{"id":"25a1b3d8-ab91-46a1-9d33-dd68d9d89cc9","arxiv_id":"1908.11627","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For non-integrable NLS on R, the paper proves existence of space-time quasi-periodic, infinite-energy, smooth solutions with two frequencies for a large Cantor set of parameters.","lead":"This paper constructs smooth, infinite-energy solutions to nonlinear Schrödinger equations on the entire real line, quasi-periodic in both space and time and free of spatial symmetry. It is the first existence proof of this kind for non-integrable equations, extending Bourgain's semi-algebraic method to a non-compact setting.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The decisive Green's-function estimates in region (i) are imported from [W5] via an asserted 'same setting' claim; if the bi-rational, two-frequency adaptation fails, estimates (2.47)-(2.48) and the Newton scheme collapse.","rationale":"I read the paper as a serious extension of Bourgain's semi-algebraic method to space-time quasi-periodic solutions on R, with a genuinely new bi-rational approximation step in §3.4. The statement is precise, the measure estimates are standard in form, and the Newton framework is coherent. The single load-bearing weakness is exactly what the reader identified: the linear estimates in region (i), where |θ| and |φ| are polynomial in log N, are not proved in this paper. Section 2.8 imports the Main Lemma from [W5] by asserting that the setting is the same, but the present operator has a ±(n·ω+θ) term and the parameters are constrained by bi-rational maps. These are nontrivial differences: the W5 lemma was for standing waves, and the proof there did not need to control the composition with M_N or the ±-sector asymmetry. If the adaptation fails, the Green's function estimates and hence the Newton scheme collapse. I see no internal contradiction elsewhere, and I do not believe the result is false; rather, the proof as written is not self-contained at the critical point. The reader's CONDITIONAL verdict is appropriate, so I recommend no change. A complete verification would require a self-contained proof of the Main Lemma in the bi-rational, two-frequency setting, or a precise theorem-by-theorem reduction to [W5] with all hypotheses checked.","tokens_in":29966,"tokens_out":7456,"duration_ms":74429,"concrete_test":"Independently re-derive the Main Lemma in the present setting: take T_N(λ,m,M) from (2.8)-(2.10), double the auxiliary variables as in §2.8, and run the three-scale argument of [W5] with bi-rational maps M_N satisfying (2.38)-(2.41). In particular, write out every application of Lemmas 2.2-2.4 for the bad sets Θ_N arising from |±(n·ω+θ)+M+(j·λ+φ+m)^2| < δ^{2p-1}, and check that Lemma 2.4's hypotheses hold with n=4, d=17 and the degree bounds (2.41) after composing with M_N. If any step fails or requires extra hypotheses, such as absence of the ±n term or exact rather than approximate bi-rationality, the theorem is not proved from this paper plus [W5] as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The theorem rests on the Main Lemma in §2.6, whose region-(i) proof is the core of the linear analysis. In §2.8, after doubling (θ,φ) to (θ̃,φ̃), the text says: 'In the (λ,ω)-coordinates, we are now in the same setting as the Main Lemma in [W5]. The Lemma and the proof there apply.' No proof of this adaptation is given. The [W5] lemma was proved for standing waves, without the ±(n·ω+θ) term and without a family of bi-rational maps M_N relating (λ,m,M) to (λ,ω); the present operator (2.14)-(2.16) has both features. Property B and Property V3 are assumed rather than verified for the actual Newton iterates, apart from the approximate rational-approximation statement in Lemma 3.3. If the degree/determinant bounds (2.38)-(2.41) are not preserved under composition with M_N^{-1}, or if the covariance structure used in Lemmas 2.3-2.4 fails because of the ±n term, the Green's function bounds (2.47)-(2.48) cannot be established. Since the Newton iteration at every scale N=A^r inverts T_N using those bounds, the whole construction is unsupported at that point. The paper explicitly says it does not repeat the proof, so this is a verification gap rather than an internal contradiction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs smooth, infinite-energy, space-time quasi-periodic solutions of