{"id":"8f2807ac-441c-43a4-8aea-d45bf11c0d7c","arxiv_id":"2411.16610","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Local well-posedness of the cubic NLS on the half-space R^n_+ is proved for s > n/2 - 1 using the Fokas method, with uniqueness and Lipschitz dependence on the data.","lead":"This paper proves that the cubic nonlinear Schrödinger equation is locally well-posed on a higher-dimensional half-space when the initial data live in natural Sobolev spaces and the boundary data in specially adapted Bourgain spaces. It matters because it extends rigorous existence, uniqueness, and continuous dependence results from lower-dimensional settings to all dimensions at the same regularity threshold as the whole-space problem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 rests on Lemma 3.1, an unproved boundary-data extension lemma; if its compatibility count or the B^s anisotropic extension fails, the linear estimates and the fixed-point argument collapse.","rationale":"I read the paper in good faith and find the central strategy coherent: solve the forced linear ibvp via Fokas, derive the boundary Bourgain space B^s, prove linear and trilinear estimates, and close a fixed-point argument. The proofs of Theorem 2.1, the trilinear estimates, and the contraction argument are presented in detail, and I did not find a fatal flaw. The single most load-bearing weakness is the unproved Lemma 3.1, exactly as the reader identified. It is load-bearing because Proposition 3.3 depends entirely on it, and Proposition 3.3 is the bridge from the reduced pure ibvp estimate to the full linear estimates used in Theorem 1.1. The paper itself flags the omission, so this is not a manufactured concern. The lemma is plausible: the compatibility count floor((2s-1)/4) matches the temporal regularity (2s+1)/4 of B^s, so a standard Sobolev-type extension should work. However, the B^s norm is anisotropic and the uniform-in-ξ' extension is not immediate from the cited theorem. Thus a conditional verdict is appropriate: the theorem should be accepted only after Lemma 3.1 is proved or replaced by a direct argument. I also agree with the reader's secondary point: in the proof of Theorem 1.2, β_2 is set to s/8 while (1.25) gives a different expression; this affects the lifespan formula (1.14) but does not threaten the small-data central claim. No change to the reader's conditional verdict is needed.","tokens_in":52847,"tokens_out":29961,"duration_ms":282630,"concrete_test":"Complete the omitted proof of Lemma 3.1. Concretely: for s ≥ 0, split ‖h‖_{B^s}^2 as the displayed X^{0,(2s+1)/4} + X^{s,1/4} sum; for each ξ', extend the temporal function e^{iξ'^2t}\\hat h_{x'}(ξ',t) from (0,T) to a function supported in (0,2) using the standard Sobolev extension operator that requires exactly floor((2s-1)/4) vanishing derivatives at t=0; then recombine and check that the two resulting terms are controlled by the two components of ‖G0‖_{B^s_T}^2 with constants independent of ξ'. If the low-frequency (X^{0,(2s+1)/4}) component needs one extra derivative at a particular s, Lemma 3.1 is false; if both components extend with the stated count, Proposition 3.3 is valid and the central theorem stands.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 3.1 is the only unproved link in the chain from the reduced pure ibvp estimates (Theorem 2.1) to the full linear estimates (Theorem 1.3). Proposition 3.3 is proved in three sentences: extend G0 by Lemma 3.1 to h with supp h ⊂ (0,2) and ‖h‖_{B^s} ≲ ‖G0‖_{B^s_T}, then apply (2.3)/(2.4). Proposition 3.3 is then the sole input for the G0 term in the decomposition (3.14) used in the proof of Theorem 1.3; (1.21)/(1.22) feed directly into the fixed-point estimate (9.4)-(9.6) for Theorem 1.1. The paper explicitly says \"we omit the proof of Lemma 3.1 here.