{"id":"9ebb99c9-042b-4e1f-9228-48012d17a8dd","arxiv_id":"2604.00252","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Subtracting the mean from the particle density improves orthonormal Strichartz estimates and shifts the optimal well-posedness threshold for the cubic NLS system on the circle from Schatten exponent 1 to 2.","lead":"This paper introduces a renormalised particle density for a system of nonlinear Schrödinger equations on a circle and proves it satisfies stronger estimates than the usual density. It then uses these estimates to show the renormalised system is well-posed for a wider class of initial data than the original system.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The well-posedness application for (RNLSS) is not self-contained: Lemma 4.8 and Lemma 4.10 depend on two technical results from the unpublished preprint [6]; the core renormalised Strichartz estimates (Theorems 1.12, 1.14) appear sound.","rationale":"The paper's main contribution is Theorem 1.12 and Theorem 1.14; I read these proofs as consistent. Lemma 3.2 is a standard Fourier argument; the counting estimate in Lemma 3.3, while intricate, appears correct; the interpolation to obtain the L^3 threshold is valid. The application is where the proof is exposed. Proposition 4.2(i) is proved by a contraction on rho in L^2_{t,x}, using Lemma 4.8 for the a priori L^2 bound on the renormalised density and Lemma 4.10 for the difference. Both lemmas rely, without proof, on two results from [6], an unreviewed arXiv preprint by (among others) the first author. This is a genuine gap in self-containedness: the advertised 'critical Schatten exponent' depends on external results. The reader's specific complaint about alpha<2 is not correct: since S^alpha is contained in S^2 for alpha<2, the S^2 solution automatically has the stated regularity; the real issue is that the S^2 contraction must first be valid. The abstract's 'minimal improvement' claim for d>=2 is also overstated: Proposition 5.5 only provides a necessary condition (1/alpha >= 1 - sigma/(d-1)) that is weaker than the non-renormalised threshold, not a sharp converse. This is secondary but worth noting. Overall, the verdict should remain CONDITIONAL: the core estimates may be correct, but the well-posedness theorem needs either a verification of [6] or a self-contained proof.","tokens_in":29582,"tokens_out":25628,"duration_ms":211442,"concrete_test":"Independently inspect arXiv:2504.19552: verify the statements and proofs of [6, Cor. 4.9] and [6, Lemma 4.1]. In particular, check whether [6, Cor. 4.9] legitimately implies (4.7), i.e., that the operator with integral over [0,t] is bounded in S^2(L^2_{t,x}->L^2_x) by the same expression with [0,T], and whether [6, Lemma 4.1] applies to the double integral in Lemma 4.8. If either is invalid, or requires extra assumptions (e.g., trace-class gamma0 or V in a smaller space), the contraction argument fails. Alternatively, attempt to prove Lemma 4.8 using only Lemmas 3.2, 4.4, and 4.5; failure suggests the proof is not self-contained.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The renormalised-density estimates in Theorems 1.12 and 1.14 are argued carefully and appear sound. The load-bearing weakness is in the advertised application: Proposition 4.2 (the precise form of Theorem 1.19) constructs the fixed point for (RNLSS) using Lemma 4.8 (density bound for U_V) and Lemma 4.10 (difference estimate). Both proofs invoke [6, Cor. 4.9] and [6, Lemma 4.1] — technical Schatten-space estimates from an unreviewed preprint co-authored by one of the present authors — to replace a full time integral by a truncated one and to bound a double integral in S^2 operator norms. The paper gives no proof of these inputs and no alternative derivation. If either [6] result is false, or if its hypotheses are not satisfied in the present setting (e.g., with V in L^2_{t,x} and gamma0 in S^2), then Lemma 4.8/4.10 fail and the existence of the S^2 fixed point, hence the critical exponent alpha=2, is unsupported. The alpha<2 half of Theorem 1.19 would then also be unjustified, since it relies on the same S^2 solution. (Note: the reader's stated concern that S^2 well-posedness does not extend to alpha<2 is not the issue — S^alpha is contained in S^2 for alpha<2, so the S^2 solution automatically lies in S^alpha; the issue is entirely upstream.)","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a renormalised density for the NLS system on the circle and proves orthonormal Strichartz estimates for it. The main new estimates are Theorem 1.12, an L^2_{t,x} bound for the renormalised density of U(t)γ0U(t)* with sharp Schatten threshold α=2, and Theorem 1.14, an L^3_{t,x} estimate for the frequency-truncated renormalised density with sufficient conditions for α∈[2,3] and a matching necessary condition for α∈[3/2,3]. The authors apply these estimates to the cubic NLS system: Proposition 4.1 gives well-posedness in S^1 for the unrenormalised system and ill-posedness for α>1, while Proposition 4.2 gives S^2 well-posedness for the renormalised