{"id":"de9d49c8-c87d-4622-916e-066fd7a2fb18","arxiv_id":"2507.03891","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Sharp H^s thresholds for a.e. convergence of complex-time Schrodinger operators along alpha-Holder curves are established for all alpha in (0,1), with fractional generalizations announced but not proved.","lead":"This math paper finds how smooth a starting wave must be so a damped quantum wave returns to it almost everywhere, even along wiggly paths.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Upper-bound proof is not justified at the table step: the displayed kernel bound (2.4)-(2.9) does not by itself yield the claimed I(x) bounds, and the beta1=0 row for alpha=1/2, gamma<1 produces a spurious log divergence.","rationale":"I read the paper as a proof of sharp L^2 maximal estimates for the complex-time operator along curves. The main theorems are plausible, and the counterexamples in Section 3 check out against the piecewise thresholds: each branch of (1.8) is matched by either Theorem 3.1 or Theorem 3.2. The delicate point is the passage from the TT* kernel bound (2.4) to the finite table of I(x) bounds. I verified most table rows against (2.9): with standard epsilon adjustments, the exponents in regimes such as gamma in [2alpha,1), [1,1/(2alpha)), and [1/(2alpha),2) match the square roots stated in the Propositions. However, the printed proof does not justify treating the second summand of (2.4) as a globally valid integrable bound: near x=y it is infinite, and outside the case in which it was derived, no domination argument is supplied. The row alpha=1/2, gamma in (0,1), beta1=0 makes the gap concrete, since the displayed integration yields a log lambda contribution rather than O(1). The true kernel is integrable, so this is a proof gap rather than a contradiction, but the table is effectively an unproved lemma on which the whole sufficiency argument depends. The fractional theorems in Section 4 are also stated without proof; that is a separate completeness issue, not the main load-bearing concern. The reader's conditional verdict is therefore appropriate, and my stress-test does not change it.","tokens_in":10928,"tokens_out":31663,"duration_ms":368137,"concrete_test":"Recompute I(x) from (2.9) in the row alpha=1/2, gamma in (0,1) with beta1=0, beta2=1/(2gamma): the first-term integral over lambda^{-2alpha}<=y<=1 gives log lambda, so the table's I(x)<=1 does not follow from the displayed bound. Then compute the true kernel in the case t(x)=t(y)=0 using |K| <= C_N lambda (1+lambda|x-y|)^{-N}; if this gives integral |K|dy <= 1, the row is true but the written derivation is incomplete, and the proof needs an explicit nonstationary refinement or a different beta1. Repeat the same check at the borderline values gamma=2alpha and gamma=1/(2alpha) to verify the epsilon/log bookkeeping.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The sufficiency half of The main theorems rests on the table of parameter choices in Section 2. The key estimate (2.4) is stated for all x,y, but its second summand is derived only under the case condition |x-y| lesssim lambda |t(x)-t(y)|. In passing to (2.9), this singular bound is integrated over the full interval 0<=|x-y|<=1, including regions where the first case or the nonstationary case applies, and no dominance argument is given. At x=y the second summand is infinite, so the integral must be truncated by hand. The table rows are then asserted without computation. Concretely, for alpha=1/2, gamma in (0,1), beta1=0, beta2=1/(2gamma), the first-term contribution over lambda^{-2alpha}<=|x-y|<=1 in (2.9) is integral_{lambda^{-1}}^1 dy/y, which is about log lambda, whereas the table claims I(x) <= 1. A nonstationary-phase estimate in |x-y| can rescue the row, but that missing argument is exactly what the proof needs to supply. Because each row feeds Schur's test and hence the maximal L^2 estimate, an unjustified row leaves the sufficiency statement in Propositions 2.1-2.3 and Theorem 1.3 unproved as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies almost-everywhere convergence of the complex-time Schrödinger operator P_γ f(Γ(x,t),t) along curves Γ that are bilipschitz in x and α-Hölder in t. The main results, Theorems 1.1–1.3, give a sharp