the nonlinear Schrödinger equation i∂_t u = −∂_x^2 u + |u|^{2p}u on R for every integer p ≥ 1. The Ansatz is a two-frequency time and two-frequency space Fourier series of the form (A); the parameters are the spatial frequencies λ ∈ (0,2π)^2 and the phases (m,M), while the time frequencies ω are determined by four Q-equations. The main theorem asserts that for non-parallel h1,h2 ∈ Z^2 and sufficiently small fixed amplitudes a, there is a Cantor set G ⊂ (0,2π)^4 of measure at least (2π)^4(1−|log δ|^{−1/2}) such that the P-equations can be solved by a Newton scheme, yielding a solution close to the linear two-wave solution with exponentially decaying Fourier coefficients. The proof combines a Lyapunov–Schmidt decomposition, Green's function estimates for truncated linearized operators, and semi-algebraic set exclusions; the nonlinear analysis is carried out in Section 3.","tokens_in":30178,"tokens_out":7128,"duration_ms":67261,"significance":"If the main theorem is correct, it represents a notable advance: it provides global, non-decaying, spatially non-symmetric quasi-periodic solutions for a non-integrable NLS on the line, in a setting where the linearized operator is non-elliptic and the problem is non-compact. The manuscript also offers a serious extension of Bourgain's semi-algebraic method beyond the compact/lattice setting. There are genuine contributions in the paper: Lemma 3.1 gives a self-contained proof of nonvanishing of the relevant diagonal matrix elements under h1 ∦ h2; Lemma 3.3 constructs rational approximations of the inverse frequency map with degree control of the form e^{C(log N)^3}, which is the scale required by Property B; and Section 3 contains a coherent induction with explicit scales and measure losses. The main limitation is that the decisive linear estimate in region (i) of the Main Lemma is imported from the author's previous work [W5], and the adaptation to the present more general operator is asserted rather than proved. The significance of the result is therefore conditional on completing that verification.","major_comments":[{"comment":"The proof of the Main Lemma in region (i) is not contained in this paper. After doubling (θ,φ) to (θ̃,φ̃), the text states: 'In the (λ,ω)-coordinates, we are now in the same setting as the Main Lemma in [W5]. The Lemma and the proof there apply.' According to the manuscript's own description, [W5] treated space quasi-periodic standing waves, i.e., time-independent solutions, with no term ±(n·ω+θ) in the diagonal operator (2.14)-(2.16) and no family of bi-rational maps M_N. The present operator has both features, and the covariance structure is altered by the doubling. Since estimates (2.47)-(2.48) are the input for the Newton step (3.53), the paper must either provide a proof of region (i) in the present setting or cite a version of [W5]'s Main Lemma that explicitly covers two-frequency operators with ±(n·ω+θ) and bi-rational conjugacies, and explain why Lemmas 2.3-2.4 and the Yomdin-Gromov triangulation argument remain valid after this adaptation.","section":"§2.8"},{"comment":"Properties B and V3 are assumptions in the Main Lemma, but their verification for the actual Newton iterates is incomplete. Lemma 3.3 constructs rational maps μ_r satisfying μ_r M_r = I + O(δ_r^2) and M_r μ_r = I + O(δ_r^2), but the Main Lemma requires genuine bi-rational maps M_N with exact inverse M_N^{-1} and the determinant and degree bounds (2.38)-(2.41), as well as deg(w_N∘M_N^{-1}) ≲ e^{(log N)^3} in Property V3. It is not shown that the approximate inverse supplied by Lemma 3.3 can be replaced by an exact rational inverse without changing the Green's function estimates, nor that the degree bounds remain uniform in N after all Newton corrections. This is load-bearing because the Main Lemma is stated for TN(λ,ω;u_N,v_N) and TN(λ,m,M;u_N,v_N) after conjugation by M_N^{-1}.","section":"§2.6 and §3.4"},{"comment":"The Diophantine conditions (Hiv,b2) are only verified for finite frequency ranges j,n ∈ [−A^r,A^r]^2 at each step, while Lemma B in §2.4 assumes global conditions ‖j·λ‖_T ≥ 1/(|log δ||j|^ρ) and ‖n·ω‖_T ≥ 1/(|log δ|^2|n|^ρ) for all nonzero j,n. The transition from finite-range to all-scales Diophantine control, including the dependence of ω^{(r)} on the Newton step through (3.1)-(3.2), is only sketched in the measure estimate (2.49). The authors