system and ill-posedness for α>2, presented in the introduction as the critical-exponent result Theorem 1.19. A final section gives an alternative proof of a torus result of Nakamura and a necessary condition showing that in d≥2 the renormalised density yields only a small improvement over the non-renormalised one.","tokens_in":29977,"tokens_out":15113,"duration_ms":136533,"significance":"If the central estimates are correct, the paper makes a genuine contribution: the renormalisation of the density is a natural and useful device, and the sharp L^2 threshold α=2 on T is a clear improvement over the non-renormalised α=1 result. The L^3 estimate in Theorem 1.14 is technically interesting, and the counting argument behind Lemma 3.3 is elegant. The paper also gives a clean alternative proof of Nakamura's torus result and a transparent counterexample in higher dimensions. These parts are strong and, as far as I have checked, the proofs of Theorems 1.12 and 1.14 are self-contained and convincing. However, the advertised optimal well-posedness application is not fully established within the manuscript: two key lemmas in Section 4 depend on an unpublished preprint, and the precise statement proved in Proposition 4.2 is weaker in two respects than the theorem announced in the introduction.","major_comments":[{"comment":"Both lemmas are load-bearing for Proposition 4.2: Lemma 4.8 supplies the L^2 density bound for U_V, and Lemma 4.10 supplies the difference estimate needed for the contraction argument. The proof of Lemma 4.8 explicitly invokes [6, Cor. 4.9] and [6, Lemma 4.1] to replace a full time integral by a truncated one and to bound a double integral in S^2; Lemma 4.10 again invokes [6, Cor. 4.9]. These are results from an unpublished preprint, arXiv:2504.19552, and are not proved or stated in the present paper. Since the fixed-point argument for (RNLSS) has no independent estimate without these inputs, the well-posedness claim is conditional on results outside the manuscript. Please either include proofs of the two [6] results in an appendix or explicitly state Proposition 4.2 as conditional on [6].","section":"§4.1, Lemmas 4.8 and 4.10"},{"comment":"Theorem 1.19 advertises global well-posedness in S^α for every α∈[1,2], but Proposition 4.2(i), cited as the precise statement, only proves existence, uniqueness, and Lipschitz dependence in S^2. No argument is given that the data-to-solution map is continuous in the S^α topology for α<2, nor that the S^2 solution obtained for γ0∈S^α is the unique solution in C_t S^α. The inclusion S^α⊂S^2 and unitarity of U_V make a repair plausible, but the statement as written is not proved. The introduction should either restrict the well-posedness claim to S^2 or include the S^α continuity argument.","section":"§1.4, Theorem 1.19 vs §4.2, Proposition 4.2(i)"},{"comment":"The ill-posedness half of Theorem 1.19 is stated in terms of discontinuity of the solution map, but Proposition 4.2(ii) proves only that the data-to-density map S_2:E^α_R→L^2_{t,x} has no continuous extension. This is a weaker property: the sequence γ0,N has S^α norm tending to 0 while the associated density has L^2 norm bounded below, which obstructs continuity of γ0↦ργ but does not immediately obstruct continuity of γ0↦γ(t) in C_t S^α. If 'ill-posed' is intended in the usual flow-map sense, an additional argument is needed; otherwise the theorem should be reformulated to define ill-posedness via the density map, as Proposition 4.2 does.","section":"§4.2, Proposition 4.2(ii) vs Theorem 1.19 ill-posedness"}],"minor_comments":[{"comment":"In the expansion of Lemma 4.8, the second term is printed as 'ϱ_1 + ϱ_1 + ϱ_2'; presumably the two ϱ_1 terms are ϱ_1 and its conjugate (or DγU* and UγD*). The notation should be cleaned up to avoid ambiguity.","section":"§4.1, Eq. (4.6)"},{"comment":"The introduction says the improvement for d≥2 is 'minimal', but Proposition 5.5 only establishes a necessary condition, leaving a gap between 1/α ≥ 1−σ/(d−1) and the known sufficient condition 1/α > 1−σ/d. The wording 'minimal' overstates what is proved; 'the improvement is at most of order σ/(d(d−1))' would be more accurate.","section":"§5.2, Proposition 5.5"},{"comment":"The remark that proving the α=3/2 case with σ>0 suffices for the conjecture is not justified in the text. The interpolation argument should be indicated, or the remark removed.","section":"§3.3, Conjecture 1.16"},{"comment":"The necessity part of Proposition 1.6 is asserted without proof, with a reference to [31]. Since the proposition is stated as a result of this paper, it would be helpful to include the short counterexample or make the reference more precise.","section":"§1.3.1, Proposition 1.6"}],"recommendation":"major_revision","confidential_remarks":"The core Strichartz estimates appear sound and publishable, but the advertised critical-exponent application is not self-contained and the theorem statements in the introduction are stronger than the propositions proved. The dependence on [6] should be resolved before publication; at