\" The lemma is not obviously false: the displayed equivalence at the start of its justification indicates that B^s has temporal regularity (2s+1)/4, so the compatibility count floor((2s-1)/4) is the natural one for extending a function supported in (0,2). But the equivalence is only sketched, and the anisotropic, ξ'-dependent nature of the norm means that the claimed uniform extension is not a direct consequence of the cited scalar Sobolev extension theorem. If the trace count or the norm bound in Lemma 3.1 is wrong, the pure ibvp estimates (3.15)-(3.16) fail, and Theorem 1.1 lacks a core step. The reader's beta_n/lifespan inconsistency is real but secondary; it affects the n=2 large-data result (1.14), not the small-data theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the initial-boundary value problem for the linear and cubic nonlinear Schrödinger equations on the half-space R^n_+, n ≥ 2. The linear problem is solved via the Fokas unified transform, and the solution formula is used to derive linear estimates in spatial and temporal Bourgain-type spaces, introducing a boundary data space B^s. The main result, Theorem 1.1, claims local well-posedness for small data at the same Sobolev threshold s > n/2 - 1 as the whole-space Cauchy problem, with a locally Lipschitz data-to-solution map, under compatibility conditions. A second result, Theorem 1.2, claims arbitrary-size data well-posedness in dimension n = 2 for 0 < s < 1/2. The proof strategy is a contraction argument on the Fokas-based iteration map, supported by trilinear estimates in X^{s,b} and Y^{s,b} spaces and by an optimality argument for the trilinear exponents.","tokens_in":53181,"tokens_out":13965,"duration_ms":132958,"significance":"If correct, the main theorem is a significant advance: it matches the known whole-space Sobolev threshold in all dimensions n ≥ 2 for a half-space IBVP, and the boundary data space B^s is a natural object emerging from the Fokas formula. The paper contains substantial original technical work: the detailed microlocal proof of the linear estimates for the reduced pure IBVP, the trilinear estimates in spatial and temporal Bourgain spaces, and the optimality construction in Section 8. The derivation of the homogeneous estimates leading to (2.13) is a genuine strength, as is the explicit change-of-variables structure that motivates the B^s norm. However, the main theorem rests on an unproved extension lemma and on a fixed-point argument whose compatibility requirements are not fully addressed, so the significance is contingent on those gaps being closed.","major_comments":[{"comment":"Lemma 3.1 is load-bearing but is not proved: the paper states 'we omit the proof of Lemma 3.1 here' immediately after (3.18). Proposition 3.3, which supplies the pure-IBVP estimates (3.15)-(3.16), is proved in three sentences by extending G0 via Lemma 3.1 and then applying Theorem 2.1. These pure-IBVP estimates are the only input for the boundary term in the decomposition (3.14), and hence they feed directly into the fixed-point estimates (9.4)-(9.6) for Theorem 1.1. The displayed equivalence before the omission is only a sketch and does not by itself establish the claimed anisotropic extension with compact support in (0,2) and controlled B^s norm. The cited Sobolev extension theorem in [50] is not directly applicable to the anisotropic, ξ'-dependent weight in (1.4). This gap must be closed by a full proof or by a precise reference that covers the B^s spaces used here.","section":"Section 3, Lemma 3.1 and Proposition 3.3"},{"comment":"The contraction map Φ(u)=S[u0,g0; ∓|u|^2u] is defined on the ball B(r) in X^{s,b}∩Y^{s,b}, but the linear estimates of Theorem 1.3 are stated only under compatibility condition (3.21), and the proof of Proposition 3.3 requires the trace conditions (3.17) on the reduced boundary data G0. For an arbitrary iterate u in B(r), the forcing ∓|u|^2u need not satisfy the compatibility conditions needed to make G0 trace-free at t=0; moreover, functions in X^{s,b} with b<1/2 do not in general have the temporal traces needed to impose those conditions. The manuscript does not define a subspace