{"id":"700cc858-bfd3-4f6d-b005-e843d38b9ffa","arxiv_id":"2507.01317","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"A physical-space bilinear estimate method reproduces the sharpest known local well-posedness thresholds for the 2d and 3d Zakharov system without Bourgain spaces.","lead":"This paper gives a new proof, without Bourgain spaces, of the best known low-regularity well-posedness results for the 2d and 3d Zakharov system of plasma physics. It shows that a physical-space div-curl and bilinear estimate method can replace the standard Fourier-analysis machinery for these nonlinear dispersive equations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The induction rests on forced-iterate bilinear estimates (59)-(60), (63)-(64) that are asserted without derivation, and the printed (63) is not the direct consequence of the balance laws (61)-(62).","rationale":"The reader's weakest assumption identifies the same load-bearing point: the forced-iterate bilinear estimates are asserted rather than proved. I examined Theorem 4.3 itself, including the step that compares ∫( |∇Pλ n|² + |∂t Pλ n|² ) dy against λ² ∫|Pλ n|² dy; because the y-integrals are separated, Bernstein's inequality applies and the potential pointwise-weighted counterexample does not arise. Thus the free bilinear estimates appear plausible and are not the main gap. The gap is the passage from free solutions to the forced Picard iterates. The paper says 'As we have shown in Section 4, applying div-curl lemma to (57) and (58), we have ...', but Section 4 contains no such derivation. The formulas (59)-(60) and (63)-(64) are precisely what the induction needs, and the printed index in (63) is inconsistent with the balance laws (61) unless there is a typo. Since the contraction bound (70) and the d=3 transfer in Section 6 both inherit this unproved step, the verdict CONDITIONAL is appropriate: the advertised method may be correct, but the paper as written does not deliver the full proof it advertises.","tokens_in":23263,"tokens_out":37354,"duration_ms":395194,"concrete_test":"Independently re-derive (59)-(60) from Lemma 4.1 applied to the balance laws (57)-(58), and then (63)-(64) from (61)-(62), without importing Corollary 4.4. Specifically, track the source term G1 = ∫ P≤λ(∂tℜv^{(k+1)}) P≤λ(Δ|E^{(k)}|²) dy through the div-curl lemma and determine whether the resulting estimate is linear or square-root in ∥G1∥_{L1}; and check whether the E-source factor is P_{μ,e1}E^{(k+1)} rather than P_{μ,e1}E^{(k)}. If the source enters linearly, or if the index must be k+1 and (63) cannot be corrected by a trivial change, then Proposition 5.2 does not follow.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central induction in Proposition 5.2 and the contraction bound (70) depend on bilinear estimates for the forced Picard iterates. Theorem 4.3 and Corollary 4.4 prove only the free case with zero sources. Equations (59)-(60) and (63)-(64) are introduced by 'applying div-curl lemma to (57)-(58)/(61)-(62)', but the application is not carried out. This matters for two reasons. First, Lemma 4.1 is stated for sources G1,G2 entering linearly, whereas the printed estimates place the L1 norm of a quadratic source under a square root. Obtaining that square root requires an additional argument, for instance absorbing the L-infinity_t L1_x energy growth of the forced wave component, which is nowhere supplied. Second, (63) contains an apparent index inconsistency: the E-source factor is written with P_{μ,e1}E^{(k)}, while the balance laws (61) for E^{(k+1)} have source terms involving P_{μ,e1}E^{(k+1)}; as printed, (63) is not the direct consequence of (61). If the correct factor is E^{(k+1)}, then the subsequent bound in (65) uses exactly that; if the printed E^{(k)} is meant, the estimate is for a different quantity and cannot close the induction for ∥v^{(k+1)}E^{(k+1)}∥. Since Proposition 5.3 also relies on the same asserted estimates for differences, the missing derivation is load-bearing for both d=2 and the d=3 adaptation in Section 6, which explicitly omits the remaining argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a physical-space, Bourgain-space-free proof of sharp local well-posedness for the Zakharov system: in 2D for data (E0,n0,n1) in L2 × H^{-1/2} × H^{-3/2} and in 3D for s>0, l=s-1/2, with Lipschitz data-to-solution maps. The