{"id":"ba50a8f8-cfb7-47fe-8e2b-7d855a9e7975","arxiv_id":"1908.08358","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For Schrödinger equations with sub-quadratically perturbed harmonic oscillators, the wave front set at any time is characterized by high-frequency decay of the wave packet transform of the initial data along the scaled classical flow.","lead":"This paper shows where the sharp, singular features of a quantum wave go when the wave evolves in a harmonic trap with a weaker extra force. It proves that these singularities travel along very high-energy classical orbits, a partial confirmation of a 1995 conjecture by Yajima.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The unproved reduction to 0<t0≤π is load-bearing: Lemmas 4.1–4.2 only cover the half-period t0−π≤s≤t0, so the induction P(σ) does not establish Theorem 1.2 for |t0|>π without an additional argument.","rationale":"The reader's weakest assumption is exactly the point I would stress. The central claim is Theorem 1.2, and its proof rests on the induction P(σ) in Section 5, which relies on Lemmas 4.1 and 4.2. Those lemmas are explicitly proved only for 0<t0≤π, and the proof of Lemma 4.1 uses the fact that [0,t0] lies in the single half-period [t0−π,t0]. For larger t0 the displayed estimates do not cover the initial part of the backward flow. Since the equation is not time-periodic and the perturbation v is time-dependent, one cannot simply translate time; any repair would require a separate argument, most likely an induction over π-length intervals. The rest of the proof is detailed and the estimates inside the half-period appear sound, which is why I regard this as a conditional-accept issue rather than a rejection: the result is plausible and likely fixable, but the written proof does not establish the full statement. I agree with the reader's assessment and recommend CONDITIONAL.","tokens_in":16402,"tokens_out":9796,"duration_ms":107204,"concrete_test":"Re-derive Lemma 4.1 with t0=3π/2, keeping the same interval decomposition. The split after (19)–(20) covers only −π≤s−t0≤0, i.e. π/2≤s≤3π/2; isolate the untreated range 0≤s<π/2 and check whether the bound (17) follows there with the same δ. If it does, state the missing estimate; if not, exhibit a v and u satisfying the lemma's hypothesis for which (17) fails in that range, or show that the induction argument cannot proceed. This directly settles whether the WLOG reduction is valid.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 5 begins 'We may assume without loss of generality that 0<t0≤π', and Lemmas 4.1 and 4.2 are proved only under that assumption. This is not a harmless normalization: v(t,x) is time-dependent, so the equation has no translation or periodicity invariance, and shifting the initial time would replace u0 by u(t0−mπ), not the data of the theorem. The proof genuinely uses the restriction. In Lemma 4.1, after (19) the estimate (20) for |x(s)| is obtained by splitting into −π+λ^{−2b}≤s−t0≤−λ^{−2b} and |sin(s−t0)|≤λ^{−2b}; this dichotomy only exhausts t0−π≤s≤t0. For t0>π the remaining interval 0≤s<t0−π is never treated, and the displayed bound (17) is not proved there. The induction P(σ) in Section 5 needs (16)–(17) for all 0≤s≤t0; since Lemmas 4.1–4.2 stop at one half-period, the step P(σ)→P(σ+δ) fails for |t0|>π. The same problem occurs for t0<−π by symmetry. Consequently Theorem 1.2, stated for every t0∈R, is not established as written for times beyond the first half-period, including t0=mπ with |m|≥2. The gap may be repairable by an induction over intervals of length π using the theorem on shifted equations, but the paper does not supply that argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the Schrödinger equation with a harmonic oscillator potential plus a time-dependent sub-quadratic perturbation v(t,x) satisfying Assumption 1.1, i.e. |∂_x^α v(t,x)| ≤ C_α (1+|x|)^{ρ−|α|} with 0 ≤ ρ < 2. Using a time-dependent wave packet transform adapted to the harmonic oscillator propagator, the authors derive an integral equation (14) for the transformed solution and prove Theorem 1.2: for 0 < b < min(1/2, (2−ρ)/2), a point (x0, ξ0) is not in WF(u(t0,·)) if and only if the wave packet transform of the initial data, W_{φ_λ(−t0)}u0, evaluated along the backward classical