of functions satisfying the compatibility conditions, nor does it prove that Φ maps such a subspace to itself. Without this, the application of (1.21)-(1.22) in (9.4) is not justified, and the contraction argument as written is incomplete.","section":"Section 9, proof of Theorem 1.1, equations (9.1)-(9.6)"},{"comment":"The proof of Theorem 1.2 sets β2=s/8 and uses Lemma 9.1 with b-b'=s/16, which leads to the lifespan bound T0 ∼ [1+‖u0‖+‖g0‖]^{-32/s} in (1.14). However, the definition of β_n in (1.25) gives β2=1/16 for every n=2 and 0<s<1/2, because the second term in the minimum is always larger than 1/16. With β2=1/16, the same argument gives b-b'=1/32 and hence the different lifespan exponent -64. The two choices are inconsistent, and consequently Theorem 1.2 as stated is not actually proved by the argument in Section 9.","section":"Section 9, proof of Theorem 1.2, after (9.15); compare with (1.25) and (1.14)"}],"minor_comments":[{"comment":"The definition of the region A2 appears to be misprinted: it reads |ζ| ≤ |ξ-ζ-η| < |ξ-ζ-η|, which is not meaningful; in view of the definition of A1 in (6.11), it should presumably read |ζ| ≤ |ξ-ζ-η| < |η|.","section":"Section 6, equation (6.12)"},{"comment":"The identity used for |u|^2u - |v|^2v is written with missing conjugate symbols and is hard to parse; please write the standard decomposition (u-v)|u|^2 + v (ar{u}-ar{v})(u+v) or the equivalent form, and specify exactly how the extensions \tilde{u} and \tilde{v} in (9.10) are chosen from u and v.","section":"Section 9, equations (9.9)-(9.11)"},{"comment":"The sentence 'The extension in B^s space is similar to the extension in Sobolev spaces Hs' is not a substitute for a proof, especially for the negative range -3/2 < s < 0 where the displayed equivalence is not given; please at least state the precise extension theorem being used and explain how its trace count matches floor((2s-1)/4).","section":"Section 3, after (3.18)"},{"comment":"The chain '≲ ‖h‖²_{Ḃs} ≲ ‖h‖²_{B^s}' contains a redundant repetition; the intended intermediate step is the extension of the τ-integration, which should be stated explicitly.","section":"Section 2, proof of Theorem 2.1, around (2.21)"}],"recommendation":"major_revision","confidential_remarks":"The paper has a substantial and credible core: the Fokas-based linear analysis and the trilinear estimates are developed in detail, and the main theorem would be an important result if the gaps are closed. The two decisive issues are the unproved extension lemma and the compatibility of the iterates in the fixed-point argument; both are fixable in principle but require real work, not merely local editing. The n=2 lifespan inconsistency in Theorem 1.2 should also be corrected, but it is secondary to Theorem 1.1."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The paper proves local well-posedness for cubic NLS on R^n_+ at s > n/2 - 1, the same threshold as the whole-space IVP, for all n >= 2. That is a real advance over the half-line and half-plane results. The genuinely new objects—the boundary Bourgain space B^s and the temporal Bourgain spaces Y^{s,b}—come out of the Fokas formula rather than being imposed; the change of variables in (2.13) makes that clear. The linear and trilinear estimates are worked out in considerable detail, and the optimality section is a useful check.\n\nSoft spots. Lemma 3.1 is the load-bearing one. It is the only step that gets boundary data from the restriction space B^s_T to a compactly supported extension h with the same B^s norm, and Proposition 3.3 is just 'apply Lemma 3.1 then Theorem 2.1'. The paper explicitly says the proof is omitted. That is not fatal in itself—the lemma looks plausible, and the displayed equivalence B^s ~ X^{0,(2s+1)/4} + X^{s,1/4} explains the compatibility count—but the B^s norm is anisotropic in xi', so the uniform extension is not a one-line consequence of scalar Sobolev extension theory. A referee needs to see the details. If the lemma fails in the stated range, Theorem 1.1 loses its core linear input.