Sobolev threshold s(γ) for α in [1/2,1], (0,1/4], and (1/4,1/2), respectively, with a five-regime piecewise formula (1.8) in the intermediate range. The sufficiency proofs are via a linearized maximal operator, TT* duality, pointwise kernel estimates, and Schur's test; the necessity proofs use Nikishin–Stein-type counterexamples with Γ(x,t)=x−t^α. Section 4 states seven analogous sharp results for the fractional Schrödinger operator P^m_γ, without proof.","tokens_in":11206,"tokens_out":14819,"duration_ms":157969,"significance":"If the main theorems are correct, the paper closes a gap left by Niu–Xue [8] and gives a sharp threshold for all α∈(0,1), including the previously open tangential range α∈(1/4,1/2). The explicit counterexamples in Theorems 3.1 and 3.2 are a genuine strength: they are concrete, falsifiable, and do not rely on circular reasoning. The claimed five-regime formula (1.8) is also a clean quantitative statement. However, the significance is currently limited by the fact that the upper-bound argument is not actually carried out at the load-bearing step, and the fractional results in Section 4 are asserted without proof.","major_comments":[{"comment":"The passage from the pointwise kernel bound (2.4) to the integral bound (2.9) is not justified. The first summand in (2.4) is derived under the case condition |x−y|≲|t(x)−t(y)|^α, while the third summand is derived under the case condition |x−y|≫|t(x)−t(y)|^α together with |x−y|≲λ|t(x)−t(y)|. In (2.9) these bounds are integrated over intervals that depend only on |x−y|, including the full interval 0≤|x−y|≤1, with no dominance argument showing that the pointwise bounds hold on the regions over which they are integrated. The second and third integrands are singular at x=y, so the integrals require truncation, and the displayed inequality I(x)≤... is not established as written.","section":"Section 2, Eq. (2.9)"},{"comment":"The table entry for α∈[1/2,1], γ∈(0,1) chooses β1=0, β2=1/(2γ) and claims I(x)≲1. With this choice, the middle integral in (2.9) over λ^{-2α}≤|x−y|≤1 is ∫_{λ^{-1}}^1 y^{-1}dy≈logλ when α=1/2, and the third integral has the same logarithmic behavior near y=0. Thus the claimed bound does not follow from (2.9). A nonstationary-phase estimate or a different choice of β1 (for instance β1>0 in this regime) may repair the row, but that argument is not supplied. Because every row feeds Schur's test and hence Propositions 2.1–2.3, the sufficiency halves of Theorems 1.1–1.3 are unproved as written.","section":"Section 2, table, row α=1/2, γ∈(0,1)"},{"comment":"The proof does not check the borderline parameter cases γ=2α, γ=1/(2α), and the cases where the denominator exponent in the integrals in (2.9) is exactly 1. These are precisely the regimes where logarithmic factors can appear or where the claimed λ-power estimates change. The text asserts the rows of the table without showing the required integral computations or the necessary ε-absorption arguments, so even if the main row above were fixed, the sharp threshold (1.8) would still lack a complete verification at these boundaries.","section":"Section 2, borderline cases in the table"},{"comment":"Seven sharp theorems are stated with no proof; the text explicitly says 'we will only present the results in Section 4 while omit the proof.' Since these theorems are part of the paper's claims of sharp convergence results, this is not a mere presentation issue. At minimum, the analogue of the kernel estimate (2.4), the corresponding choice table, and the counterexample constructions must be provided or the statements must be clearly marked as conjectural. As written, the fractional results cannot be checked.","section":"Section 4, Theorems 4.1–4.7"}],"minor_comments":[{"comment":"The notation line 'We write A /greaterorsimilarB to mean...' contains a corrupted LaTeX command and should read A≳B.","section":"Section 1, Notation"},{"comment":"There are several typos, including 'extentsivly' for 'extensively', 'prensent' for 'present', and 'Combing' for 'Combining' before (3.9).","section":"Page 3 and Section 4"},{"comment":"In the second half of Theorem 3.2, the choice of t_x satisfying x−t_x^α−2R^2t_x=0 makes the quadratic phase t_xR^2ξ^2 of size comparable to xξ^2; the proof should explicitly