flow (3), decays like λ^{−N} uniformly in a neighborhood of x0 and a conic neighborhood of ξ0. The proof is a bootstrap P(σ) with exponent increment δ = min(2−2b−ρ, 2b), using Lemmas 4.1 and 4.2 to control the Taylor remainders of the perturbation. Corollaries treat the cases ρ < 1, ρ = 1, and recover Yajima's smoothing result at t = π.","tokens_in":16760,"tokens_out":22205,"duration_ms":216411,"significance":"If Theorem 1.2 is fully established, the paper is a substantial contribution: it determines the wave front set of solutions at all times, including the caustic times t = mπ, for time-dependent sub-quadratic perturbations of the harmonic oscillator. This goes beyond earlier results by Mao and Nakamura, Mao, and Okaji, and is directly aligned with Yajima's conjecture on propagation of singularities along infinite-energy classical orbits. The proof strategy is elegant: the time-dependent wave packet transform converts the second-order equation into a first-order transport equation plus remainder, and the bootstrap exponent δ is explicit. The argument uses no fitted constants and rests on the standard Folland characterization of wave front sets. The paper is concise and mostly well organized. However, the full-time claim is currently not proved because of an unjustified reduction to 0 < t0 ≤ π in Section 5.","major_comments":[{"comment":"The reduction 'We may assume without loss of generality that 0 < t0 ≤ π' at the start of the proof of Theorem 1.2 is not justified and is load-bearing. Since v(t,x) in (1) is time-dependent, the equation has neither time-translation nor periodicity symmetry; shifting the time origin would replace u0 by a different Cauchy datum. Lemmas 4.1 and 4.2 are proved only under this restriction, and their proofs use a dichotomy covering s−t0 ∈ [−π,0]: the estimates (21)–(22) treat −π+λ^{−2b} ≤ s−t0 ≤ −λ^{−2b} and the two subintervals of [−π,0] where |sin(s−t0)| ≤ λ^{−2b}. For t0 > π the segment 0 ≤ s ≤ t0−π, which contains s = 0 when t0 = mπ with m ≥ 2, is never estimated; for t0 < 0 the interval [t0,0] is not treated at all. The same defect appears in the proof of Lemma 4.2. The bootstrap P(σ) in §5 needs the bounds (16)–(17) and (23) on the full interval, so the step P(σ) → P(σ+δ) does not go through for |t0| > π. Consequently Theorem 1.2, stated for every t0 ∈ R, is not established as written beyond one half-period, and the (i)⇒(ii) direction, said to follow 'in the same way', inherits the same gap. The problem may be repairable, for example by an induction over intervals of length π, but the needed argument is not in the manuscript.","section":"§5 (Proof of Theorem 1.2); Lemmas 4.1–4.2"}],"minor_comments":[{"comment":"In the estimate for I_{α,2}, the text says 'Hence (34), (31), (32) and Scwartz's inequality shows...', but (34) is the estimate being proved; the intended reference is likely (30).","section":"Lemma 4.2"},{"comment":"Equation (41) reads 'ξ(s) = (s;π,x,λξ)'; the function symbol ξ is missing.","section":"Corollary 1.7"},{"comment":"Remark 1.3 asserts that the theorem remains valid for u0 ∈ H^{−s} and u ∈ C(R;H^{−s}) without giving a proof or a reference; as stated this extension is unsupported.","section":"Remark 1.3"},{"comment":"In the proof of Theorem 1.8, the displayed conclusion (50) says |x(0;π,x,λξ)| = O(λ^{ρ−1}), but the preceding lower bound on ⟨x(0),ξ⟩ gives a lower bound of order λ^{ρ−1}, not an upper bound; the argument needs |x(0)| ≳ λ^{ρ−1} to obtain the decay in (46).","section":"Theorem 1.8"},{"comment":"There are many typographical errors, including 'Scwartz' after (31), 'Prinsto n' in reference [6], and inconsistent umlauts in 'Schrödinger'; a careful copyedit is needed.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The half-period gap is the only substantive issue I found; the rest of the proof is coherent, and the paper would fit the journal once the full-time statement is either proved or appropriately restricted. I would not require a complete reworking if the authors can supply the missing interval argument, but the current WLOG is not a harmless normalization and must be addressed head-on."