\n\nSecondary: the beta_n definition and its use in n = 2 do not line up. In Theorem 1.4, beta_2 = min{1/16, ...} is 1/16 for the relevant s < 1/2, while the proof of Theorem 1.2 sets beta_2 = s/8, and the lifespan (1.14) has exponent -32/s, which corresponds to beta_2 = s/8. It is fixable, but as written the reader cannot tell which definition the authors intend.\n\nI do not see circularity. The data spaces are natural consequences of the Fokas solution formula, not reverse-engineered, and the compatibility conditions for s > 1/2 are handled carefully.\n\nBottom line: this is for people working on dispersive IBVPs with the Fokas method and Bourgain spaces. It deserves a serious referee, but I would not accept it as is. The referee should ask for a complete proof of Lemma 3.1, or a precise citation, and for the beta_n inconsistency to be resolved.","headline":"Solid extension of the Fokas-method well-posedness program to half-space NLS, but the main theorem leans on an omitted boundary-extension lemma and a sloppy beta_n definition.","tokens_in":53754,"tokens_out":3698,"would_cite":true,"duration_ms":36014,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55","35G31","35G16","37K10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves local well-posedness of the cubic NLS on the half-space $\\mathbb{R}^n_+$ in dimensions $n\\ge 2$ for Sobolev data above $s=n/2-1$, the same threshold as the whole-space problem, using the unified transform and sharper…","keywords":["cubic nonlinear Schrödinger equation","half-space initial-boundary value problem","unified transform method","well-posedness in Sobolev spaces","Bourgain spaces","trilinear estimates","boundary compatibility conditions","temporal Bourgain spaces"],"falsifier":"Take an admitted exponent $s$ with $\\lfloor(2s-1)/4\\rfloor\\ge 0$ and set $G_0(x',t)=t^{m}\\varphi(x')\\psi_T(t)$ with $m=\\lfloor(2s-1)/4\\rfloor+1$, $\\varphi\\in H^s(\\mathbb{R}^{n-1})$ nonzero, and $\\psi_T$ supported in $(0,T)$. Compute the ratio of the optimal compactly supported extension norm $\\|h\\|_{\\mathcal{B}^s}$ to $\\|G_0\\|_{\\mathcal{B}^s_T}$; if this ratio is unbounded as $T\\to 0$, Lemma 3.1 fails and the proof of the linear estimates (and hence Theorem 1.1) loses its core step. The paper's own optimality construction already shows the trilinear estimate fails at $s=n/2-1$, which rules out lowering the threshold by this method.","tokens_in":52614,"feed_emoji":"🌊","tokens_out":11269,"duration_ms":96411,"temperature":0.7,"pith_summary":"The paper claims that the cubic nonlinear Schrödinger equation on the half-space $\\mathbb{R}^n_+$ is locally well-posed for initial data in $H^s$ and boundary data in a natural boundary Bourgain space $\\mathcal{B}^s_T$ whenever $s>n/2-1$, for every dimension $n\\ge 2$, provided the data are small and satisfy compatibility conditions. This is the same Sobolev threshold as the whole-space Cauchy problem, so the claim is that the boundary costs no loss of regularity. The proof constructs the solution by a fixed point around the unified transform solution formula, splitting the forced linear problem into an initial-value part, a forcing part, and a pure boundary part, and uses new trilinear estimates in Bourgain spaces with $b<1/2$. In dimension $n=2$ and regularity $0<s<1/2$, the small-data restriction is removed and an explicit lifespan bound is given. The weakest point, flagged by the paper itself, is an omitted proof of the boundary-data extension lemma that connects the reduced pure boundary estimate to the full linear estimates.","feed_headline":"Half-space cubic NLS well-posed at whole-space threshold","feed_subtitle":"Boundary data in natural Bourgain spaces; proof uses the unified transform and sharp trilinear estimates.","key_machinery":"The machinery is the unified transform solution formula (1.16) for the forced linear ibvp, expressed as a full-space Fourier integral minus a