{"id":"4fe1b0d5-58d1-4027-89ad-970bdb3ca3a0","arxiv_id":"2508.19451","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In d=3, the maximal estimate (1.8) is proved for all s > max{s_d(q), s_d(2β)}, confirming the conjecture; sharp β-ranges are also obtained for d≥4 and d=2.","lead":"This paper proves sharp maximal estimates for the wave equation with orthonormal initial data, settling the conjectured regularity range in dimension three. The result yields optimal Schatten-exponent bounds for pointwise convergence of fermionic density operators.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Prop 2.3 rests on an invalid 'cone transfers ξ-localization to τ' step; the Bernstein inequality used has no temporal support to justify it.","rationale":"The reader identified Proposition 2.3 as the load-bearing assumption, and my reading independently converges on the same point. The specific gap is in the proof of Proposition 2.3: the functions g^k_j are defined only by spatial frequency cutoffs, so they have no temporal frequency support. The proof asserts that the cone geometry transfers the spatial localization to temporal localization, but this transfer is not valid for arbitrary L^2_xL^1_t functions. The subsequent Bernstein inequality and the summation over k depend on this temporal support claim. If this step cannot be repaired, the bilinear estimate fails as proven, and Proposition 3.1—the sharp β=2 estimate—has no support. This makes the central claim conditional on a nontrivial fix. The secondary issue of applying Proposition 2.3 with δ=1/2 outside its stated range is real but easily addressed by extending the proposition's range, so it is not the primary concern. The paper's strategy is plausible and the rest of the structure is coherent, so the appropriate verdict remains CONDITIONAL, matching the reader's assessment.","tokens_in":14329,"tokens_out":7071,"duration_ms":65125,"concrete_test":"Independently prove or disprove the Bernstein step: take g(x,t)=a(x)b(t) with a∈L^2, b∈L^1, and \\hat{b} supported near |τ|=2^N for N≫k; compute \\|g^k\\|_{L^2_xL^2_t} versus \\|g\\|_{L^2_xL^1_t}. If the claimed ≲2^{k/2} fails, revise the proof of Prop 2.3 (e.g., retain the (1+||ξ|±τ|)^{-M} factor and use weighted Cauchy–Schwarz, or insert a temporal cutoff and pay an extra δ factor). Also verify whether the argument can be extended to δ∈(0,1) so that δ=2^{l-k} with l=k-1 is covered.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Proposition 2.3, g^k_j is defined by a spatial Littlewood–Paley cutoff only: \\hat{g^k_j}(ξ,τ)=\\hat{g_j}(ξ,τ)\\phi_{2^k}(|ξ|). The next line claims 'for k≥1, the cone structure transfers the additional localization in ξ to τ' and concludes \\hat{g^k_j} is supported in |ξ|∼|τ|∼2^k. This is false for arbitrary g_j∈L^2_xL^1_t: temporal Fourier support of \\hat{g_j} is unconstrained by spatial localization. The subsequent Bernstein bound \\|g^k_j\\|_{L^2_xL^2_t}≲2^{k/2}\\|g_j\\|_{L^2_xL^1_t} requires Fourier support in |τ|≲2^k; without it, the factor 2^{k/2} can fail. The proof of (3.12) then applies Prop 2.3 with δ=2^{l-k}, which for l=k-1 gives δ=1/2, outside the stated range δ<2^{-2}; this is a smaller, fixable issue, but the temporal-support gap is load-bearing: if Proposition 2.3 cannot be justified, Proposition 3.1 and Theorems 1.4/1.6/1.8 do not follow.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves maximal estimates for orthonormal systems of half-wave equations, i.e., bounds for the L^{q/2}_x L^∞_t norm of the density Σ_j λ_j |e^{it√−Δ}f_j|^2 in terms of ∥λ∥_{ℓ^β} and the Sobolev regularity of the orthonormal family (f_j). The main result, Theorem 1.4, establishes the conjectured sharp threshold in dimension d=3 for all β≥1, up to endpoint. In d≥4 and d=2, Theorems 1.6 and 1.8 give sharp results for β∈[2,∞] and β∈[1,2] respectively, improving on the earlier work [24]. The proof centers on a new bilinear estimate, Proposition 2.3, for a smoothed thickened-cone kernel, and a resulting frequency-localized L^2 estimate, Proposition 3.1, from which the full range follows by duality and interpolation. The paper also includes a careful analysis of the Fourier decay of the conic measure in Lemma 2.2.","tokens_in":14651,"tokens_out":17343,"duration_ms":154070,"significance":"If the main results are correct, they confirm the natural conjecture for orthonormal wave maximal estimates in d=3 up to the endpoint and give the first sharp bounds for the Schatten exponent in several parameter ranges. The approach via a smooth thickened-cone kernel is a genuine methodological novelty compared with the geometric arguments in [24]. The reduction to Proposition 3.1 and the