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe thing to know: this paper has the right result but a load-bearing gap in the proof. It determines the wave front set of solutions to Schrödinger equations with sub-quadratic perturbations including the singular times t=mπ, which extends Mao's earlier work and recovers Yajima's theorem as a corollary. That part is genuine and worth having.\n\nWhat it does well: the wave-packet-transform representation from the authors' earlier work is applied cleanly, and the bootstrap induction P(σ) in Section 5 is coherent. Lemmas 4.1 and 4.2 are the technical heart, and the estimates λ^{−σ−δ} and λ^{−N} with δ=min(2−2b−ρ,2b) are worked out in detail. I see no circularity or fitted constants. The result is a solid extension rather than a new method, and it is honestly positioned against the literature.\n\nWhere it goes wrong: the proof begins with 'We may assume without loss of generality that 0<t0≤π' and never justifies it. That is not a harmless normalization. The potential v(t,x) is time-dependent, so there is no translation or periodicity invariance to shift the initial time. Lemmas 4.1 and 4.2 are proved only on a half-period [t0−π,t0], and the induction P(σ) in Section 5 needs the estimates for all 0≤s≤t0. For |t0|>π the interval [0,t0−π] is simply not treated. So Theorem 1.2, stated for every t0∈R, is not established by the proof as written. The gap is repairable by an induction over intervals of length π, but the paper does not supply that argument. The converse direction (i)⇒(ii) is also only sketched, which is a minor but real omission.\n\nWho this is for: people working on microlocal analysis of Schrödinger equations. It deserves a serious referee, not a desk reject. The referee should ask for the t0 reduction to be justified or the statement restricted, and for the converse to be written out.\n\nRegards,","headline":"Solid extension to singular times with a real but repairable proof gap: the reduction to 0<t0≤π is unjustified and the theorem as stated for all t0 is not proved.","tokens_in":17308,"tokens_out":3862,"would_cite":true,"duration_ms":31062,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q41","35A18","35S30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A point is outside the wave front set at time $t_0$ exactly when the initial wave packet transform decays faster than any power of $\\lambda$ along the backward classical flow, including the singular times $t_0=m\\pi$.","keywords":["wave front set","Schrödinger equation","harmonic oscillator","sub-quadratic perturbation","wave packet transform","singularity propagation","classical flow","microlocal analysis"],"falsifier":"For the exactly solvable case $v(t,x)=q(t)\\cdot x$ with $\\rho=0$, evaluate both sides of Theorem 1.2 at $t_0=2\\pi$ for an initial datum with a known singularity. Since the propagator is explicit, this is a direct calculation: if the $\\lambda^{-N}$ estimate fails at a point that is smooth, or holds at a point that is singular, the claimed equivalence is refuted; if it holds, the unproved restriction to $0<t_0\\le\\pi$ is shown not to hide a counterexample.","tokens_in":16177,"feed_emoji":"⚛️","tokens_out":12338,"duration_ms":110241,"temperature":0.7,"pith_summary":"This paper aims to determine, for the Schrödinger equation with a harmonic oscillator plus a sub-quadratic potential $v(t,x)$, exactly where solutions are singular at every time. It claims that a phase-space point $(x_0,\\xi_0)$ is outside the wave front set of $u(t_0,\\cdot)$ precisely when the wave packet transform of the initial data, evaluated along the backward classical flow of the perturbed oscillator, decays faster than any power of the scaling parameter $\\lambda$. The characterization is uniform in a neighborhood of $x_0$ and a conic neighborhood of $\\xi_0$, and it covers the singular times $t_0=m\\pi$ where the unperturbed oscillator refocuses singularities. A sympathetic reader would care because the result turns a question about solutions of a PDE into a checkable decay estimate on the initial data, and it gives a concrete sense in which singularities travel along limits of classical orbits as the energy tends to infinity.","feed_headline":"Wave packet decay determines wave front sets at every time","feed_subtitle":"For sub-quadratic perturbations, singularities move along classical-orbit limits, including the times t = mπ.","key_machinery":"The central object