{"id":"828c8647-db29-47e8-a71d-09d587652aa3","arxiv_id":"2608.07866","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"For the 1D periodic cubic NLS, there exists a nonzero singular weak solution with zero initial data in every C_t^0 H_x^alpha below H^{1/6}, making the H^{1/6} uniqueness threshold sharp in the stated singular solution class.","lead":"A new proof constructs a nonzero, time-compactly supported solution of the one-dimensional periodic cubic NLS that starts from zero data while remaining below the H^{1/6} regularity threshold. The result sharpens where non-uniqueness begins for a specific singular notion of the cubic nonlinearity.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-carrier cancellation is algebraically false for complex b: the constant term in Lemma 6.1 should be a^2\\bar b, not a^2b, so with b=-σa^{-2}E_q the zero slab output reproduces \\bar E_q, not E_q, and the induction does not close as written.","rationale":"The reader's verdict is careful, and the negative-Wiener residual framework, the intermittent slab estimates, and the block-disjointness argument are plausible building blocks. The problem is not the framework but the exact algebraic cancellation at the center of the induction. Lemma 6.1 is not merely a display typo: its constant term is wrong for every non-real b, and the whole construction chooses b to be the non-real error E_q. Consequently the residual equation (6.31) is false, the one-step residual bound Proposition 6.8 is not grounded, and the induction in §6.5 cannot be completed as written. The same missing conjugation appears in Definition 2.5 and in the proof of Lemma 2.7, so the claimed compatibility with the ordinary product |u|^2u at C^0_tL^3 is false for complex-valued u; the endpoint sharpness statement, which relies on Lemma 2.7 and on Guo-Kwon-Oh uniqueness, also does not follow. That said, the strategy may be repairable by taking b_{q+1}=-σa^{-2}_{q+1}\\bar E_q and conjugating the middle factor throughout the signed-Fourier definitions; the absolute-value form of C_s makes such a repair natural. This is a substantive revision, not a display-only correction. The reader identified a related limitation (singular product versus ordinary product), but treated it as an honest caveat; the stress test shows a concrete algebraic failure inside the construction itself, so the agreement is only partial. I recommend REJECT of the current version, with resubmission after a consistent conjugation-corrected proof.","tokens_in":26295,"tokens_out":15621,"duration_ms":168599,"concrete_test":"Evaluate the claimed polarization identity with a=1, b=i, z=1: the left side is |1+i|^2(1+i)=2+2i, while the printed right side is i+(1+2)+i(2+1)+i^2=2+4i. Then recompute (6.25) for a nonzero complex E_q, e.g. the q=0 error E_0=η(iχ'_0-4π^2χ_0)e^{2πix}+ση^3χ_0^3e^{2πix}, using b=-σa^{-2}E_q: the zero slab mode of σ|w|^2w is -σ\\bar E_q, not -σE_q, so E_{O,q+1} contains 2i\\,{\\rm Im}E_q. This check settles whether the two-carrier mechanism actually reproduces the full old error. If a consistent conjugation-corrected formulation is intended, the missing bars must be traced through (6.4), Definition 2.5, Lemma 6.11, and (6.63)-(6.66).","verdict_should_be":"REJECT","load_bearing_attack":"The engine of the proof is the two-carrier perturbation w_{q+1}=h_{q+1}Θ_{q+1}(a_{q+1}z+b_{q+1}z^2), whose zero-carrier, zero-slab-output mode must reproduce the full old error E_q. Lemma 6.1 states |az+bz^2|^2(az+bz^2)=a^2b+a(a^2+2|b|^2)z+b(2a^2+|b|^2)z^2+ab^2z^3. For a>0 real, b∈C, |z|=1, the correct expansion has constant term a^2\\bar b, because |az+bz^2|^2=a^2+|b|^2+a\\bar b z^{-1}+ab z. Any non-real b makes the two differ. With b_{q+1}=-σa^{-2}_{q+1}E_q in (6.4), the claimed constant mode is -σE_q, but the actual one is -σ\\bar E_q. Hence (6.25), (6.29), and the residual decomposition (6.31) fail, leaving an unabsorbed term 2i\\,{\\rm Im}E_q of size comparable to δ_q in A^{-s}. E_q is genuinely complex already at q=0: u_0=ηχ_0(t)e^{2πix} gives the coefficient iχ'_0-4π^2χ_0. Separately, Definition 2.5 omits the conjugate on the middle Fourier factor; for complex u this does not define the Fourier coefficient of |u|^2u, so Lemma 2.7 and the sharpness conclusion via Guo-Kwon-Oh do not follow as stated. A repaired version replacing b by -σa^{-2}\\bar E and conjugating the middle factor may close, but the written induction is invalid.