proof is organized as a Picard iteration reduced to Proposition 1.4, with the key input a div-curl-type bilinear estimate proved in physical space (Theorem 4.3) and refined Strichartz estimates. The paper develops the 2D induction in some detail (Sections 5.2-5.4) and only sketches the 3D case (Section 6). The claimed results match the thresholds of Bejenaru-Herr-Holmer-Tataru and Bejenaru-Herr.","tokens_in":23650,"tokens_out":3267,"duration_ms":35409,"significance":"The results, if correct, would demonstrate that the sharp low-regularity local well-posedness thresholds for the Zakharov system can be reached without Bourgain spaces, using instead a div-curl lemma and mixed-norm Strichartz estimates. The method is genuinely different and has no fitted parameters; the free-evolution bilinear estimates in Theorem 4.3 and Corollary 4.4 are presented with detailed proofs. However, the significance is conditional because several load-bearing forced-iterate estimates are asserted without proof, and the 3D induction is explicitly omitted. If the missing arguments can be supplied, the paper would be an important methodological contribution to nonlinear dispersive equations.","major_comments":[{"comment":"The forced-iterate bilinear estimates are stated as consequences of applying Lemma 4.1 to the balance laws (57)-(58) and (61)-(62), but the derivation is not carried out. Lemma 4.1 is linear in the source terms G1 and G2, whereas the printed estimates place the L1 norm of a quadratic source under a square root. Obtaining such a square-root bound requires an additional argument, for instance controlling the L∞_t L1_x growth of the forced wave energy, and that argument is not supplied. Since these estimates are used to close the induction in Proposition 5.2 and the contraction bound (70), they are load-bearing and cannot be left as assertions.","section":"Section 5.3, equations (59)-(60) and (63)-(64)"},{"comment":"There is an index inconsistency in (63): the E-source factor is written as P_{μ,e1} E^{(k)}, while the balance laws (61) for E^{(k+1)} have source terms involving P_{μ,e1} E^{(k+1)}. As printed, (63) is not the direct consequence of (61). If the correct factor is E^{(k+1)}, then the subsequent bound in (65) uses that; if E^{(k)} is intended, the estimate is for a different quantity and cannot close the induction for ||v^{(k+1)}E^{(k+1)}||. This must be clarified and corrected in a revision.","section":"Section 5.3, equation (63)"},{"comment":"For d=3, the paper only proves the bound for E^{(1)} and then states 'Therefore, we omit the rest of the proof.' Since the refined Strichartz estimate (33) differs from the 2D analogue (32) by introducing T^{s/2}-type factors and different scaling in the N1 norm, the induction for all k, the bounds (13)-(15), and the contraction estimates (16)-(18) are not established. As Proposition 1.4 is the core of Theorem 1.1, the 3D claim is not proved by the current text.","section":"Section 6"},{"comment":"The proof of the contraction estimate (70) for R_kb relies on bilinear estimates for differences of forced iterates, which are asserted without derivation. In particular, the chain (73)-(74) uses a div-curl estimate for (ℜv^{(k)}-ℜv^{(k-1)}) and (E^{(k)}-E^{(k-1)}) that is not written out. Without these forced-difference versions of Corollary 4.4, the recursion (70) has no foundation.","section":"Section 5.4, Proposition 5.3"}],"minor_comments":[{"comment":"There is a typo: 'non-neagtive' should be 'non-negative'.","section":"Section 2"},{"comment":"The line 'sup_a LHS ≳ (μ−1) ... ≳ μ ...' appears to involve a factor of order one; please verify that the lower bound is written with the correct constants, since the subsequent use of μ≫1 is essential.","section":"Section 4, proof of Theorem 4.3"},{"comment":"The proof invokes [19, Lemma 4.3] for the sup-over-translations lower bound, but the lemma is not stated in this paper. Since it is a key tool, either restate it or give a reference with the precise statement.","section":"Section 4, Theorem 4.3"},{"comment":"The definition of the dual norm N*_1(T) has λ^{-4s}, but in the estimates for I1 and I3 the factors σ^{-s} and μ^{2s} are used; please check the homogeneity and ensure the powers of λ are consistent.","section":"Section 6"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's central claim hinges on the unsupported forced-iterate bilinear estimates (59)-(60), (63)-(64) and on an omitted d=3 induction. I would ask the authors to provide complete proofs of these statements, or at least a detailed derivation of the source-term corrections, before the paper can be accepted. The mathematical approach is original and worth pursuing, but in its current state the proof is incomplete."