interpolation framework are clean, and Lemma 2.2 is rigorously proved. However, the proof of the load-bearing bilinear estimate, Proposition 2.3, contains an unjustified support-transfer assertion and an improperly justified dyadic summation, so the paper cannot be accepted in its current form.","major_comments":[{"comment":"The assertion after (2.3) that 'for k ≥ 1, the cone structure transfers the additional localization in ξ to τ' and hence that \\hat{g^k_j} is supported in |ξ|∼|τ|∼2^k is false. The functions g^k_j are defined by a spatial Littlewood–Paley cutoff only; for arbitrary g_j∈L^2_xL^1_t the temporal Fourier support is unconstrained. For example, take g_j(x,t)=f(x)h(t) with \\hat f supported on |ξ|∼2^k and \\hat h supported on |τ|∼2^K with K≫k and h∈L^1∩L^2. Then ∥g_j∥_{L^2_xL^1_t}<∞ but \\hat{g^k_j} is supported in |τ|∼2^K, not |τ|∼2^k. Consequently the Bernstein bound ∥g^k_j∥_{L^2_xL^2_t}≲2^{k/2}∥g_j∥_{L^2_xL^1_t} is unjustified and can fail. This step is load-bearing: Proposition 3.1 and Theorems 1.4, 1.6, and 1.8 depend on it. The proof can likely be repaired by inserting a temporal cutoff χ(τ/2^k) using the τ-decay and cone factors in Lemma 2.2 and verifying that the L^2_xL^1_t norm of the trun","section":"Section 2, proof of Proposition 2.3"},{"comment":"The reduction to a single dyadic sum is not justified. Decomposing both g_1 and g_2 yields a double sum Σ_{k,l} over the spatial frequency pieces. The displayed bound after (2.3) is a single sum over k. The cross terms k≠l are not discussed. If the intended device is the support or decay of \\widehat{K^N_δ} (localization in |ξ| and near the cone), that argument should be stated; without it the bound does not follow. This is a technical but necessary step in the proof.","section":"Section 2, proof of Proposition 2.3 (dyadic summation)"},{"comment":"Proposition 2.3 is applied with δ=2^{l-k}. For l=k-1 this gives δ=1/2, which is outside the range 0<δ<2^{-2} stated in Proposition 2.3. The proof of the proposition appears to work for any δ<1 with a minor adjustment of the threshold L, but as stated the application is outside the hypothesis. The authors should either extend the proposition to the full range δ<1 or handle the case l=k-1 separately.","section":"Section 3.2, Eq. (3.12)"}],"minor_comments":[{"comment":"The inclusion 'ℓ^{q/2} ⊂ ℓ^β' appears reversed. When q/2 ≥ β, the correct inclusion is ℓ^β ⊂ ℓ^{q/2}. The argument still works, but the statement should be corrected.","section":"Section 3, first paragraph"},{"comment":"The remark contains an unresolved citation '[?, Proposition 3.1]'. Please replace it with a proper reference.","section":"Remark after Proposition 2.3"},{"comment":"The notation for the Fourier transform of the kernel is inconsistent (\\widehat{KN_δ} vs \\widehat{K^N_δ}). Please unify.","section":"Proof of Lemma 2.2"},{"comment":"The notation φ^2_{2^k}(|ξ|) is ambiguous; it should mean φ_{2^k}(|ξ|)^2.","section":"Section 3.2, Eq. (3.8)"}],"recommendation":"major_revision","confidential_remarks":"The paper reports a significant advance and the main results are plausible. The referee's main concern is the proof of Proposition 2.3, specifically the unjustified temporal-support transfer step; this is load-bearing because Proposition 3.1 and the main theorems rely on it. The estimate itself may be salvageable with a temporal-cutoff argument, and the dyadic-summation and δ-range issues are fixable. This is a major revision rather than a rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline result is real: Ko, Lee, and Shiraki confirm the conjectured sharp regularity for orthonormal wave maximal estimates in d=3, and get best-known results in d≥4 (β∈[2,∞]) and d=2 (β∈[1,2]). The new ingredient is a β=2 estimate via the Fourier decay of the conic measure, which is a genuine improvement over the geometric approach in [24]. If the proof stands, it settles a natural open question.\n\nWhat the paper does well: the high-level strategy is clear. The reduction to β=2 by interpolation is standard but correctly executed. Lemma 2.2 is a nice kernel estimate, and the use of Plancherel is sensible. The improvements are not incremental; they reach the conjectured threshold in the main case.\n\nThe soft spot is serious. Proposition 2.3 is the load-bearing new input, and its proof is not just compressed—one step is wrong. The functions g^k_j are defined by a spatial Littlewood–Paley cutoff only. The text says the cone structure transfers the ξ-localization to τ, so \\hat{g^k_j} is supported in |τ|~2^k. For arbitrary g_j∈L^2_xL^1_t this is false; the temporal Fourier support is unconstrained. The subsequent Bernstein bound \\|g^k_j\\|_{L^2L^2}≲2^{k/2}\\|g_j\\|_{L^2L^1} requires temporal support (or something like it) and can fail. This affects Proposition 3.1 and therefore the main theorems. The smaller issue is that (3.12) applies Proposition 2.3 with δ=2^{l-k}, which for l=k-1 gives δ=1/2, outside the stated range; that looks fixable.