusion is only as secure as an unpublished manuscript. Lemma 3.2 and Proposition 5.1 are sketched or referenced to [16,17], which is less concerning because those are published and the adaptation is plausible, but the paper is not self-contained.\n\nThe embedded-eigenvalue concern is real but is an explicit hypothesis in Theorem 1.2, so it is not a hidden flaw. And the self-citation pattern is heavy but not circular: the earlier papers do not contain the L^p theorem.\n\nWho should read this: anyone working on dispersive estimates or scattering for higher-order or fractional Schrödinger operators. A serious referee can add value, mainly by checking the imported estimates. Send it to review, but ask for a revision that supplies or verifies Propositions 3.3 and 3.4, fixes the placeholder reference, and reconciles the Theorem 1.1 assumption.","headline":"First L^p wave operator result for non-integer α is legitimate in design, but its central estimates are imported from an unpublished placeholder referenced preprint.","tokens_in":790,"tokens_out":828,"would_cite":false,"duration_ms":31168,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P25","47A40","35R11"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the wave operators for the fractional Schrödinger operator (−Δ)^α+V, with α>1 and n>2α, extend to bounded operators on L^p(ℝ^n) for every 1≤p≤∞, provided the real-valued potential decays sufficiently and the operator h","keywords":["wave operators","L^p boundedness","fractional Schrödinger operator","dispersive estimates","Strichartz estimates","resolvent estimates","Born series","embedded eigenvalues"],"falsifier":"Take a real-valued potential V with |V(x)|≲⟨x⟩^{−β}, β>n_*, for which H=(−Δ)^α+V is known to have a positive embedded eigenvalue (examples of this kind exist for α>1). If the weighted resolvent bound ∥⟨x⟩^{−1/2−}R_V(λ^{2α})⟨y⟩^{−1/2−}∥_{L^2→L^2}≲λ^{1−2α} fails near such an eigenvalue, Proposition 3.4 and hence the high-energy step of the proof cannot hold, confirming the assumption is necessary; if the bound somehow persists, the theorem could be improved to drop or relax assumption (iv).","tokens_in":26011,"feed_emoji":"🌊","tokens_out":9719,"duration_ms":93375,"temperature":0.7,"pith_summary":"The paper proves that the wave operators for the fractional Schrödinger operator H=(−Δ)^α+V(x), with α>1 and n>2α, extend to bounded operators on L^p(ℝ^n) for every 1≤p≤∞, assuming the real-valued potential V decays sufficiently and the operator has no positive eigenvalues and a regular zero energy. The wave operators compare the free fractional evolution with the perturbed evolution, and their L^p boundedness is the standard route to turning dispersive and Strichartz estimates for the free operator into estimates for the perturbed operator via the intertwining identity. The argument is the first to cover non-integer α, where the resolvent splitting identities used for integer orders are unavailable; instead the paper develops direct bounds on the Fourier multipliers appearing in a Born series expansion of the stationary representation.","feed_headline":"Wave operators bounded on all L^p for fractional Schrödinger","feed_subtitle":"For α>1 and n>2α, decaying potentials give full L^p continuity, unlocking dispersive and Strichartz estimates.","key_machinery":"The main engine is the stationary representation of the wave operator as W_+ = I − (1/2πi)∫_0^∞ R_V^+(λ)V[R_0^+(λ)−R_0^−(λ)] dλ, recast after the change of variables λ↦λ^{2α}. The perturbed resolvent R_V is expanded in a Born series via the second resolvent identity, and each summand is controlled on L^p by writing its integral kernel in terms of the functions h_k = F^{-1}(p_ω), where p_ω(ξ)=(|ξ−ω|^2−|ξ|^2)/(|ξ−ω|^{2α}−|ξ|^{2α}). Lemma 2.2 supplies uniform L^1 bounds and convergence for these multipliers, which replace the algebraic splitting identity that exists for integer α. For low energies, the tail of the series is dominated by absolutely bounded kernels using the resolvent representat","core_discovery":"The central claim is that L^p-continuity of wave operators, a property long known for the classical Schrödinger operator and later extended to integer powers (−Δ)^m, holds for every real α>1 in the fractional setting. Fix α>1 and n>2α. If V is real-valued with pointwise decay |V(x)|≲⟨x⟩^{−β} for β>n_* (where n_*=n+4 for odd n, n+3 for even n), with the appropriate