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a real methodological contribution—a physical-space, div-curl based bilinear estimate that bypasses Bourgain spaces—but as it stands the proof of the main theorem has load-bearing gaps. The d=2 iteration depends on forced-iterate bilinear estimates that are stated without derivation, and the d=3 section only sketches the E^(1) step and omits the rest. My own reading agrees with the stress-test note: equation (63) has an index mismatch relative to (61), and the square-root terms in (59)–(60) do not follow directly from the div-curl lemma as stated.\n\nWhat is actually new: the directional phase-space decomposition, the refined Strichartz estimates in Section 3, and the free bilinear estimates in Theorem 4.3 and Corollary 4.4. These are clean and worth studying. The paper is also honest: it reproduces the sharp thresholds from Bejenaru–Herr–Holmer–Tataru and Bejenaru–Herr, and it does not claim new well-posedness statements. That is a legitimate contribution if the proofs close.\n\nWhere the soft spots are: the transition from free solutions to forced Picard iterates is the heart of the matter, and it is skipped. Lemma 4.1 controls products of quantities satisfying balance laws with source terms entering linearly. The estimates (59), (60), (63), and (64) put an L^1 norm of a quadratic source under a square root, and the reader never sees the energy-growth or absorption argument that would produce that. Additionally, (63) appears to involve E^(k) and v^(k) where the balance laws (61)–(62) would require E^(k+1) and v^(k+1); as printed it is not a direct consequence. The d=3 section proves only the first estimate and then says the rest parallels d=2, but the d=2 parallel is precisely what is missing. These gaps are fixable in principle, but they are not cosmetic.\n\nThe citation pattern is fine; using the authors' own prior lemmas is normal, and those lemmas are stated explicitly enough to check. The paper leans on its companion works, but that is not a problem here.\n\nWho this is for: experts in low-regularity dispersive well-posedness who are interested in proof techniques that avoid Bourgain spaces. A serious referee should engage with it, because the core idea is worth developing. I would not cite it in its current form, but I would want to see a revised version where the forced estimates and the d=3 argument are written out. Send it to review; just ask for a complete or at least an honestly flagged revision.","headline":"A genuinely new physical-space bilinear-estimate framework that reproduces known Zakharov thresholds, but the forced-iterate and d=3 parts are asserted rather than proved, so the preprint is promising but incomplete.","tokens_in":24169,"tokens_out":4628,"would_cite":false,"duration_ms":50380,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55"],"pacs":[],"model":"deepseek-v4-flash","headline":"Using only physical-space estimates, the paper proves that the 2d and 3d Zakharov system are locally well-posed at the same sharp Sobolev regularities previously obtained with Bourgain spaces, with Lipschitz data-to-solution maps.","keywords":["Zakharov system","Local well-posedness","Bilinear estimate method","Div-curl lemma","Physical space proof","Refined Strichartz estimates","Sharp regularity","Bourgain spaces"],"falsifier":"Compute both sides of the forced bilinear estimate (59) for the first Picard iterate in $d=2$, taking $E_0$ concentrated near frequency $\\mu$ in the $e_1$ direction and $v_0$ at a much lower frequency $\\lambda$. If the claimed $\\mu^{-1/2}$ gain, or the stated dependence on the source norm, fails for any large frequency ratio, the induction in Proposition 5.2 and the contraction bound (70) collapse.","tokens_in":23046,"feed_emoji":"🌊","tokens_out":11787,"duration_ms":113837,"temperature":0.7,"pith_summary":"This paper sets out to show that the sharp local well-posedness thresholds for the Zakharov system can be proved without using Bourgain spaces, that