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a convex-integration construction of nonzero compactly supported singular weak solutions of the one-dimensional periodic cubic nonlinear Schrödinger equation with zero initial datum. The constructed function lies in C_t^0 H_x^\\alpha for every \\alpha<1/6 and in C_t^0 L_x^p for every 1\\le p<3, and the cubic nonlinearity is interpreted through a signed absolute-Fourier product N(u). The proof adapts the intermittent slab of Gismondi--Ma--Pathak--Radu and introduces a two-carrier perturbation whose zero-output interaction is intended to reproduce the full old error, including the zero Fourier mode. The paper also proves a compatibility lemma between N(u) and the ordinary product |u|^2u at C_t^0 L_x^3, and uses the Guo--Kwon--Oh unconditional uniqueness theorem at H^{1/6} to claim sharpness of the threshold 1/6 within the singular solution class.","tokens_in":26754,"tokens_out":8278,"duration_ms":85290,"significance":"If the technical issues identified below are repaired, this would be a substantial contribution to the convex-integration approach to dispersive equations. The construction is nontrivial: the spatially constant output of the cubic nonlinearity is a genuine obstruction, and the proposed two-carrier mechanism is an original way to address it. The manuscript is unusually detailed and self-contained, with explicit Fourier bookkeeping, a complete placement-by-placement analysis of the signed Fourier sums, and a transparent statement of the parameter recursion. The authors also honestly disclose that the constructed solution solves the equation with the modified singular product rather than with the ordinary product unless it lies in C_t^0 L_x^3. The main concern is that the central algebraic identity and the definition of the cubic product contain a complex-conjugation error that breaks the induction as written.","major_comments":[{"comment":"The displayed identity (1.4)/(6.11) is false for complex b. Direct computation gives |az+bz^2|^2(az+bz^2) = a^2\\bar b + a(a^2+2|b|^2)z + b(2a^2+|b|^2)z^2 + ab^2z^3, with the constant term a^2\\bar b, not a^2 b. With the choice b_{q+1}=-\\sigma a_{q+1}^{-2}E_q in (6.4), the zero carrier output is therefore -\\sigma\\bar E_q rather than -\\sigma E_q. Consequently (6.25), (6.29), the residual decomposition (6.31), and the pure-new estimate in Lemma 6.11 do not close as written, and an unabsorbed term proportional to 2i\\operatorname{Im}E_q remains. The induction is invalid in its printed form. A repair is plausible: if b_{q+1}=-\\sigma a_{q+1}^{-2}\\bar E_q and the constant term in the polarization identity is corrected to a^2\\bar b, then the zero-output cancellation is restored; however, this change must be propagated through (6.4), (6.5), (6.63)--(6.66), and all estimates involving b_{q+1}.","section":"§6.1, Lemma 6.1 and Eq. (6.4)"},{"comment":"The signed cubic product in Definition 2.5 is defined by summing \\hat u(t,n_1)\\hat u(t,n_2)\\hat u(t,n_3) with no complex conjugate on the middle Fourier factor. For complex-valued u this is not the Fourier coefficient of the ordinary product |u|^2u. As a result, the proof of Lemma 2.7 compares the wrong multiplier factor, and the conclusion N(u)=|u|^2u for u\\in C_t^0 L_x^3 does not follow. The subsequent sharpness argument using the Guo--Kwon--Oh uniqueness theorem at H^{1/6} therefore is not established as written. If a conjugate on the middle factor was intended, it must be inserted consistently in Definitions 2.4--2.5, in the expansions in Section 6.4, and in all applicable lemmas. Without such a correction, the compatibility lemma is false for complex solutions.","section":"Definitions 2.4--2.5 and Lemma 2.7"}],"minor_comments":[{"comment":"The sentence ``The complex conjugate in (1.5) is essential'' is inconsistent with the displayed equation (1.5) and with (6.4), which contain no conjugate on E_q. Please clarify the intended formula and ensure the surrounding text and equations agree.","section":"§1.4 and §6.1"},{"comment":"The statement ``non-uniqueness ... to the ... cubic NLS'' should be qualified, as is already done