should spell out the total measure loss after intersecting the exclusions over all scales and show explicitly that it is compatible with the claimed measure bound (1.2).","section":"§3.3 and §2.4"}],"minor_comments":[{"comment":"In the definition of a bi-rational map, \"an open set I ∈ R^d\" should read \"an open set I ⊂ R^d\".","section":"§2.5"},{"comment":"In the statement of the Main Lemma, the notation is inconsistent: G_N is said to be a semi-algebraic set in (λ,ω), but is also written as G_N ⊂ I, where I is an interval in (λ,m,M); the relation Γ_N = M_N^{-1}(G_N) suggests G_N should be a subset of the image M_N(I). This should be clarified.","section":"§2.6"},{"comment":"The convolution notation (u∗v)^{∗p} is used repeatedly but never defined explicitly; a combinatorial definition would help the reader verify the form of the linearized operator H in (2.10).","section":"§2.1"},{"comment":"In the passage deriving sectional measures after doubling, the assertion that \"the set S is independent of θ̃_1−θ̃_2\" is not justified; if this independence is needed for the measure comparison, it should be proved.","section":"§2.8"},{"comment":"Lemma A states the estimate (2.22) with exponents σ > τ > 0, but the proof requires additional conditions 0 < sτ < 1 and sσ > 1 relating s and the initial scales (2.18); these hypotheses should be included in the lemma statement.","section":"§2.3"},{"comment":"The theorem states that there is a diffeomorphism (λ,m,M) ↦ (λ,ω) on (0,2π)^4, but the map is singular on the set (h1−h2)·λ = 0, as shown in (3.26) and (3.30); the statement should either restrict the domain or explicitly include this exclusion in the definition of G.","section":"Theorem, §1"}],"recommendation":"major_revision","confidential_remarks":"The paper depends heavily on the author's own works [W4] and especially [W5]; the latter is listed as 'to appear' and the decisive region-(i) proof is not reproduced. This creates both a technical verification gap and an accessibility concern for referees and readers. The core issue is not a demonstrated internal contradiction, but a missing proof of an adaptation that is asserted to be routine. If the author can supply the missing proof or a precise reference covering the two-frequency bi-rational setting, the paper could become acceptable. Given the foundational role of the Main Lemma, I recommend major revision rather than rejection at this stage."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper you asked about is a serious piece. It proves existence of smooth, infinite-energy, space-time quasi-periodic solutions (two frequencies each) for the non-integrable NLS on R, for any p≥1. That is new: prior constructions were either space-periodic (Bourgain, Wang) or relied on integrability (Damanik-Goldstein, Boutet de Monvel-Egorova). The method is a real extension of Bourgain's semi-algebraic set method to a non-compact, two-frequency setting, and the overall architecture—Lyapunov-Schmidt splitting, Newton iteration, Green's function estimates—is coherent.\n\nThe soft spot is exactly where the reader says it is. The central Main Lemma in Section 2.6, which gives the Green's function estimates in region (i), is not proved in this paper. In Section 2.8 the text says that after changing to the (λ,ω) coordinates, \"we are now in the same setting as the Main Lemma in [W5]\" and the proof there applies. But the operator here has the ±(n·ω+θ) term and a family of bi-rational coordinate changes, neither of which appears in [W5]. Properties B and V3 are assumed, not verified in detail for the actual Newton iterates; the verification in Section 3.4 is plausible but terse. If the adaptation fails, the Green's function bounds (2.47)–(2.48) collapse and so does the Newton construction.\n\nThat said, I don't see an internal contradiction or a circular argument. The proof does not assume the conclusion. The dependence on [W5] is a verification gap, not a sleight of hand. It is common in this area to cite technical lemmas from earlier papers, and the author is the one who proved [W5], so the referral is not lazy. But for a result of this significance, particularly with the two-frequency non-compact twist, the referee should insist on a self-contained proof of the Main Lemma, or at minimum a precise statement of what in [W5] is being used and why the bi-rational map preserves the required degree and determinant bounds.