is, without the frequency-weighted function spaces tailored to the dispersive evolution. It establishes that the 2d system is locally well-posed for data in $L^2_x(\\mathbb{R}^2)\\times H_x^{-1/2}(\\mathbb{R}^2)\\times H_x^{-3/2}(\\mathbb{R}^2)$ and the 3d system for $s>0$ with $l=s-1/2$, the same regularities as the best known results. The proof runs a Picard iteration in adapted spaces $S^1(T)$ and $N^1(T)$ built from directional frequency decompositions, controlling the quadratic nonlinearities with bilinear estimates obtained from a div-curl lemma and from the mass and momentum balance laws of the free Schr\\\"odinger and wave equations. The data-to-solution map is Lipschitz continuous on the stated data classes, and the paper claims these thresholds reflect physical-space structure rather than the Bourgain-space method.","feed_headline":"Sharp Zakharov thresholds reached without Bourgain spaces","feed_subtitle":"Div-curl estimates give Lipschitz well-posedness at the sharp Zakharov regularity in 2d and 3d.","key_machinery":"The load-bearing object is a div-curl type bilinear estimate: when $(f_{11},f_{12})$ and $(f_{21},f_{22})$ solve $\\partial_t f_{11}+\\partial_{x_1} f_{12}=G_1$ and $\\partial_t f_{21}-\\partial_{x_1} f_{22}=G_2$, the spacetime integral of $f_{11}f_{22}+f_{12}f_{21}$ is bounded by the product of $L^1$-type norms of the initial data and sources (Lemma 4.1). Proposition 4.2 provides the needed balance laws: the mass and momentum densities of the free Schr\\\"odinger and wave equations satisfy exactly such first-order systems in the $(t,x_1)$ variables. Applying the lemma to these laws after a directional dyadic decomposition $P_{\\lambda,\\omega_i}$, in which the distinguished direction $\\omega_i$ carries most of the frequency, gives the bilinear estimates of Theorem 4.3 with the $\\mu^{-1/2}$ and $\\lambda^{-1/2}$ gains that close the gap to the optimal regularity. Refined Strichartz estimates (Theorem 3.2), in the mixed norms $L^q_t L^2_{\\omega_i} L^\\infty_{\\omega_i^\\perp}$, then define the iteration spaces $S^1(T)$ and $N^1(T)$ on which the Picard contraction runs.","core_discovery":"The central claim is Theorem 1.1: the Zakharov system, written with $v=n+i\\Lambda^{-1}\\partial_t n$ so that the wave equation becomes first order, has a unique local solution with Lipschitz data-to-solution map for $(E_0,n_0,n_1)$ in $L^2_x(\\mathbb{R}^2)\\times H_x^{-1/2}(\\mathbb{R}^2)\\times H_x^{-3/2}(\\mathbb{R}^2)$ when $d=2$, and in $H^s_x(\\mathbb{R}^3)\\times H_x^{s-1/2}(\\mathbb{R}^3)\\times H_x^{s-3/2}(\\mathbb{R}^3)$ when $d=3$ and $s>0$. The paper's discovery is the route, not the regularity: these thresholds are obtained by a physical-space bilinear estimate method, with a div-curl lemma supplying the derivative gains and refined Strichartz estimates in mixed spatial norms supplying the remaining control. The argument is a contraction in the $S^1(T)\\cap L^\\infty_t L^2_x$ type spaces, so existence, uniqueness, and Lipschitz dependence are proved together. If correct, it shows the sharp well-posedness is not an artifact of Bourgain-space calculus.","pith_inferences":["Our extension: the same balance-law plus div-curl route would likely produce bilinear estimates for any system whose quadratic nonlinearity couples a Schr\\\"odinger and a wave component, for instance the Klein-Gordon-Schr\\\"odinger system; the paper does not make this claim.","Our observation: Section 6 transfers the $d=2$ proof to $d=3$ by asserting that the remaining arguments parallel the $d=2$ case, so the 3d theorem as written rests on an unprinted induction; a reader relying on the 3d result should verify those steps.","Our testable extension: apply the method to $d\\geq 4$, where the sharp well-posedness range has been determined by Fourier-based methods; whether the directional decomposition still closes at the boundary is not addressed here."],"forward_implications":["The 2d Zakharov system is locally well-posed, with Lipschitz data-to-solution map, for $(E_0,n_0,n_1)\\in L^2_x(\\mathbb{R}^2)\\times H_x^{-1/2}(\\mathbb{R}^2)\\times H_x^{-3/2}(\\mathbb{R}^2)$.","The 3d system is locally well-posed for every $s>0$ with $l=s-1/2$, matching the known boundary of the well-posed region.","The sharp thresholds are not tied to Bourgain-space calculus: the same results follow from physical-space bilinear estimates