later, by saying that the equation is satisfied in the singular weak sense of Definition 2.6 with nonlinearity N(u), not necessarily with the ordinary product |u|^2u for the constructed solution. The current wording in the abstract is potentially misleading, even though Remark 3.2 and the discussion after Lemma 2.7 are transparent about this point.","section":"Abstract and Theorem 3.1"},{"comment":"The notation \\alpha_{q+1}=1/6-\\varepsilon_{q+1} is used before the reader knows that \\varepsilon_{q+1} tends to zero; consider stating explicitly that \\varepsilon_{q+1}=2^{-(q+1)} immediately after (4.2) rather than later.","section":"§4, Eq. (4.2)"}],"recommendation":"major_revision","confidential_remarks":"The paper is not acceptable in its current form because the central polarization identity and the definition of the nonlinear product contain conjugate errors that break the induction. However, both issues appear locally repairable and the rest of the construction is detailed and internally coherent. I recommend major revision rather than rejection, with the expectation that the authors either fix the conjugation consistently or clearly state a modified solution concept for which the proofs are valid."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. The two-carrier mechanism is a real idea, and the negative-Wiener framework is well executed. But the central cancellation is algebraically wrong as written, and the singular product definition is missing a conjugate. Both are load-bearing, not typos.\n\nWhat's new: the paper adapts the GMPR intermittent slabs to full cubic NLS rather than gKdV. The zero Fourier mode of |u|^2 u is a genuine obstruction, and the idea of using a second carrier to let the zero-carrier, zero-slab-output interaction reproduce the full old error is clever. The signed absolute-Fourier bookkeeping in Section 6.4 is detailed, and the carrier-separation lemma is a solid piece of work. The negative Wiener norm for the residual is a sensible way to sidestep the usual dispersive loss.\n\nThe soft spot: Lemma 6.1 states |az+bz^2|^2(az+bz^2) = a^2 b + ... . The correct expansion has a^2 \\bar b. The paper's own line for |az+bz^2|^2 includes a \\bar b z^{-1} and a b z; multiplying by az+bz^2 gives a^2 \\bar b. With b_{q+1} = -\\sigma a^{-2} E_q, the claimed cancellation in (6.25) actually produces -\\sigma \\bar E_q at constant mode, not -\\sigma E_q. So (6.29), (6.31), and the residual estimate do not close as written; an unabsorbed term of the form (\\bar E_q - E_q)\\Theta^3 remains. This is fixable by setting b = -\\sigma a^{-2} \\bar E_q, but the written induction is invalid.\n\nSecond issue: Definition 2.5 defines N(u) with b u(n1) b u(n2) b u(n3) and sign n1-n2+n3. For complex u this is not the Fourier coefficient of |u|^2 u, which requires \\overline{b u(n2)}. The paper even says the sign is dictated by the conjugated middle factor, but the conjugation is missing from the coefficient. Lemma 2.7 (compatibility with the ordinary cubic) therefore does not hold as stated, and the sharpness argument via Guo-Kwon-Oh loses its bridge. Same fix: put the conjugate on the middle factor.\n\nThe reader's CONDITIONAL verdict is too mild if it treats this as a display typo. The algebra error is in the engine of the induction. That said, the strategy looks repairable, and the paper is honest about its solution class. For a specialist in convex integration or dispersive PDE, there is real value in the two-carrier idea and the carrier-separation analysis. I would send it to a serious referee, but with the expectation that the referee check the polarization identity and the product definition line by line. I would not cite it in its current form.","headline":"A promising two-carrier convex integration scheme for cubic NLS, but the central polarization identity is miscomputed and the singular product definition is missing a conjugate; both are load-bearing.","tokens_in":27233,"tokens_out":9452,"would_cite":false,"duration_ms":86547,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35Q55","35A02","35D30","35B30"],"pacs":[],"model":"deepseek-v4-flash","headline":"Zero initial data admits a nonzero singular solution of the cubic NLS below the $H^{1/6}$ threshold.","keywords":["cubic nonlinear Schrödinger