\n\nWho is this for? People working in KAM theory for PDEs, quasi-periodic solutions, and Anderson localization in the nonlinear setting. It deserves a serious referee. I would condition acceptance on the author closing the gap, but the question is worth refereeing, not desk rejecting.","headline":"New construction of infinite-energy space-time quasi-periodic solutions for NLS on R; the key semi-algebraic lemma is imported from the author's prior work, leaving a verification gap that needs closing.","tokens_in":30807,"tokens_out":2487,"would_cite":true,"duration_ms":22398,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55","35B15","37K55"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs smooth infinite-energy, space-time quasi-periodic solutions to nonlinear Schrödinger equations on the real line.","keywords":["nonlinear Schrödinger equation","quasi-periodic solutions","infinite energy","space-time quasi-periodic","Newton scheme","Anderson localization","semi-algebraic geometry","small divisors"],"falsifier":"Compute, for a single large scale $N$ in region (i), the exceptional set $\\Theta_N$ arising from the local Lipschitz eigenvalue curves after composing with the bi-rational map $M_N$; if its sectional measure in $\\theta$ or $\\varphi$ exceeds $e^{-N^\\tau}$ or if a bad box remains at distance at least $N/10$, then the estimates (2.47)-(2.48) fail and the Newton construction cannot be continued.","tokens_in":29670,"feed_emoji":"🌊","tokens_out":6822,"duration_ms":61609,"temperature":0.7,"pith_summary":"The paper seeks smooth global solutions to the nonlinear Schrödinger equation $i\\partial_t u = -\\partial_x^2 u + |u|^{2p}u$ on $\\mathbb R$ that are quasi-periodic in both time and space, with two independent frequencies in each direction. Such solutions do not decay as $|x|\\to\\infty$, so they carry infinite energy and mass, and they have no spatial symmetry. Earlier constructions of this type either required integrability, an exponentially decaying nonlinearity, or a compact spatial domain. The paper claims that for every integer $p\\ge 1$ and for a parameter set whose measure is close to full, these infinite-energy solutions exist as convergent series with exponentially decaying Fourier coefficients.","feed_headline":"Infinite-energy quasi-periodic waves solve NLS on the real line","feed_subtitle":"Smooth nonlocalized solutions with two space and two time frequencies exist for every power nonlinearity.","key_machinery":"The main mechanism is a Newton scheme built on a Lyapunov-Schmidt decomposition, together with Green's function estimates for the linearized operator $F'(\\theta,\\varphi)=D(\\theta,\\varphi)+\\delta^{2p}H$, where $D_{\\pm}=\\mathrm{diag}[\\pm(n\\cdot\\omega+\\theta)+M+(j\\cdot\\lambda+\\varphi+m)^2]$. The auxiliary pair $(\\theta,\\varphi)$ cannot be bounded by the spatial scale $N$, so the proof splits into a large-$(\\theta,\\varphi)$ region, handled by Lipschitz eigenvalue variation, and a small-$(\\theta,\\varphi)$ region, handled by semi-algebraic geometry. A key new ingredient is the use of bi-rational maps $(\\lambda,m,M)\\leftrightarrow(\\lambda,\\omega)$ that approximate the diffeomorphism defined by the Q-equations, allowing the algebraic and measure estimates to be transferred back to the original parameters.","core_discovery":"The central claim is that for any $p\\ge 1$ and any non-parallel $h_1,h_2\\in\\mathbb Z^2$, there exists $\\delta_0>0$ such that for all $0<\\delta<\\delta_0$ and fixed $a\\in(0,\\delta)^2$, there is a Cantor set $G\\subset(0,2\\pi)^4$ with relative measure at least $1-|\\log\\delta|^{-1/2}$ and a diffeomorphism $(\\lambda,m,M)\\mapsto(\\lambda,\\omega)$ satisfying $|\\omega_k-(h_k\\cdot\\lambda+m)^2-M|=O(\\delta^{2p})$, such that for every parameter point in $G$ the equation admits a space-time quasi-periodic solution of the form $u(t,x)=\\sum a(n,j)e^{i(n\\cdot\\omega+M)t}e^{i(j\\cdot\\lambda+m)x}$ with exponentially decaying Fourier coefficients, close to the two-mode linear wave. The solutions are bifurcations of linear solutions with two time frequencies, and they appear to be the first smooth, global, non-localized solutions without spatial symmetry for a non-integrable equation of this type.","pith_inferences":["If the small-divisor transfer from the standing-wave setting holds as