minimum, the authors should prove the two cited results from [6] or explicitly mark the well-posedness theorem as conditional. The mismatch between Theorem 1.19 and Proposition 4.2 on the α<2 range and on the ill-posedness notion also needs to be fixed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The core of the paper is the renormalised density estimate on T: Theorem 1.12 (L2, S^2) and Theorem 1.14 (L3 with N^σ, essentially the expected interpolation threshold). These are new, the Fourier/diophantine counting argument in Lemma 3.3 is clean, and the sharpness examples are convincing. The renormalisation trick — subtracting the spatial mean before duality — is simple and effective; it genuinely lowers the Schatten exponent from 1 to 2 for the L2 density estimate.\n\nThe application to (RNLSS) is more delicate. Proposition 4.2 proves the fixed point in S^2, and the ill-posedness for α>2 is fine. The introduction's Theorem 1.19 advertises well-posedness for α∈[1,2]; the reader worried this wasn't proven, but that worry is misplaced: for α<2, S^α embeds into S^2, the flow is unitary, so the S^2 solution is automatically an S^α solution with the same Schatten norm. That part is a free corollary.\n\nThe real soft spot is self-containment, not scope. Lemma 4.8 and Lemma 4.10, which carry the well-posedness proof, rely on two technical statements from the unpublished preprint [6] by one of the authors. If those statements are false or don't apply, the S^2 fixed point argument collapses. I have no reason to think they're wrong, but a referee should ask for proofs or for the preprint to be posted/accepted before this is published. This is a moderate weakness, not a fatal one, because the central estimates are independent.\n\nOne more: the 'minimal improvement' claim for d≥2 is based only on the necessary condition in Proposition 5.5. The example is a good negative result, but 'minimal' is a bit stronger than 'we can't improve much.' Fine as a remark, not as a theorem.\n\nBottom line: the paper deserves a serious referee. The main estimates are new, correct-looking, and potentially useful for many-fermion problems. For anyone working on orthonormal Strichartz estimates or density-matrix NLS, this is the paper to read. I'd send it out, with a request to either prove the [6] inputs or make the dependency explicit and checked. I'd cite the L2/L3 estimates even before the application issue is resolved.","headline":"New renormalised-density Strichartz estimates on the circle that look right; the advertised NLS application has a self-containment gap, but the main claims hold up better than the reader's report suggests.","tokens_in":30459,"tokens_out":7928,"would_cite":true,"duration_ms":75326,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42B37","35Q55","35B45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that subtracting the spatial mean from the density — a renormalised density — strictly improves orthonormal Strichartz estimates on the circle, yielding an L² estimate up to the Hilbert–Schmidt class and an L³ estimate in","keywords":["Orthonormal Strichartz estimates","Renormalised density","Cubic NLS system","Schatten classes","Global well-posedness","Ill-posedness","Torus","Dispersive estimates"],"falsifier":"Take γ_{0,N}=N^{−1/2}Σ_{n=1}^N |φ_n><φ_n| with φ_n=2^{−1/2}(e_n+e_{−n}); the paper's own computation gives ∥ρ(U(t)γ_{0,N}U*(t))∥_{L²}∼1 while ∥γ_{0,N}∥_{S^α}→0 for α>2, proving sharpness of the L² estimate. To test the missing well-posedness range, compute the contraction constant of the map V↦ρ(U_{±V}(t)γ0U*_{±V}(t)) on S^{3/2} for small T: a Lipschitz constant >1 for all small T would show the S²-based proof cannot reach α<2.","tokens_in":29475,"feed_emoji":"🌀","tokens_out":6382,"duration_ms":55915,"temperature":0.7,"pith_summary":"The paper introduces a renormalised density on the circle: subtract the spatial mean, equivalently a multiple of the trace, from the usual density. Because constants commute with everything, the renormalised density generates the same NLS system for trace-class states, but it satisfies better orthonormal Strichartz estimates than the standard density. The paper proves an L² estimate for all Hilbert–Schmidt data and an L³ estimate for Schatten exponents between 2 and 3, with the remaining lower range left as a conjecture. As an application, it claims the cubic renormalised NLS system on the circle is globally well-posed for Schatten exponents up to 2 and ill-posed above 2. On tori of dimension at least two, the same renormalisation is shown to yield almost no improvement over the non-renormalised estimates.","feed_headline":"Renormalised density sets NLS system critical exponent at 2","feed_subtitle":"Subtracting the mean from the density upgrades orthonormal Strichartz estimates and marks where well-posedness fails.","key_machinery":"The key object is the renormalised density ρA(x)=ρA(x)−(1/2π)TrA, defined for trace-class A and extended by density to S². Its value is that subtracting the constant removes the zero-frequency Fourier mode; duality then only needs