equation","non-uniqueness","convex integration","intermittent slab","singular solution","signed absolute-Fourier product","sharp threshold"],"falsifier":"Evaluate the series defining $C_s(u)$ on the limit function produced by the induction; divergence for some $s>3$ would invalidate the existence claim. Alternatively, any nonzero singular weak solution with zero initial datum belonging to $C_t^0 H_x^{1/6}$ would directly contradict the sharpness statement.","tokens_in":26121,"feed_emoji":"🌊","tokens_out":9062,"duration_ms":88626,"temperature":0.7,"pith_summary":"This paper establishes that the one-dimensional periodic cubic nonlinear Schrödinger equation loses uniqueness below the Sobolev exponent $1/6$, within a carefully defined class of singular weak solutions. It constructs, for either sign of the cubic term, a nonzero solution that is compactly supported in time, starts from zero initial data, and belongs to every $C_t^0 H_x^\\alpha$ with $\\alpha<1/6$ and every $C_t^0 L_x^p$ with $p<3$. The construction uses convex integration with an intermittent slab profile, and the sharpness comes from an existing unconditional uniqueness theorem at $H^{1/6}$: any singular solution of this class with zero initial datum at that higher regularity is forced to vanish. The cost of the construction is that the cubic nonlinearity is defined by a signed absolute-Fourier sum rather than the pointwise product $|u|^2u$, and the two agree only when the solution is regular enough to belong to $C_t^0 L_x^3$.","feed_headline":"Below a Sobolev threshold, zero initial data yields two cubic NLS solutions","feed_subtitle":"A convex-integration construction makes 1/6 the exact dividing line between uniqueness and non-uniqueness for singular weak solutions.","key_machinery":"The engine is an intermittent slab $\\Theta_{\\lambda,\\varepsilon}$: a finitely band-limited, mean-zero, even periodic profile with nonnegative Fourier coefficients on the sparse lattice $\\mu\\mathbb{Z}$, exact cubic moment $\\int_\\mathbb{T}\\Theta^3\\,dx=1$, small $L^1$ and $L^2$ norms, and $L^p$ norms that scale like $\\lambda^{(1-\\varepsilon)(1/3-1/p)}$. The paper places this slab on two carrier waves, writing the perturbation as $w=h\\Theta(a e^{2\\pi i K x}+b e^{4\\pi i K x})$. The key algebraic identity is $|az+bz^2|^2(az+bz^2)=a^2b+a(a^2+2|b|^2)z+b(2a^2+|b|^2)z^2+ab^2z^3$; choosing $b=-\\sigma a^{-2}E$ makes the constant term $a^2b$ reproduce the old error $E$, including its zero Fourier mode, through the resonance $K-2K+K=0$. The residual is measured in a negative Wiener norm $A^{-s}_{x,t}$, the $\\ell^1$ norm of time-continuous Fourier coefficients with weight $\\langle n\\rangle^{-s}$, whose negative weight absorbs the two derivatives in the Schrödinger operator and turns high-frequency output into gains. A complete placement-by-placement analysis of the eight terms in the cubic expansion keeps every low-output interaction under control.","core_discovery":"The central claim is Theorem 3.1: for either $\\sigma\\in\\{-1,1\\}$ and any $s>3$, there is a nonzero function $u$ with $u(0,\\cdot)=0$, compact time support, membership in $\\bigcap_{\\alpha<1/6} C_t^0 H_x^\\alpha \\cap \\bigcap_{1\\le p<3} C_t^0 L_x^p$, and finite $C_s(u)$, solving the equation in the singular weak sense of Definition 2.6. Consequently, zero initial data has both the zero solution and this nonzero singular solution. Conversely, every singular weak solution with zero initial data that belongs to $C_t^0 H_x^{1/6}$ is identically zero, by the embedding $H^{1/6}(\\mathbb{T})\\hookrightarrow L^3(\\mathbb{T})$ and the existing unconditional uniqueness statement. The theorem therefore makes $1/6$ the sharp threshold inside this solution class. The nonlinear term $N(u)$ is defined as the signed absolute-Fourier cubic, is independent of the admissible Fourier cutoff, and coincides with the ordinary product exactly when the solution lies in $C_t^0 L_x^3$.","pith_inferences":["The two-carrier cancellation is likely to transfer to other dispersive equations whose cubic-type nonlinearity produces a low-frequency or constant component that a one-amplitude convex integration scheme cannot discard.","The use of a negative Wiener norm for the residual suggests a general recipe: measure errors in a