{"id":"a791de89-da17-4657-9b6c-9b6def9ec91c","arxiv_id":"2506.16378","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For higher-order Schrödinger operators with a threshold eigenvalue in dimensions 2m < n ≤ 4m, the wave operators are bounded on L^p for the natural range, with extended ranges and a new L∞ endpoint under extra orthogonality conditions.","lead":"This paper proves new L^p boundedness ranges for the wave operators of higher-order Schrödinger operators H = (−∆)^m + V in dimensions 2m < n ≤ 4m when zero energy is an eigenvalue but not a resonance. It extends the authors' previous high-dimensional results, covers the classical m=1 case, and adds a new L∞ endpoint under extra orthogonality conditions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 omits the no-resonance hypothesis on which the Section 4 threshold classification and representation (8) depend; as displayed, the stated Lp ranges are not established in mixed eigenvalue-resonance cases.","rationale":"The reader's CONDITIONAL verdict is appropriate. I agree that the threshold classification is the weakest point, but I would sharpen it to a concrete statement-level defect: Theorem 1.1 drops the no-resonance condition that the Section 4 proof explicitly invokes and that the abstract includes. This does not invalidate the intended theorem, because the surrounding text supplies the missing hypothesis, but it does mean the displayed theorem overclaims in the mixed eigenvalue-resonance case. Under the intended no-resonance hypothesis I do not see an internal inconsistency: the dimensional bookkeeping k0=2m−⌈n/2⌉+1 reproduces the stated ranges, Lemma 4.3 provides the needed expansion, and Proposition 2.2 coherently converts the singular terms into Schur-test bounds. The additional high-energy assumptions from [6] are also omitted from the abstract but are present in Corollary 1.4, so that is a presentation issue rather than a mathematical obstruction. No change to the reader's conditional verdict is needed.","tokens_in":27051,"tokens_out":19642,"duration_ms":183516,"concrete_test":"Run the Section 4 inversion of B(λ) (Eqs. 27 and 35) for a concrete mixed obstruction in 2m<n≤4m, e.g. m=1, n=3, with V chosen so that T0 has kernel dimension at least two, one element in S2 (eigenvalue) and one outside S2 (resonance). If the leading term of B(λ) is not −T_{k+1} with T_{k+1} invertible on S1, then representation (8) and the p-ranges of Theorem 1.1 fail without the no-resonance hypothesis; if it is, the omitted hypothesis is harmless and the statement stands as printed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The displayed Theorem 1.1 asserts the Lp range 1≤p<2n/(n−1) (n odd) or 1≤p<2n/(n−2) (n even) under only 'an eigenvalue at zero, but no positive eigenvalues.' The proof, however, requires the eigenvalue-only classification stated in Section 4: for 2m<n≤4m, 'having only an eigenvalue at zero (no resonances) means S1=Sk+1 and Tk+1:=S1 v G_{1,L} v S1 is invertible on S1L2.' This identification is exactly what makes B(λ) in (27) invertible and produces the singular structure (8); it is used in Proposition 4.2 and Section 5. In these dimensions a zero-energy resonance can coexist with a zero-energy eigenvalue, and in that mixed case S1≠Sk+1 or T_{k+1} fails to be invertible, so the expansion of M(λ)^{-1} and the resulting p-ranges change. Thus Theorem 1.1 as stated is not supported. The preceding paragraph and the abstract do say 'no resonances,' so the intended theorem is likely correct once that hypothesis is added; this is a statement-level gap rather than a proof error, but it is load-bearing because the entire low-energy argument rests on the classification.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves Lp boundedness of the wave operators for H=(-Δ)^m+V in dimensions n>2m when the perturbed operator has a zero-energy eigenvalue. The main new contribution is the low-energy regime 2m<n≤4m, where threshold resonances are possible: under a no-resonance assumption, the authors obtain the range 1≤p<2n/(n-1) for n odd and 1≤p<2n/(n-2) for n even. With additional orthogonality of the eigenspace to {x^α V: |α|≤k0-1}, the range is extended to 1≤p<n/(2m-k0), with endpoint ranges [1,∞) and [1,∞] when k0=2m and k0=2m+1 respectively. The proof uses the stationary representation of the wave operator, the Jensen-Nenciu resolvent inversion scheme, and a finite-rank decomposition of the singular part of M(λ)^{-1}; the remaining