contour integral over $\\partial D^+$ plus a boundary-data term; the decomposition (3.14) that separates the solution into a homogeneous initial-value problem, an inhomogeneous initial-value problem, and a pure boundary problem; the reduced pure ibvp estimate (Theorem 2.1); and the extension lemma (Lemma 3.1) that moves boundary data from $(0,T)$ into compactly supported data on $(0,2)$. The temporal Bourgain spaces $Y^{s,b}$ defined in (1.10) are the new ingredient forced by the boundary, and the trilinear estimates of Theorem 1.4 are what allow the iteration map to contract despite $b<1/2$.","core_discovery":"The central discovery is that the initial-boundary value problem (1.1) with Dirichlet boundary data is locally well-posed in the same Sobolev range as the whole-space problem: for $s>s_n=n/2-1$ and $(2s-1)/4\\notin \\mathbb{N}_0$, small data $(u_0,g_0)$ satisfying the compatibility conditions (3.21) yield a unique solution $u\\in X^{s,b}\\cap Y^{s,b}$ for some $b\\in(0,1/2)$, with a locally Lipschitz data-to-solution map. The quantitative heart is the pair of linear estimates (1.21)--(1.22) and the trilinear estimates (1.23)--(1.24), whose admissible Bourgain exponents occupy $1/2-\\beta_n\\le b'\\le b<1/2$, with $\\beta_n$ given by (1.25). Theorem 1.5 shows the trilinear estimate is optimal, failing at $s\\le n/2-1$, so the threshold is not an artifact of the method. For $n=2$ and $0<s<1/2$, the small-data assumption is eliminated and the lifespan satisfies (1.14).","pith_inferences":["The same inductive structure should yield analogous thresholds for the other cubic nonlinearities $\\pm u^3$ and $\\pm |u|^2\\bar u$ on the half-space, with different low-regularity limits, just as on the half-line.","If Lemma 3.1 is proved as stated, the method likely transfers to other dispersive equations with explicit unified-transform formulae, where the boundary will again force temporal Bourgain spaces and $b<1/2$ estimates.","The small-data restriction in Theorem 1.1 for $n\\ge 3$ may be an artifact of controlling the full $X^{s,b}\\cap Y^{s,b}$ norm at once; a lifespan-shortening argument like the one used for $n=2$ might extend the large-data result to higher dimensions."],"forward_implications":["If Theorem 1.1 is correct, cubic NLS on the half-space is locally well-posed at the same Sobolev exponent as the whole-space problem in every dimension $n\\ge 2$, so the boundary does not force a loss of regularity for small data.","In two dimensions with $0<s<1/2$, the result applies to data of any size and gives an explicit lifespan $T_0=c_0[1+\\|u_0\\|_{H^s}+\\|g_0\\|_{\\mathcal{B}^s_T}]^{-32/s}$.","The data-to-solution map is locally Lipschitz continuous, so nearby initial and boundary data produce nearby solutions in the Bourgain norms.","The compatibility conditions (3.21) are part of the well-posedness statement: boundary data must match the solution's time derivatives at the corner, determined recursively from the equation.","Since the trilinear estimate is optimal, no well-posedness below $s=n/2-1$ can be obtained by this contraction method in these Bourgain spaces."],"supporting_citations":[{"why":"Defines the spatial Bourgain spaces $X^{s,b}$ and the restriction/extension conventions that all linear and trilinear estimates use.","marker":"[6]"},{"why":"Supplies the unified transform solution formula for the forced linear Schrödinger ibvp on the half-line, which the paper reduces to and then lifts to the half-space.","marker":"[21]"},{"why":"First implementation of the unified transform for the NLS ibvp on the half-line; the well-posedness proof there is the template this paper extends.","marker":"[23]"},{"why":"Prior unified-transform well-posedness on the half-plane; the present work generalizes it to all dimensions and to the boundary Bourgain spaces $\\mathcal{B}^s$.","marker":"[33]"},{"why":"Standard Sobolev extension theory invoked to justify the analogous $\\mathcal{B}^s$ extension lemma (Lemma 3.1), the omitted step