and refined Strichartz estimates.","The Picard iteration contracts at a geometric rate in the $S^1$ and $N^1$ spaces, so existence, uniqueness, and continuous dependence on the data are obtained simultaneously.","The method, if accepted, gives a template for treating other semilinear dispersive equations where Bourgain spaces had seemed indispensable."],"supporting_citations":[{"why":"Supplies the sharp 2d local well-posedness threshold for the Zakharov system with $L^2$ Schr\\\"odinger data that Theorem 1.1 reproduces.","marker":"[2]"},{"why":"Supplies the 3d well-posedness range, near the boundary of which Theorem 1.1 places $s>0$, $l=s-1/2$.","marker":"[1]"},{"why":"Introduces the Bourgain-space method; the paper's entire point is to give an alternative that avoids it.","marker":"[15]"},{"why":"The first part of the series, which develops the bilinear estimate method and supplies Lemma 4.3 used inside the proof of Theorem 4.3.","marker":"[19]"},{"why":"Introduces the div-curl type lemma due to the third author that is the paper's central tool.","marker":"[24]"},{"why":"Proves the div-curl lemma in the form quoted as Lemma 4.1 in this paper.","marker":"[22]"},{"why":"Christ-Kiselev lemma used in the proof of the refined Strichartz estimates to pass from the kernel estimate to the norm estimate.","marker":"[7]"}],"fun_headline_variants":["No Bourgain spaces needed for sharp Zakharov results","Div-curl lemma yields sharp Zakharov well-posedness","Physical-space proof bypasses Bourgain method","Bilinear estimates in physical space settle sharp Zakharov","Bypassing Bourgain spaces: new route to sharp Zakharov"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs the unshown assertion that the div-curl bilinear estimates, proved for solutions of the homogeneous, source-free Schr\\\"odinger and wave equations, extend with the same gains to the forced Picard iterates; the three-dimensional case is then transferred from the two-dimensional argument by assertion rather than by a written derivation.","fun_headline_variants_meta":{"raw":{"variants":["No Bourgain spaces needed for sharp Zakharov results","Div-curl lemma yields sharp Zakharov well-posedness","Physical-space proof bypasses Bourgain method","Bilinear estimates in physical space settle sharp Zakharov","Bypassing Bourgain spaces: new route to sharp Zakharov"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000741,"raw_usage":{"total_tokens":3325,"prompt_tokens":984,"completion_tokens":2341,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":2258}},"tokens_in":600,"tokens_out":2341,"duration_ms":19744,"temperature":1.0,"reasoning_tokens":2258,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:56:23.413077+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute both sides of the forced bilinear estimate (59) for the first Picard iterate in $d=2$, taking $E_0$ concentrated near frequency $\\mu$ in the $e_1$ direction and $v_0$ at a much lower frequency $\\lambda$. If the claimed $\\mu^{-1/2}$ gain, or the stated dependence on the source norm, fails for any large frequency ratio, the induction in Proposition 5.2 and the contraction bound (70) collapse.","supporting_citations":[{"cited_title":"Bejenaru, S","cited_arxiv_id":null,"evidence_quote":"Supplies the sharp 2d local well-posedness threshold for the Zakharov system with $L^2$ Schr\\\"odinger data that Theorem 1.1 reproduces."},{"cited_title":"Bejenaru and S","cited_arxiv_id":null,"evidence_quote":"Supplies the 3d well-posedness range, near the boundary of which Theorem 1.1 places $s>0$, $l=s-1/2$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the Bourgain-space method; the paper's entire point is to give an alternative that avoids it."},{"cited_title":"Physical Space Proof of Bilinear Estimates and Applications to Nonlinear Dispersive Equations","cited_arxiv_id":"2410.13656","evidence_quote":"The first part of the series, which develops the bilinear estimate method and supplies Lemma 4.3 used inside the proof of Theorem 4.3."},{"cited_title":"Wang and Y","cited_arxiv_id":null,"evidence_quote":"Proves the div-curl lemma in the form quoted as Lemma 4.1 in this paper."},{"cited_title":"Christ and A","cited_arxiv_id":null,"evidence_quote":"Christ-Kiselev lemma used in the proof of the refined Strichartz estimates to pass from the kernel estimate to the norm estimate."}],"review_version":1}