state how small the constant c is chosen so that this phase is negligible on the set B, since the current sentence 'we can find ... φ_R small enough' is too compressed.","section":"Theorem 3.2, γ∈[2,∞) case"},{"comment":"The derivation of s≥1/2−α/γ from the measure of the set A should state explicitly that |A|≳R^{-2α/γ}, since this measure is what turns the size of the set into the Sobolev exponent after the R→∞ limit.","section":"Theorem 3.1, equation (3.5)"},{"comment":"The phrase 'All results are sharp up to the endpoints' is accurate only in the sense that the theorems give s>s(γ) for sufficiency and s<s(γ) for necessity; the endpoint s=s(γ) itself is left open. The wording could be clarified to avoid suggesting endpoint sharpness.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The core idea is plausible and the necessity constructions are concrete, but the sufficiency proof has a genuine gap in Section 2 that affects the main theorems. The Section 4 fractional theorems are asserted without proof, which is too large a portion of the paper to be left unverified. I would recommend a major revision rather than rejection, because the gap appears fixable with additional kernel estimates and a detailed parameter-optimization table, but the current manuscript does not establish its central claims as written."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new result is real: Niu–Xue only covered α∈[1/2,1], and Theorems 1.2 and 1.3 fill in α≤1/4 and the delicate range 1/4<α<1/2 with a piecewise sharp s(γ). The thresholds look right, the counterexamples in Section 3 are concrete and standard, and the method is a sensible continuation of the TT* kernel approach. I do not see circularity, and the self-citation [3] is background only.\n\nThe soft spots are in the sufficiency half. The whole upper-bound proof in Section 2 leans on a table of (β1,β2) choices with no computation of the integrals. The stress-test example is fair: for α=1/2, γ∈(0,1), β1=0, the middle integral in (2.9) is ∫_{λ^{-1}}^1 dy/y ~ log λ, not O(1) as the table claims. The row is probably salvageable — choosing β1=ε>0 or using the exponential decay that (2.6) discards would fix it — but the paper does not say which, and the display as written is wrong. The same section also integrates singular kernels down to |x−y|=0 and relies on truncation by K≲λ without spelling it out. Borderline cases γ=2α, γ=1/(2α), and denominator exponent exactly 1 are not checked.\n\nThe bigger problem is Section 4. Seven fractional theorems, including piecewise formulas, are stated as sharp with only “proof method is similar” and no proof. For a research paper that is not acceptable as is; either prove them or clearly mark the section as an announcement of results to appear elsewhere.\n\nI would send this to a referee, but with instructions to demand a fully written Section 2 and a decision on Section 4. The main range α∈(0,1/2) is a genuine completion and worth defending. If the table is fixed and Section 4 is either proved or explicitly announced, I would use the main theorems.","headline":"Real new range (α<1/2) with plausible sharp thresholds, but the sufficiency proof rests on an uncomputed parameter table and Section 4 states seven fractional theorems without proof.","tokens_in":11762,"tokens_out":8102,"would_cite":true,"duration_ms":92205,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B25"],"pacs":[],"model":"deepseek-v4-flash","headline":"For curves that are bilipschitz in $x$ and $\\alpha$-Hölder in $t$, the sharp Sobolev exponent for complex-time Schrödinger convergence is an explicit piecewise function of $\\gamma$, with all results sharp up to endpoints.","keywords":["Schrödinger operator","complex time","maximal functions","pointwise convergence","curves","Sobolev spaces","oscillatory integrals","fractional Schrödinger"],"falsifier":"Compute the three integrals that make up $I(x)$ in (2.9) at $\\gamma=2\\alpha$ and $\\gamma=1/(2\\alpha)$ for $\\alpha\\in(1/4,1/2)$; if the row for $\\gamma\\in[1,1/(2\\alpha))$ gives an upper bound larger than $C\\lambda^{1-2\\alpha}$ (or with a logarithm), then the corresponding $L^2$ maximal estimate and hence the sufficiency statement would be