is the wave packet transform $W_{\\phi_\\lambda(t)}u(t,x,\\xi)=\\int \\phi_\\lambda(t,y-x)u(t,y)e^{-iy\\cdot\\xi}\\,dy$, where the wave packet $\\phi_\\lambda(t)=e^{\\frac i2 t(\\triangle-|x|^2)}\\phi_{0,\\lambda}$ is itself evolved by the unperturbed harmonic oscillator. Under this transform, the equation becomes a first-order transport equation in the phase-space variables $(x,\\xi)$, and the method of characteristics converts it into the integral equation (14), whose leading term is the wave packet transform of the initial data evaluated on the backward orbit. The Taylor expansion of $v$ around the moving center $x(s)$ splits the interaction into a linear part absorbed into the flow and remainders of order $|\\alpha|\\ge2$; Lemmas 4.1 and 4.2 show these remainders decay like $\\lambda^{-\\sigma-\\delta}$, with $\\delta=\\min(2-2b-\\rho,2b)$. An induction on $\\sigma$ then lifts the assumed decay of the initial data to decay of the evolved wave packet transform at every time up to $t_0$.","core_discovery":"Under Assumption 1.1, for any $0<b<\\min(1/2,(2-\\rho)/2)$, Theorem 1.2 states that $(x_0,\\xi_0)\\notin WF(u(t_0,\\cdot))$ is equivalent to the estimate $$|W_{\\phi_\\$\\lambda$(-t_0)}u_0(x(0;t_0,x,\\$\\lambda$\\xi),\\xi(0;t_0,x,\\$\\lambda$\\xi))|\\le C_{N,a,\\phi_0}\\$lambda^{{-N}}$$$ holding for all $N$, uniformly for $x$ in a neighborhood $K$ of $x_0$ and $\\xi$ in a conic neighborhood $\\Gamma$ of $\\xi_0$ with $a^{-1}\\le|\\xi|\\le a$. Here $(x(s),\\xi(s))$ solves $\\dot x=\\xi$, $\\dot\\xi=-x-\\nabla v(s,x)$, with final data $x(t_0)=x$, $\\xi(t_0)=\\lambda\\xi$. The equivalence holds for every real $t_0$, including the singular times $t_0=m\\pi$. The proof transforms the equation with time-dependent wave packets evolved by the harmonic oscillator, writes the solution as an integral equation along the classical flow, and shows that all Taylor-remainder terms of the perturbation gain powers of $\\lambda$.","pith_inferences":["If the unproved reduction to $0<t_0\\le\\pi$ is supplied, the same transport-and-remainder scheme should also work for time-periodic or almost-periodic subquadratic perturbations, since only the Taylor remainder estimates use the specific form of Assumption 1.1.","The explicit formula in Corollary 1.6 suggests a numerical way to locate singular directions: compute the backward flow and measure the decay rate of the wave packet transform; directions whose decay is slower than every power are candidates for the wave front set.","The gain exponent $\\delta=\\min(2-2b-\\rho,2b)$ indicates that the optimal choice of the scaling parameter $b$ may sharpen the result to Sobolev-type wave front sets rather than $C^\\infty$ wave front sets."],"forward_implications":["For $0\\le\\rho<1$, the wave front set of the perturbed solution coincides with the wave front set of the pure harmonic oscillator with the same initial data (Corollary 1.6).","At the singular time $t_0=\\pi$ and for $\\rho=1$ with an asymptotically homogeneous gradient, a point outside $WF(u(\\pi,\\cdot))$ forces an explicitly shifted point outside $WF(u_0)$ (Corollary 1.7).","For $1<\\rho<2$ and a positive definite Hessian, the theorem recovers the known result that the fundamental solution is smooth at $t=\\pi$ (Theorem 1.8).","The equivalence supplies a uniform test for membership in the wave front set at any fixed time: integrate the classical flow backward from $(x,\\lambda\\xi)$ and check whether the wave packet transform of $u_0$ decays like $\\lambda^{-N}$ for all $N$."],"supporting_citations":[{"why":"supplies the representation of the Schrödinger evolution operator through the wave packet transform that underlies the transformed equations.","marker":"[12]"},{"why":"provides the wave-packet characterization of the wave front set used in Proposition 2.2 to convert decay estimates into microlocal regularity.","marker":"[6]"},{"why":"gives the inversion formula and time-frequency analysis background needed to control remainder terms.","marker":"[8]"},{"why":"introduces wave packets as a phase-space tool for partial differential equations, the transform on which the proof is built.","marker":"[1]"},{"why":"earlier determination of wave front sets for the free particle and harmonic oscillator that the present theorem