\n\nI suspect the bilinear estimate is true and can be proved by a different argument, perhaps using the L^∞_τ bound from the L^1_t norm rather than trying to localize \\hat{g} in τ. But as written, the proof doesn't establish it.\n\nBottom line: the paper is important and likely correct, but it needs a serious revision of Section 2. Send it to a qualified referee; don't desk reject. The referee should press hard on Proposition 2.3. If that gets fixed, this is a publishable result at a good journal.","headline":"Sharp d=3 orthonormal wave maximal estimates, but the key bilinear estimate has a proof gap that needs fixing.","tokens_in":15137,"tokens_out":4897,"would_cite":true,"duration_ms":49114,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L05","35B65"],"pacs":[],"model":"deepseek-v4-flash","headline":"In dimension 3, this paper confirms the conjectured sharp regularity threshold for maximal estimates that control pointwise convergence for orthonormal systems of wave equations, and it extends sharp ranges in dimensions 2 and ≥4.","keywords":["wave equation","orthonormal systems","maximal estimates","Schatten classes","pointwise convergence","sharp regularity","bilinear cone estimates","Bessel asymptotics"],"falsifier":"Compute B_δ on the pair its dyadic step is designed to make sharp: g1,g2 with Fourier transforms on unit-thick cone pieces |τ|=|ξ| at scale |ξ|∼δ^{-1}, τ-supports just touching. If the true d=3 bound is δ^{1/2}‖g1‖‖g2‖ (the earlier geometric method's rate) rather than δ‖g1‖‖g2‖, Proposition 2.3 is false and Theorem 1.4 does not follow. A separate algebraic check settles the range issue: does the claimed decay still hold at δ=1/2, the value used in (3.12), which lies outside the stated range δ<2^{-2}?","tokens_in":14241,"feed_emoji":"🌊","tokens_out":19012,"duration_ms":167694,"temperature":0.7,"pith_summary":"This paper addresses a quantum many-body question: for an infinite system of non-interacting fermions evolving under the wave equation, does the particle density converge pointwise to its initial value as t→0? The controlling object is a maximal estimate for sums of squares of wave solutions with orthonormal initial data, indexed by a summability parameter β; the conjectured sharp condition is that the Sobolev regularity s exceed max{s_d(q), s_d(2β)}, where s_d(σ)=max{d/2−d/σ,(d+1)/4−(d−1)/(2σ)}. In dimension 3, the paper proves this conjecture for every β≥1, up to the endpoint (strict inequality). For d≥4 it proves the sharp range β∈[2,∞], and for d=2 the sharp range β∈[1,2], improving the previously known non-sharp results in both cases. The key new input is a bilinear estimate for a thickened-cone kernel that loses only one power of δ in d≥3, and half a power in d=2, obtained by exploiting the Fourier decay of the conic measure rather than the spatial geometry of cone intersections.","feed_headline":"Sharp regularity for wave maximal estimates confirmed in 3-D","feed_subtitle":"The conjectured threshold for pointwise convergence of many-fermion densities is now reached in dimension 3.","key_machinery":"The engine is a bilinear estimate (Proposition 2.3) for the thickened-cone kernel K^N_δ: the form B_δ(g1,g2)=∫∫∫∫ g1 g2 K^N_δ(x−x′,t−t′) is bounded by δ^{θ_d}‖g1‖_{L²_xL¹_t}‖g2‖_{L²_xL¹_t}, with θ_d=1 for d≥3 and θ_d=1/2 for d=2. The δ-power comes from Lemma 2.2, which reads the kernel's Fourier transform off the conic measure's Bessel asymptotics, replacing the spatial cone-intersection analysis of earlier work (δ^{1/2} in d=3). Around it: a duality principle (Proposition 2.1) turning the orthonormal maximal estimate into a Schatten-2 bound on W T_k W, Littlewood–Paley localization, and a spatial decomposition feeding each kernel piece into B_δ at δ=2^{l−k}. Interpolation of the β=1, β=2, a","core_discovery":"The paper's central claim is Theorem 1.4: in dimension 3, the maximal estimate for orthonormal wave data holds at every regularity s above the conjectured threshold max{s_d(q), s_d(2β)}, for every β≥1 and q≥2. Since earlier counterexamples force s≥s_d(2β), the sharp regularity exponent is pinned down up to the endpoint. The decisive case is β=2: the frequency-localized maximal estimate with ℓ² weights holds for s>max{d/4,5/8}, exactly the conjectured critical value in d=3. Interpolating with the elementary β=1 and β=∞ bounds yields the full β-range; d=3 