false.","tokens_in":10726,"feed_emoji":"⚛️","tokens_out":10085,"duration_ms":99213,"temperature":0.7,"pith_summary":"This paper determines the minimal Sobolev regularity needed for the almost-everywhere pointwise convergence of the damped Schrödinger operator $P_\\gamma f(\\Gamma(x,t),t)$ to $f(x)$ as $t\\to0$, for curves $\\Gamma$ that are bilipschitz in $x$ and $\\alpha$-Hölder in $t$. For every $\\alpha\\in(0,1]$ it gives a piecewise constant exponent $s(\\gamma)$ such that convergence holds for all $f\\in H^s$ when $s>s(\\gamma)$, and fails for some admissible curve when $s<s(\\gamma)$; the results are sharp up to the endpoints. The main new contribution is the range $1/4<\\alpha<1/2$, which was previously open; the cases $0<\\alpha\\le 1/4$ and $1/2\\le\\alpha\\le 1$ are also treated, and the proof of the latter is simplified. Fractional analogues for $e^{it(-\\Delta)^{m/2}}$ and all $m>0$ are stated in the same spirit.","feed_headline":"A five-piece formula settles sharp Schrödinger convergence","feed_subtitle":"Complex-time damping improves almost-everywhere convergence; thresholds cover all Hölder orders and are sharp.","key_machinery":"The argument linearizes the maximal operator by a measurable function $t(x)$, then applies the $TT^*$ method, so the central object is the kernel $K(x,y,t(x),t(y))$, an oscillatory integral whose phase contains $\\lambda(\\Gamma(x,t(x))-\\Gamma(y,t(y)))\\xi+\\lambda^2(t(x)-t(y))|\\xi|^2$ and whose integrand contains the Gaussian factor $e^{-\\lambda^2(t(x)^\\gamma+t(y)^\\gamma)|\\xi|^2}$. Van der Corput's lemma, applied with the bilipschitz and Hölder conditions, gives the decay bound (2.4), and the Gaussian factor supplies decay in $|t(x)-t(y)|^\\gamma$ for any power $\\beta$ via $e^{-y}\\lesssim y^{-\\beta}$. The proof then optimizes a family of $\\beta_1,\\beta_2$ chosen according to the regime of $(\\alpha,\\gamma)$ to bound the integral $I(x)$ in (2.9), yielding the $L^2$ maximal estimates. The counterexamples use $\\Gamma(x,t)=x-t^\\alpha$ and frequency-localized bumps to force the lower bounds on $s$.","core_discovery":"The central claim is that the threshold for almost-everywhere convergence of (1.6) is exactly $s(\\gamma)$ given by the five-piece formula (1.8). Concretely, for $1/4<\\alpha<1/2$: convergence holds whenever $s>0$ for $\\gamma<2\\alpha$; whenever $s>\\frac12-\\frac{\\alpha}{\\gamma}$ for $2\\alpha\\le\\gamma<1$; whenever $s>\\frac12-\\alpha$ for $1\\le\\gamma<\\frac1{2\\alpha}$; whenever $s>\\frac12(1-\\frac1\\gamma)$ for $\\frac1{2\\alpha}\\le\\gamma<2$; and whenever $s>\\frac14$ for $\\gamma\\ge2$. The same statement is proved in Theorem 1.2 for $0<\\alpha\\le1/4$ with $s(\\gamma)=\\min\\{(\\frac12-\\frac{\\alpha}{\\gamma})_+,\\frac12-\\alpha\\}$, and in Theorem 1.1 for $1/2\\le\\alpha\\le1$ with $s(\\gamma)=\\min\\{\\frac12(1-\\frac1\\gamma)_+,\\frac14\\}$; each is matched by a curve for which convergence fails below the threshold. The fractional versions in Section 4 assert the analogous thresholds for all $m>0$.","pith_inferences":["A direct check of the boundary values $\\gamma=2\\alpha$, $\\gamma=1/(2\\alpha)$, and of the cases where the denominator exponent in (2.9) equals $1$, would determine whether logarithmic factors are needed; the paper does not display those computations.","The same two-term kernel estimate might extend to higher dimensions or to curves with weaker regularity, but the paper only treats $\\mathbb{R}$.","The fractional theorems in Section 4 are stated without proof; a reader who wants to use them should verify that the analogous kernel estimates survive when $|\\xi|^m$ replaces $|\\xi|^2$."],"forward_implications":["The threshold $s(\\gamma)$ for almost-everywhere convergence is now known for every Hölder exponent $\\alpha\\in(0,1]$ and every $\\gamma>0$.","The Gaussian damping $e^{-t^\\gamma|\\xi|^2}$ lowers the required Sobolev exponent relative to the undamped case in several regimes; for example, when $\\gamma<2\\alpha$ no Sobolev regularity beyond $L^2$ is needed.","The paper's fractional statements give explicit thresholds for all $m>0$, extending the range of known results for $e^{it(-\\Delta)^{m/2}}$ along curves.","The endpoint cases $s=s(\\gamma)$ remain undecided; the theorems are sharp only in the sense of $s>s(\\gamma)$ versus $s<s(\\gamma)$."],"supporting_citations":[{"why":"Proved the sharp threshold for $\\alpha\\in[1/2,1]$ that this paper extends to all $\\alpha$ and whose proof is simplified.","marker":"[8]"},{"why":"Introduced the study of the Schrödinger operator along curves and the associated kernel method.","marker":"[7]"},{"why":"Introduced the complex-time maximal operator and the almost-everywhere convergence problem in this setting.","marker":"[13]"},{"why":"Gave the sharp threshold for complex time without curves, which is the baseline case in the non-tangential regime.","marker":"[1]"},{"why":"Supplied the sharp tangential-curve result for the undamped Schrödinger operator that the complex-time result is compared with.","marker":"[4]"},{"why":"Initiated the almost-everywhere convergence problem for the Schrödinger operator.","marker":"[2]"},{"why":"Constructed the classical counterexample showing that $s=1/4$ is sharp in one dimension.","marker":"[6]"},{"why":"Provided the sharp fractional Schrödinger baseline $s\\ge1/4$ for $m>1$ used in Section 4.","marker":"[12]"}],"fun_headline_variants":["Sharp thresholds for Schrödinger with complex time","Exact convergence boundary for Schrödinger curves","Five-piece rule sets Schrödinger convergence limits","Complex-time damping yields sharp almost-everywhere results","Schrödinger convergence thresholds now exact for all orders"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The upper-bound half of the theorems rests on the table in Section 2 that chooses $(\\beta_1,\\beta_2)$ for each regime of $(\\alpha,\\gamma)$; if any row of that table is wrong, or if a borderline case such as $\\gamma=2\\alpha$ or $\\gamma=1/(2\\alpha)$ produces an extra logarithmic factor, the claimed sufficiency result fails.","fun_headline_variants_meta":{"raw":{"variants":["Sharp thresholds for Schrödinger with complex time","Exact convergence boundary for Schrödinger curves","Five-piece rule sets Schrödinger convergence limits","Complex-time damping yields sharp almost-everywhere results","Schrödinger convergence thresholds now exact for all orders"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000187,"raw_usage":{"total_tokens":1271,"prompt_tokens":832,"completion_tokens":439,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":448,"completion_tokens_details":{"reasoning_tokens":369}},"tokens_in":448,"tokens_out":439,"duration_ms":5722,"temperature":1.0,"reasoning_tokens":369,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:01:47.973082+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the three integrals that make up $I(x)$ in (2.9) at $\\gamma=2\\alpha$ and $\\gamma=1/(2\\alpha)$ for $\\alpha\\in(1/4,1/2)$; if the row for $\\gamma\\in[1,1/(2\\alpha))$ gives an upper bound larger than $C\\lambda^{1-2\\alpha}$ (or with a logarithm), then the corresponding $L^2$ maximal estimate and hence the sufficiency statement would be false.","supporting_citations":[{"cited_title":"Estimates for Schr¨ odinger maximal ope rators along curve with complex time","cited_arxiv_id":null,"evidence_quote":"Proved the sharp threshold for $\\alpha\\in[1/2,1]$ that this paper extends to all $\\alpha$ and whose proof is simplified."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced the study of the Schrödinger operator along curves and the associated kernel method."},{"cited_title":"Maximal operators of Schr¨ odinger type with a com plex parameter","cited_arxiv_id":null,"evidence_quote":"Introduced the complex-time maximal operator and the almost-everywhere convergence problem in this setting."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gave the sharp threshold for complex time without curves, which is the baseline case in the non-tangential regime."},{"cited_title":"Some analytic problems related to statistical mechanics","cited_arxiv_id":null,"evidence_quote":"Initiated the almost-everywhere convergence problem for the Schrödinger operator."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Constructed the classical counterexample showing that $s=1/4$ is sharp in one dimension."},{"cited_title":"Regularity of solutions to the Schr¨ odinger equation","cited_arxiv_id":null,"evidence_quote":"Provided the sharp fractional Schrödinger baseline $s\\ge1/4$ for $m>1$ used in Section 4."}],"review_version":1}