singular terms are controlled by a new Proposition 2.2. The authors state that the arguments apply uniformly to the classical m=1 case and yield a new L∞ endpoint for n>3.","tokens_in":27173,"tokens_out":7882,"duration_ms":68358,"significance":"If the results are correct as intended, this is a substantial extension of the higher-order wave-operator theory: it covers the low-dimensional range 2m<n≤4m where threshold resonances and logarithmic resolvent terms appear, and it does so with a unified even/odd-dimensional treatment. The paper contains genuine technical work: Lemma 4.3 provides the requisite resolvent expansion with logarithmic terms, Proposition 2.2 is proved in-text via the cancellation Lemmas 3.2–3.4, and the dimensional bookkeeping correctly reproduces the claimed endpoint ranges from k0=2m-⌈n/2⌉+1. The claimed new L∞ boundedness in the classical m=1, n>3 eigenvalue-only case is clearly identified and is a concrete advance over the existing p<∞ results. There are no fitted parameters or post hoc exclusions in the derivation. The main limitation is that the theorem statements, as displayed, do not fully state the no-resonance structural hypothesis on which the low-energy inversion argument depends.","major_comments":[{"comment":"The displayed statement of Theorem 1.1 assumes only a zero-energy eigenvalue and no positive eigenvalues, but the proof requires the additional no-resonance classification stated in Section 4: for 2m<n≤4m, 'having only an eigenvalue at zero (no resonances)' means S1=Sk+1 and T_{k+1}:=S1 v G_{1,L} v S1 is invertible on S1L2. This invertibility enters directly in the Jensen-Nenciu inversion (26)-(27), in the expansion of M(λ)^{-1} in Proposition 4.2, and ultimately in the singular structure (8) used in Section 5. Since zero-energy resonances can coexist with a zero-energy eigenvalue in this dimension range, the stated Lp ranges are not established under the hypotheses as written. The abstract and the paragraph preceding Theorem 1.1 do say 'no resonances,' so this is a statement-level gap rather than a proof error, but it is load-bearing and should be fixed by adding the no-resonance hypothesis to the theorem statement.","section":"Section 1, Theorem 1.1"},{"comment":"The same gap affects Theorem 1.2 in dimensions 2m<n≤4m. The orthogonality condition on the eigenspace, {x^α V: |α|≤k0-1}, does not by itself rule out zero-energy resonances; the proof requires the stronger structural assumption S1=S_{k0+1}, which is used in Lemma 4.5 and Proposition 4.2. Consequently, the statement of Theorem 1.2 should include the no-resonance (or S1=S_{k+1}) hypothesis for 2m<n≤4m, and Corollary 1.4, which invokes Theorem 1.2, should be revised accordingly. This is a fixable statement-level issue, but it is central because the entire low-energy argument rests on the eigenvalue-only classification.","section":"Section 1, Theorem 1.2 and Corollary 1.4"}],"minor_comments":[{"comment":"The sentence 'We note that this has a slightly less stringent assumption than [3], which requires β>n+4k+5=8m−n+11 for n odd' appears to have an arithmetic slip: with the paper's k=2m−⌈n/2⌉+1, n+4k+5 = 8m−n+7 for n odd. Please verify the formula for the assumption in [3] or clarify the definition of k used there.","section":"Section 1, p. 3"},{"comment":"The sentence 'We note that the classical case and dimensions 4m≥n>2m, (n=3,4), the existence of threshold eigenvalues limits...' has a grammatical issue; 'and dimensions' should likely be 'in dimensions'.","section":"Section 1, p. 2"},{"comment":"The statement 'In the case 2m<n≤4m, with k=2m−⌈n/2⌉+1, having only an eigenvalue at zero (no resonances) means S1=Sk+1 and T_{k+1}:=S1 v G_{1,L} v S1 is invertible on S1L2' is an external structural classification from [9]. Since it is load-bearing, it would help to label it explicitly as an assumption taken from [9] rather than presenting it as a consequence of the immediately preceding definitions.","section":"Section 4, p. 18"}],"recommendation":"major_revision","confidential_remarks":"The missing no-resonance hypothesis in Theorems 1.1 and 1.2 is the only substantive issue I found; it is a statement-level gap that appears fixable by adding the hypothesis already used in the proof. The paper's reliance on the authors' earlier work for the high-energy bounds and for Proposition 2.1 is standard and appropriately cited. I would ask the authors to double-check the β comparison with [3] and to ensure Corollary 1.4 inherits the corrected hypotheses."