on which the linear estimates depend.","marker":"[50]"},{"why":"Supplies the Bourgain-space multiplier lemmas and the $L^4$ space-time estimates used in proving the trilinear estimates and the lifespan argument.","marker":"[59]"}],"fun_headline_variants":["Half-space cubic NLS well-posed at whole-space threshold","Cubic NLS on half-space: optimal threshold via unified transform","Sharp well-posedness for NLS on half-space matches whole-space","Bourgain spaces meet boundary: NLS well-posedness","Half-space NLS: sharp trilinear estimates at optimal regularity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole proof leans on Lemma 3.1, which asserts that boundary data in $\\mathcal{B}^s_T$ with vanishing time derivatives up to order $\\lfloor(2s-1)/4\\rfloor$ at $t=0$ can be extended to a compactly supported function with controlled $\\mathcal{B}^s$ norm; the paper states this lemma without proof.","fun_headline_variants_meta":{"raw":{"variants":["Half-space cubic NLS well-posed at whole-space threshold","Cubic NLS on half-space: optimal threshold via unified transform","Sharp well-posedness for NLS on half-space matches whole-space","Bourgain spaces meet boundary: NLS well-posedness","Half-space NLS: sharp trilinear estimates at optimal regularity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001083,"raw_usage":{"total_tokens":4543,"prompt_tokens":972,"completion_tokens":3571,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":588,"completion_tokens_details":{"reasoning_tokens":3481}},"tokens_in":588,"tokens_out":3571,"duration_ms":24241,"temperature":1.0,"reasoning_tokens":3481,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:55:46.542482+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an admitted exponent $s$ with $\\lfloor(2s-1)/4\\rfloor\\ge 0$ and set $G_0(x',t)=t^{m}\\varphi(x')\\psi_T(t)$ with $m=\\lfloor(2s-1)/4\\rfloor+1$, $\\varphi\\in H^s(\\mathbb{R}^{n-1})$ nonzero, and $\\psi_T$ supported in $(0,T)$. Compute the ratio of the optimal compactly supported extension norm $\\|h\\|_{\\mathcal{B}^s}$ to $\\|G_0\\|_{\\mathcal{B}^s_T}$; if this ratio is unbounded as $T\\to 0$, Lemma 3.1 fails and the proof of the linear estimates (and hence Theorem 1.1) loses its core step. The paper's own optimality construction already shows the trilinear estimate fails at $s=n/2-1$, which rules out lowering the threshold by this method.","supporting_citations":[{"cited_title":"Bourgain Fourier transform restriction phenomena for certain latti ce subsets and applications to nonlinear evolution equations","cited_arxiv_id":null,"evidence_quote":"Defines the spatial Bourgain spaces $X^{s,b}$ and the restriction/extension conventions that all linear and trilinear estimates use."},{"cited_title":"Fokas, A unified approach to boundary value problems","cited_arxiv_id":null,"evidence_quote":"Supplies the unified transform solution formula for the forced linear Schrödinger ibvp on the half-line, which the paper reduces to and then lifts to the half-space."},{"cited_title":"Fokas, A","cited_arxiv_id":null,"evidence_quote":"First implementation of the unified transform for the NLS ibvp on the half-line; the well-posedness proof there is the template this paper extends."},{"cited_title":"Himonas and D","cited_arxiv_id":null,"evidence_quote":"Prior unified-transform well-posedness on the half-plane; the present work generalizes it to all dimensions and to the boundary Bourgain spaces $\\mathcal{B}^s$."},{"cited_title":"Lions and E","cited_arxiv_id":null,"evidence_quote":"Standard Sobolev extension theory invoked to justify the analogous $\\mathcal{B}^s$ extension lemma (Lemma 3.1), the omitted step on which the linear estimates depend."},{"cited_title":"Tao, Nonlinear Dispersive Equations-Local and Global Analysis","cited_arxiv_id":null,"evidence_quote":"Supplies the Bourgain-space multiplier lemmas and the $L^4$ space-time estimates used in proving the trilinear estimates and the lifespan argument."}],"review_version":1}