Sobolev condition when n=4α−1 or Fourier-L^r condition when n>4α−1, and if H has no positive eigenvalues and zero energy is regular, then the wave operators extend to bounded operators on L^p(ℝ^n) for all 1≤p≤∞. The proof splits into a low-energy analysis, where the Birman–Schwinger-type operator M_+(λ) is inverted and the Born-ser","pith_inferences":["If the high-energy limiting absorption bound (Proposition 3.4) could be proved under weaker spectral assumptions, the theorem would extend to potentials with positive eigenvalues whose resonances are suitably controlled; the paper's reliance on the no-positive-eigenvalues assumption suggests this is the main obstacle to a fully unconditional statement.","The critical dimension n=4α−1 uses an H^{0+} condition that the paper suspects may be an artifact; a natural test is whether a slightly weaker Sobolev regularity (e.g., H^0 instead of H^{0+}) still yields L^p boundedness, which would simplify the theorem.","One expects, by analogy with the integer-order case, that zero-energy resonances or eigenvalues would shrink the range of p (typically to 1<p<n/(2α) or smaller with orthogonality conditions); the paper explicitly plans to address threshold obstructions in future work.","The Fourier-L^r condition for n>4α−1 is reminiscent of smoothness assumptions; it would be worthwhile to test numerically for the smallest σ that actually suffices, since the paper's σ is expressed with a δ margin and may be improvable."],"forward_implications":["Under the theorem's hypotheses, the dispersive bound ∥e^{−itH}P_ac(H)∥_{L^p→L^{p'}}≲|t|^{−n/α(1/2−1/p)} holds for every 1≤p≤2, including the global L^1→L^∞ decay |t|^{−n/(2α)}.","Strichartz estimates for the perturbed fractional flow follow: ∥e^{−itH}P_ac(H)f∥_{L^q_t L^r_x}≲∥f∥_{L^2} for admissible pairs satisfying 2/q=n/α(1/2−1/r), 2≤r<∞.","The weighted dispersive family ∥e^{−itH}H^{(γ−n)/(2α)}P_ac(H)∥_{L^1→L^∞}≲|t|^{−γ/(2α)} holds for 0<γ≤nα.","These estimates extend the authors' earlier dispersive results for fractional operators to all dimensions n>2α, in particular beyond the previously treated range n≤4α−1.","The small-potential version (Theorem 1.1) shows the same L^p conclusions under explicit smallness conditions on V, so the result is robust for both small and large potentials."],"fun_headline_variants":["Fractional Schrödinger wave operators L^p-bounded for all p","All L^p continuity for wave operators of fractional Schrödinger","Fractional Laplacian with potentials: wave operators on every L^p","L^p bounds for wave operators in fractional Schrödinger equations","Full L^p control of wave operators for fractional Schrödinger"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that H has no positive eigenvalues and that zero energy is regular; the decay and smoothness conditions on V alone do not rule out embedded positive eigenvalues when α is not an integer, so this is a genuine spectral hypothesis required by both the low-energy inversion of M_+(λ) and the high-energy resolvent bound.","fun_headline_variants_meta":{"raw":{"variants":["Fractional Schrödinger wave operators L^p-bounded for all p","All L^p continuity for wave operators of fractional Schrödinger","Fractional Laplacian with potentials: wave operators on every L^p","L^p bounds for wave operators in fractional Schrödinger equations","Full L^p control of wave operators for fractional Schrödinger"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00016,"raw_usage":{"total_tokens":1048,"prompt_tokens":701,"completion_tokens":347,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":445,"completion_tokens_details":{"reasoning_tokens":256}},"tokens_in":445,"tokens_out":347,"duration_ms":3619,"temperature":1.0,"reasoning_tokens":256,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T15:46:39.657862+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a real-valued potential V with |V(x)|≲⟨x⟩^{−β}, β>n_*, for which H=(−Δ)^α+V is known to have a positive embedded eigenvalue (examples of this kind exist for α>1). If the weighted resolvent bound ∥⟨x⟩^{−1/2−}R_V(λ^{2α})⟨y⟩^{−1/2−}∥_{L^2→L^2}≲λ^{1−2α} fails near such an eigenvalue, Proposition 3.4 and hence the high-energy step of the proof cannot hold, confirming the assumption is necessary; if the bound somehow persists, the theorem could be improved to drop or relax assumption (iv).","supporting_citations":[],"review_version":1}