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a solid, defensible continuation of the authors' program, and it delivers the advertised new cases. The main technical work—Proposition 2.2, Lemma 4.3, and the reduction to representation (8)—is real, with the dimensional bookkeeping internally consistent. The new L∞ endpoint for m=1, n>3 is genuinely new and will be useful.\n\nThe paper's main new content is the extended range 2m<n≤4m for the eigenvalue-only case, the k0-based orthogonality refinements that push p up to n/(2m−k0), and the treatment of the n=4m boundary including log terms. The high energy part is imported from earlier work [6,7], and that is handled honestly—Corollary 1.4 explicitly invokes Assumption 1.3, even though the abstract doesn't.\n\nThere is one load-bearing statement-level gap that needs fixing. Theorem 1.1 as displayed says only 'an eigenvalue at zero, but no positive eigenvalues.' The proof, however, depends on the eigenvalue-only classification in Section 4: for 2m<n≤4m, no resonances means S1=S_{k+1} and T_{k+1} invertible. In mixed eigenvalue-resonance cases the expansion of M(λ)^{-1} and the resulting p-ranges change. The abstract and the intro do say 'no resonances,' so the intended theorem is almost certainly correct once that hypothesis is added to the statement. But as displayed, Theorem 1.1 is not supported. This is an easy fix, but it must be made before publication.\n\nOther soft spots are minor. For odd n, the base range overlaps with Cheng–Soffer–Wu–Yao [3]; the paper acknowledges this and the sharper orthogonality results still stand. Some kernel-admissibility and pointwise bounds are deferred to prior work, so a careful reader will need all three papers on the table. The proof of Lemma 3.4's fourth bound is a little terse, but the argument is identifiable.\n\nNo circularity, no fitted parameters. The theorems should survive checking. This deserves a serious referee. I'd be happy to referee it myself; the main request is to patch Theorem 1.1's hypotheses.","headline":"Solid technical continuation that closes the threshold-eigenvalue case for 2m<n≤4m; the main fix needed is adding the no-resonance hypothesis to Theorem 1.1.","tokens_in":27962,"tokens_out":2747,"would_cite":true,"duration_ms":25530,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35P25","47A40","81U05"],"pacs":[],"model":"deepseek-v4-flash","headline":"For higher-order Schrödinger operators with a zero-energy eigenvalue, the paper proves $L^p$ boundedness of wave operators up to dimension-dependent thresholds, with extra eigenspace orthogonality raising the range and a new $L^\\infty$…","keywords":["wave operators","L^p boundedness","higher-order Schrödinger operator","threshold eigenvalue","zero-energy resonance","resolvent expansion","dispersive estimate","scattering theory"],"falsifier":"Exhibit a real-valued potential $V$ obeying $|V(x)|\\lesssim\\langle x\\rangle^{-\\beta}$ with $\\beta>\\max(8m-n,n_\\star)+2$, $2m<n\\le 4m$, such that $H=(-\\Delta)^m+V$ has a zero-energy eigenvalue, no resonance, and no positive eigenvalue, yet $W_{low,\\kappa}$ fails to be bounded at some endpoint such as $p=n/(2m-k_0)$ for a $k_0$ satisfying the orthogonality hypotheses. Equivalently, compute the kernel of the singular part in equation (8) for a rank-one model with an explicit zero-energy eigenfunction and test whether the oscillatory integral bounds saturate at the claimed endpoint.","tokens_in":26676,"feed_emoji":"🌊","tokens_out":8464,"duration_ms":63375,"temperature":0.7,"pith_summary":"This paper proves bounds on the $L^p$ operator norm of wave operators for $H=(-\\Delta)^m+V$ in dimensions $2m<n\\le 4m$ when zero energy is an eigenvalue but not a resonance. The main result is that the low-energy part of the wave operator is bounded on the natural range $1\\le p<2n/(n-1)$ for odd $n$ and $1\\le p<2n/(n-2)$ for even $n$; if the zero-energy eigenspace is orthogonal to $\\{x^\\alpha V: |\\alpha|\\le k_0-1\\}$, the range extends to $p<n/(2m-k_0)$, to $p<\\infty$ at $k_0=2m$, and to $p\\le\\infty$ at $k_0=2m+1$. The argument handles even and odd dimensions in one framework, covers the classical $m=1$ case, and yields an $L^\\infty$ endpoint for $n>3$ that had not been established before. Because wave-operator boundedness transfers mapping properties of $f((-\\Delta)^m)$ to $f(H)P_{ac}(H)$, these ranges feed directly into dispersive and Strichartz-type estimates for