norm whose negative weight absorbs derivatives, and handle the necessarily low-output terms by exact algebraic identities rather than by smallness.","If the construction can be varied over amplitudes and frequencies, the method may imply a continuum of nonzero singular solutions with zero initial data below $H^{1/6}$, although the paper itself does not state such a statement.","The mismatch between singular and ordinary cubic products below $L^3$ means the non-uniqueness can be interpreted as a consequence of the extended solution concept; a natural testable extension is whether the same iteration can produce an ordinary-product weak solution at this regularity."],"forward_implications":["Zero initial data admits both the zero solution and a nonzero singular weak solution at every regularity below $H^{1/6}$, for either sign of the nonlinearity.","The endpoint $1/6$ is sharp within the singular weak solution class: every such solution with zero initial datum in $C_t^0 H_x^{1/6}$ is identically zero.","The constructed solution lies simultaneously in all $C_t^0 H_x^\\alpha$ with $\\alpha<1/6$ and all $C_t^0 L_x^p$ with $p<3$, not merely in one fixed space.","The singular cubic $N(u)$ is robust under Fourier cutoffs: any admissible Fourier-cutoff sequence yields the same limit, so the nonlinearity is intrinsic to the solution rather than an artifact of one approximation.","In the overlapping regime $u\\in C_t^0 L_x^3$, the singular product agrees with the pointwise product $|u|^2u$, so the two interpretations of the equation become identical there."],"supporting_citations":[{"why":"Supplies the intermittent slab profile and the signed absolute-Fourier definition of rough nonlinearities that the construction adapts.","marker":"[9]"},{"why":"Provides the unconditional uniqueness theorem at $H^{1/6}$ that makes the threshold sharpness assertion possible.","marker":"[12]"},{"why":"Introduced the Fourier-cutoff product definition that motivates the singular weak formulation and the cutoff-robustness statement.","marker":"[6]"}],"fun_headline_variants":["At 1/6 Sobolev regularity, cubic NLS loses uniqueness","Zero data, two solutions: sharp break at Sobolev 1/6","Cubic NLS non-uniqueness pinned at H^1/6 threshold","Two solutions from nothing: singular NLS threshold at 1/6","Sharp threshold: below H^1/6, cubic NLS fails uniqueness"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction depends on accepting the singular weak formulation in which the cubic term is the signed absolute-Fourier product $N(u)$; if one insists on the ordinary pointwise cubic $|u|^2u$ at the constructed regularity, the nonzero solution is not known to satisfy the equation, so the non-uniqueness conclusion does not follow.","fun_headline_variants_meta":{"raw":{"variants":["At 1/6 Sobolev regularity, cubic NLS loses uniqueness","Zero data, two solutions: sharp break at Sobolev 1/6","Cubic NLS non-uniqueness pinned at H^1/6 threshold","Two solutions from nothing: singular NLS threshold at 1/6","Sharp threshold: below H^1/6, cubic NLS fails uniqueness"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000791,"raw_usage":{"total_tokens":3554,"prompt_tokens":1082,"completion_tokens":2472,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":698,"completion_tokens_details":{"reasoning_tokens":2370}},"tokens_in":698,"tokens_out":2472,"duration_ms":17527,"temperature":1.0,"reasoning_tokens":2370,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:46:15.030234+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the series defining $C_s(u)$ on the limit function produced by the induction; divergence for some $s>3$ would invalidate the existence claim. Alternatively, any nonzero singular weak solution with zero initial datum belonging to $C_t^0 H_x^{1/6}$ would directly contradict the sharpness statement.","supporting_citations":[{"cited_title":"Non-unique solutions to the periodic gKdV equation","cited_arxiv_id":"2606.06916","evidence_quote":"Supplies the intermittent slab profile and the signed absolute-Fourier definition of rough nonlinearities that the construction adapts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the unconditional uniqueness theorem at $H^{1/6}$ that makes the threshold sharpness assertion possible."}],"review_version":1}