test functions V with vanishing space-time mean, and Parseval's identity turns the L² estimate into a bound on an integral operator with kernel e^{−it(m²−n²)} whose zero-frequency term is absent. For the L³ estimate, the same mean-zero reduction converts the problem into counting solutions of a Diophantine equation ks−(k−α)s′=β, whose divisor-counting bound yields the N^{1+ε} estimate. For well-posedness, the propagator U_V(t,s) for a real L² potential and its Stri","core_discovery":"The central discovery is that the renormalised density ρ(U(t)γ0U*(t)) = ρ(U(t)γ0U*(t)) − (1/2π)Trγ0 obeys an L²_{t,x} bound on the two-torus for every Hilbert–Schmidt operator γ0, sharp in the sense that the bound fails for α>2, and a frequency-truncated L³ bound with exponent σ>2/3−1/α for α∈[2,3]. This improves the non-renormalised L³ condition, which required roughly 1/α>1−σ. Interpreting the Hilbert–Schmidt case via the unique bounded extension from trace-class data, the paper uses these estimates to build the solution map for the renormalised cubic NLS system by a fixed point in the density, yielding global well-posedness in S^α for α≤2 and discontinuity of the data-to-density map for α","pith_inferences":["A decisive open test is whether the fixed-point contraction in the paper's Proposition 4.2 can be run in S^α for 1≤α<2; if it cannot, Theorem 1.19's well-posedness range may shrink to the proved S² statement even though the density estimates themselves hold in that range.","Because the proof of the key density estimates for the potential propagator relies on a cited unpublished preprint by one of the authors, a reader who wants to rely on the application should check whether those ingredients can be replaced by published arguments; this is a reproducibility question, not a mathematical one.","The mean-zero duality trick is not specific to the circle; a natural test is whether a similar renormalisation recovers a full range of orthonormal Strichartz estimates on other compact manifolds with a spectral gap.","For the quintic NLS system, the paper notes that renormalisation produces an extra cubic term; a plausible extension is that the well-posedness threshold shifts and is governed by the cubic correction, which could be tested by the same fixed-point scheme."],"forward_implications":["If Theorem 1.19 is correct, the cubic renormalised NLS system on the circle has a sharp dichotomy: global well-posedness in S^α for 1≤α≤2 and ill-posedness for α>2, with the solution map discontinuous above the threshold.","The L² theorem gives a genuine extension of the density to Hilbert–Schmidt operators, so the ill-defined diagonal of the kernel is handled by a bounded operator construction.","The L³ threshold σ>2/3−1/α for α∈[2,3], together with the paper's counterexamples below the threshold, fixes the growth estimate for the frequency-truncated renormalised density on the circle.","On T^d with d≥2, the necessary condition 1/α≥1−σ/(d−1) means renormalisation does not substantially improve the orthonormal Strichartz range in high dimension.","If Conjecture 1.16 is true, the L³ estimate would also hold for α∈[3/2,2) with σ>2/3−1/α, completing the admissible range down to α=3/2."],"fun_headline_variants":["Renormalised NLS: global well-posed for α≤2","Density renormalisation sharpens NLS threshold to α=2","Renormalised density yields sharp NLS well-posedness exponent","Subtract mean density to get NLS critical exponent 2","Renormalised density fixes NLS well-posedness cutoff at α=2"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the contraction argument for the renormalised system, which is written only for γ0∈S² and relies on a cited unpublished preprint for a key Strichartz estimate, extends to all Schatten classes S^α with 1≤α<2; if that extension is not possible, the stated critical exponent 2 is unsupported.","fun_headline_variants_meta":{"raw":{"variants":["Renormalised NLS: global well-posed for α≤2","Density renormalisation sharpens NLS threshold to α=2","Renormalised density yields sharp NLS well-posedness exponent","Subtract mean density to get NLS critical exponent 2","Renormalised density fixes NLS well-posedness cutoff at α=2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000516,"raw_usage":{"total_tokens":2325,"prompt_tokens":715,"completion_tokens":1610,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":1515}},"tokens_in":459,"tokens_out":1610,"duration_ms":11602,"temperature":1.0,"reasoning_tokens":1515,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T16:59:57.267322+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take γ_{0,N}=N^{−1/2}Σ_{n=1}^N |φ_n><φ_n| with φ_n=2^{−1/2}(e_n+e_{−n}); the paper's own computation gives ∥ρ(U(t)γ_{0,N}U*(t))∥_{L²}∼1 while ∥γ_{0,N}∥_{S^α}→0 for α>2, proving sharpness of the L² estimate. To test the missing well-posedness range, compute the contraction constant of the map V↦ρ(U_{±V}(t)γ0U*_{±V}(t)) on S^{3/2} for small T: a Lipschitz constant >1 for all small T would show the S²-based proof cannot reach α<2.","supporting_citations":[],"review_version":1}