the perturbed evolution.","feed_headline":"Zero-energy eigenvalue sets L^p ceiling for wave operators","feed_subtitle":"Extends higher-order Schrödinger results to 2m<n≤4m and adds an L∞ endpoint for n>3.","key_machinery":"The load-bearing object is the low-energy stationary representation $W_{low,\\kappa}$, written through the symmetric resolvent identity as an integral of $R_0^+(\\lambda^{2m})v\\,\\Gamma_\\kappa(\\lambda)\\,v[R_0^+(\\lambda^{2m})-R_0^-(\\lambda^{2m})]$. The paper expands $\\Gamma_\\kappa(\\lambda)$ into a smooth part plus a singular finite-rank part $\\lambda^{-2m}\\Gamma_s(\\lambda)$ using a threshold resolvent inversion identity and the threshold classification spaces $S_1,S_2,\\dots$; the singular part is a combination of operators of type $(k_1,k_2,\\alpha)$ acting on free resolvent differences. The crucial quantitative control is Proposition 2.2, which bounds the resulting integral kernels via cancellation lemmas that convert orthogonality of $\\phi\\in S_{k+1}L^2$ against $v x^\\alpha$ into extra powers of $\\lambda$ or of $|x-y|^{-1}$, together with oscillatory integral estimates for the remaining $\\lambda$-integrals. The classification premise that at zero energy $S_1=S_{k+1}$ with $T_{k+1}$ invertible is what turns the threshold eigenvalue into this tractable finite-rank singularity.","core_discovery":"The central claim, stated as Theorem 1.1 and Theorem 1.2 and lifted to the full wave operators by Corollary 1.4, is that the only low-energy obstruction to $L^p$ boundedness is the zero-energy eigenspace, and its effect is controlled by the smallest $k_0$ such that the eigenspace is orthogonal to $x^\\alpha V$ for all $|\\alpha|<k_0$. In dimensions $2m<n\\le 4m$ the generic bound is $p<2n/(n-1)$ for odd $n$ and $p<2n/(n-2)$ for even $n$; with $k_0$ degrees of orthogonality the bound improves to $p<n/(2m-k_0)$, to all finite $p$ when $k_0=2m$, and to $p=\\infty$ when $k_0=2m+1$. For $m=1$, $n>3$, the $L^\\infty$ case is new: previous results stopped at finite $p$ under two orthogonality conditions, aside from the known three-dimensional case. The proof works uniformly in even and odd dimensions by deriving an explicit singular decomposition of $\\Gamma_\\kappa(\\lambda)$ into a smooth operator plus a finite-rank singular term, then controlling every resulting kernel by the same orthogonality-based cancellation mechanism.","pith_inferences":["The stated upper limits are likely optimal: the paper calls the ranges natural, and known low-dimensional resonance cases exhibit unboundedness at analogous endpoints; one would expect the same saturation here even at eigenvalue-only thresholds.","Proposition 2.2 appears applicable beyond the eigenvalue-only case, since its kernel bounds use only orthogonality and decay; a resonance version would likely produce different $p$-ranges and the paper's setup already points toward that extension.","The orthogonality conditions $\\langle \\psi, x^\\alpha V\\rangle=0$ are conditions that can be checked or engineered for concrete potentials; testing them numerically for radial or compactly supported potentials in low dimensions would give direct evidence on whether the ranges saturate."],"forward_implications":["For any potential satisfying Assumption 1.3 and the theorem's hypotheses, the full wave operators $W_\\pm$ are bounded on the stated $L^p$ ranges, so $L^p\\to L^{p'}$ dispersive bounds for $e^{-itH}P_{ac}(H)$ follow with rate $|t|^{-(n/m)(1/2-1/p)}$.","In dimensions $2m<n<4m$ the low-energy part is bounded for every $\\kappa\\ge 0$, so a single smooth cutoff suffices and the $p$-ranges are not an artifact of high-energy bookkeeping.","For the classical Schrödinger operator ($m=1$) the theorem subsumes and streamlines the eigenvalue-only threshold results, and in $n>3$ it produces the first $L^\\infty$ endpoint for wave operators under the orthogonality $k_0=3$.","The decomposition into a smooth operator plus a finite-rank singular term reduces threshold eigenvalue effects to explicit kernels, and the paper notes that the same tools were developed with enough flexibility to handle resonances in lower dimensions."],"supporting_citations":[{"why":"Establishes the Lp boundedness of wave operators with threshold eigenvalues in high dimensions n>4m, the result the present paper adapts downward.","marker":"[8]"},{"why":"Supplies Proposition 2.1, the free resolvent representations of Lemmas 2.3-2.4, and the kernel bounds used to control smooth parts