extends.","marker":"[13]"},{"why":"earlier wave-packet treatment of sub-quadratic potentials that supplies the Taylor-expansion remainder strategy.","marker":"[15]"},{"why":"determines wave front sets for $\\rho<1$; the present theorem generalizes this to all admissible $\\rho<2$ and all times.","marker":"[19]"},{"why":"determines wave front sets for $\\rho<2$ away from $t=m\\pi$; the present theorem removes that exclusion.","marker":"[18]"},{"why":"is the theorem on smoothness of the fundamental solution at $t=\\pi$ recovered here as a corollary.","marker":"[29]"}],"fun_headline_variants":["Wave packet decay determines singularities at all times","Including t=mπ: wave front sets from decay estimates","Subquadratic perturbation: wave fronts follow classical orbits","All-time wave front characterization for perturbed oscillators","Wave packet decay locates singularities at every time"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes without explanation that the final time lies between 0 and $\\pi$, even though the theorem is stated for every real time, and the lemmas and induction are written only for that interval; the full claim therefore rests on this unproved reduction.","fun_headline_variants_meta":{"raw":{"variants":["Wave packet decay determines singularities at all times","Including t=mπ: wave front sets from decay estimates","Subquadratic perturbation: wave fronts follow classical orbits","All-time wave front characterization for perturbed oscillators","Wave packet decay locates singularities at every time"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000664,"raw_usage":{"total_tokens":2983,"prompt_tokens":849,"completion_tokens":2134,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":2039}},"tokens_in":465,"tokens_out":2134,"duration_ms":17255,"temperature":1.0,"reasoning_tokens":2039,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:42:26.671762+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For the exactly solvable case $v(t,x)=q(t)\\cdot x$ with $\\rho=0$, evaluate both sides of Theorem 1.2 at $t_0=2\\pi$ for an initial datum with a known singularity. Since the propagator is explicit, this is a direct calculation: if the $\\lambda^{-N}$ estimate fails at a point that is smooth, or holds at a point that is singular, the claimed equivalence is refuted; if it holds, the unproved restriction to $0<t_0\\le\\pi$ is shown not to hide a counterexample.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the representation of the Schrödinger evolution operator through the wave packet transform that underlies the transformed equations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the wave-packet characterization of the wave front set used in Proposition 2.2 to convert decay estimates into microlocal regularity."},{"cited_title":"Gr¨ ochenig, Foundations of Time-Frequency Analysis , Birkh¨ auser, Boston, 2001","cited_arxiv_id":null,"evidence_quote":"gives the inversion formula and time-frequency analysis background needed to control remainder terms."},{"cited_title":"C´ ordoba and C","cited_arxiv_id":null,"evidence_quote":"introduces wave packets as a phase-space tool for partial differential equations, the transform on which the proof is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"earlier determination of wave front sets for the free particle and harmonic oscillator that the present theorem extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"earlier wave-packet treatment of sub-quadratic potentials that supplies the Taylor-expansion remainder strategy."},{"cited_title":"Mao and S","cited_arxiv_id":null,"evidence_quote":"determines wave front sets for $\\rho<1$; the present theorem generalizes this to all admissible $\\rho<2$ and all times."},{"cited_title":"Mao, Wave front set for solutions to Schr¨ odinger equations with long-range pertubed harmonic oscillators J","cited_arxiv_id":null,"evidence_quote":"determines wave front sets for $\\rho<2$ away from $t=m\\pi$; the present theorem removes that exclusion."},{"cited_title":"Yajima, On fundamental solution of time dependent Sch¨ odinger equa tions , Cotemporary Math","cited_arxiv_id":null,"evidence_quote":"is the theorem on smoothness of the fundamental solution at $t=\\pi$ recovered here as a corollary."}],"review_version":1}