claimed, the same bi-rational Newton scheme could produce analogous infinite-energy quasi-periodic solutions for other translation-invariant PDEs on $\\mathbb R$ whose linear symbol is quadratic in the spatial frequency.","The near-full measure bound suggests that at fixed small amplitude one could numerically continue from the linear two-mode wave and observe exponentially decaying Fourier coefficients, providing a concrete signature of the predicted solutions.","The bi-rational approximation step may be replaceable by a tame inverse function theorem, which would decouple the main construction from the algebraic structure of the Q-equations and widen its applicability."],"forward_implications":["The solutions have infinite energy and mass for all time, so they live outside the usual $H^1$ energy space while remaining smooth.","The existence holds for every integer $p\\ge 1$, so increasing the power of the nonlinearity does not obstruct the construction.","The parameter set has relative measure at least $1-|\\log\\delta|^{-1/2}$, so the non-exceptional parameters form most of the phase-frequency space when $δ$ is small.","The time frequencies differ from the linear frequencies by $O(\\delta^{2p})$, and the Fourier coefficients decay exponentially, giving a quantitative description of the quasi-periodic wave.","The method is stated to generalize to any dimension and any number of time frequencies independent of the space frequencies, including forced equations."],"supporting_citations":[{"why":"Supplies the Main Lemma invoked for the critical region (i) of small $(\\theta,\\varphi)$; the theorem depends on that lemma surviving the bi-rational change of variables and the non-compact two-frequency setting.","marker":"[W5]"},{"why":"Provides the semi-algebraic variable-reduction lemmas (Lemmas 2.3 and 2.4) used to control the bad parameter set in region (i).","marker":"[B3]"},{"why":"Furnishes the Green's function framework and the nonlinear Anderson localization construction that the Newton scheme follows.","marker":"[B2]"},{"why":"Contains Lemma 5.4, the Lipschitz eigenvalue variation technique used for the unbounded region (ii).","marker":"[W4]"},{"why":"Supplies the multi-scale resolvent expansion and Proposition 5.1 used to obtain the pointwise decay estimates (2.23) and (2.48).","marker":"[BGS]"},{"why":"Provides the Lipschitz implicit function theorem used in the eigenvalue analysis to construct the families of Lipschitz functions $\\theta_i$ and $\\varphi_i$.","marker":"[Cl]"}],"fun_headline_variants":["Infinite-energy quasi-periodic waves on R without symmetry","Nonlocalized quasi-periodic NLS waves on the real line","Smooth global quasi-periodic waves on R without symmetry","Nonlocalized space-time quasi-periodic NLS waves"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire construction hinges on the assertion that the small-divisor estimates proved in the author's earlier standing-wave paper still hold after replacing the flat parameter space by the bi-rational change of coordinates $(\\lambda,m,M)\\leftrightarrow(\\lambda,\\omega)$ and working with two frequencies on the whole line; no proof of that transfer is given in this paper.","fun_headline_variants_meta":{"raw":{"variants":["Infinite-energy quasi-periodic waves on R without symmetry","Nonlocalized quasi-periodic NLS waves on the real line","Smooth global quasi-periodic waves on R without symmetry","Nonlocalized space-time quasi-periodic NLS waves"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000847,"raw_usage":{"total_tokens":3652,"prompt_tokens":881,"completion_tokens":2771,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":497,"completion_tokens_details":{"reasoning_tokens":2702}},"tokens_in":497,"tokens_out":2771,"duration_ms":19054,"temperature":1.0,"reasoning_tokens":2702,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:10:22.598078+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a single large scale $N$ in region (i), the exceptional set $\\Theta_N$ arising from the local Lipschitz eigenvalue curves after composing with the bi-rational map $M_N$; if its sectional measure in $\\theta$ or $\\varphi$ exceeds $e^{-N^\\tau}$ or if a bad box remains at distance at least $N/10$, then the estimates (2.47)-(2.48) fail and the Newton construction cannot be continued.","supporting_citations":[],"review_version":1}