and the Lambda integrals.","marker":"[7]"},{"why":"Gives the high-energy and Born-series Lp bounds assumed through Assumption 1.3 and the a priori bounds on iterated resolvents A(lambda).","marker":"[6]"},{"why":"Provides the threshold classification via the spaces S_i, the eigenvalue/resonance distinction, and the invertibility of T_{k+1} used in Section 4.","marker":"[9]"},{"why":"Gives the threshold resolvent inversion identity used to expand M(lambda)^{-1} into smooth plus singular parts.","marker":"[16]"},{"why":"Supplies the cancellation lemma for S_{k+1}L^2 used in Lemma 3.2 to convert orthogonality into extra decay.","marker":"[1]"},{"why":"Handles the m=1, n=4 eigenvalue-only cases with k0=1 and k0=2, providing the comparison behind the new L-infinity endpoint.","marker":"[31]"},{"why":"Recent low odd dimension threshold-obstruction results; the paper notes its potential-decay assumption is less stringent than this work and uses it as context for resonance cases.","marker":"[3]"}],"fun_headline_variants":["Wave operators L^p-bounded despite threshold eigenvalues","New L∞ bound for wave operators when n>3","L^p bounds for wave operators: threshold eigenvalues no obstacle","Zero-energy eigenvalues no barrier to L^p wave operators","Wave operator L^p bounds extend to 2m<n≤4m with threshold eigenvalues"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that at zero energy the only obstruction is a genuine square-integrable bound state, never a resonance; if a resonance sits at zero energy instead of or alongside the eigenvalue, the resolvent expansion and the claimed $p$-ranges would need to be replaced.","fun_headline_variants_meta":{"raw":{"variants":["Wave operators L^p-bounded despite threshold eigenvalues","New L∞ bound for wave operators when n>3","L^p bounds for wave operators: threshold eigenvalues no obstacle","Zero-energy eigenvalues no barrier to L^p wave operators","Wave operator L^p bounds extend to 2m<n≤4m with threshold eigenvalues"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.003389,"raw_usage":{"total_tokens":12858,"prompt_tokens":1135,"completion_tokens":11723,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":751,"completion_tokens_details":{"reasoning_tokens":11635}},"tokens_in":751,"tokens_out":11723,"duration_ms":87938,"temperature":1.0,"reasoning_tokens":11635,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:29:34.670366+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a real-valued potential $V$ obeying $|V(x)|\\lesssim\\langle x\\rangle^{-\\beta}$ with $\\beta>\\max(8m-n,n_\\star)+2$, $2m<n\\le 4m$, such that $H=(-\\Delta)^m+V$ has a zero-energy eigenvalue, no resonance, and no positive eigenvalue, yet $W_{low,\\kappa}$ fails to be bounded at some endpoint such as $p=n/(2m-k_0)$ for a $k_0$ satisfying the orthogonality hypotheses. Equivalently, compute the kernel of the singular part in equation (8) for a rank-one model with an explicit zero-energy eigenfunction and test whether the oscillatory integral bounds saturate at the claimed endpoint.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the Lp boundedness of wave operators with threshold eigenvalues in high dimensions n>4m, the result the present paper adapts downward."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies Proposition 2.1, the free resolvent representations of Lemmas 2.3-2.4, and the kernel bounds used to control smooth parts and the Lambda integrals."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the high-energy and Born-series Lp bounds assumed through Assumption 1.3 and the a priori bounds on iterated resolvents A(lambda)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the threshold classification via the spaces S_i, the eigenvalue/resonance distinction, and the invertibility of T_{k+1} used in Section 4."},{"cited_title":"Jensen, and G","cited_arxiv_id":null,"evidence_quote":"Gives the threshold resolvent inversion identity used to expand M(lambda)^{-1} into smooth plus singular parts."},{"cited_title":"The Lp-boundedness of wave operators for 4th order Schr¨ odinger operators on R2, I, preprint 2025, arXiv:2504.11753","cited_arxiv_id":null,"evidence_quote":"Supplies the cancellation lemma for S_{k+1}L^2 used in Lemma 3.2 to convert orthogonality into extra decay."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Handles the m=1, n=4 eigenvalue-only cases with k0=1 and k0=2, providing the comparison behind the new L-infinity endpoint."}],"review_version":1}