is complete because the regime-change exponent β*=(d+1)/(d−1) equals 2 there. The same argument gives sharp ranges for β∈[2,∞] in d≥4 and β∈","pith_inferences":["Since the β=2 estimate is already sharp in every d≥3, the remaining gaps (β<2 in d≥4, β>2 in d=2) look like an artefact of interpolation: a direct ℓ^β argument that bypasses the β=1/β=2/β=∞ interpolation should close them — an extension the paper does not make.","Proposition 2.3 is stated for δ<2^{-2}, but the proof of Proposition 3.1 applies it at δ=2^{l−k}, which can reach 1/2; a checkable technical question is whether the estimate holds uniformly for δ≤1/2 with the usual δ^{-ε} losses, which would remove the only evident mismatch between the stated lemma and its use.","The endpoint s=s_d(2β) remains open even in d=3; an endpoint (restricted weak-type) version of the bilinear estimate would be the natural route to the critical case, which this paper does not address.","The Fourier-decay approach does not rely on the low-dimensional cone-intersection geometry that limited earlier work, so the same kernel analysis may transfer to half-wave-type propagators on curved hypersurfaces or to other cone-constrained dispersive equations — consequences the paper leaves implicit."],"forward_implications":["In d=3 the maximal estimate (1.8) holds for every β≥1 and q≥2 at any regularity strictly above the conjectured threshold, so pointwise convergence of densities (1.6) follows for self-adjoint initial states in the β-range of Corollary 1.5.","In d≥4 the sharp range β∈[2,∞], and in d=2 the sharp range β∈[1,2], are established, strictly broadening the previously known non-sharp results.","The single β=2 estimate (Proposition 3.1) is sharp in every d≥3, and all other β-values are obtained from it by interpolation with the elementary β=1 and β=∞ bounds.","All results hold with strict inequality; the critical case where s equals s_d(2β) (including the endpoint regime) is not settled.","The paper remarks that the same Fourier-side analysis also yields an alternative proof of the orthonormal wave Strichartz estimate at β=2."],"supporting_citations":[{"why":"Provides the duality principle in the L^q_xL^r_t form used here, together with the orthonormal Strichartz framework the remark in Section 2 extends.","marker":"[3]"},{"why":"Establishes the density-operator framework and the L^1_xL^∞_t form of the maximal estimate, fixing the norm used in the main theorems.","marker":"[5]"},{"why":"Formulates the pointwise convergence problem for orthonormal systems and the maximal estimate (1.8) that the paper targets.","marker":"[6]"},{"why":"Cowling's single-particle wave maximal estimate at q=2, s>1/2 — the β=1 interpolation endpoint.","marker":"[14]"},{"why":"Supplies the duality principle (Proposition 2.1) that turns the orthonormal maximal estimate into a Schatten-norm bound — the backbone of the proof.","marker":"[20]"},{"why":"The prior work this paper improves: it proved the necessity of s≥s_d(2β) by counterexample and gave non-sharp sufficiency in d=2,3,4 with weaker δ^{1/2}-type bounds.","marker":"[24]"},{"why":"Walther's sharpness of s>1/2 and his L^p(L^∞) oscillatory estimates — the β=1 endpoint and its optimality.","marker":"[30]"}],"fun_headline_variants":["3-D wave maximal estimates hit sharp regularity","Sharp exponent for orthonormal wave systems in 3-D","Optimal regularity for wave maximal bounds in d=3","Wave equation: endpoint regularity for orthonormal data","3-D achieves conjectured wave maximal threshold"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The paper's results stand on one new estimate: an integral over two space–time copies against a kernel hugging the light cone is bounded by δ (d≥3) or √δ (d=2) times the input norms, where δ is the cone's thickness — if that decay is actually weaker, the sharp β=2 estimate and all three main theorems fail. Its proof in Section 2 is compressed: the dyadic summation is sketched, the cone-geometry transfer of τ-localization to ξ is asserted rather than shown, and the estimate is","fun_headline_variants_meta":{"raw":{"variants":["3-D wave maximal estimates hit sharp regularity","Sharp exponent for orthonormal wave systems in 3-D","Optimal regularity for wave maximal bounds in d=3","Wave equation: endpoint regularity for orthonormal data","3-D achieves conjectured wave maximal threshold"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000163,"raw_usage":{"total_tokens":1057,"prompt_tokens":702,"completion_tokens":355,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":446,"completion_tokens_details":{"reasoning_tokens":280}},"tokens_in":446,"tokens_out":355,"duration_ms":4559,"temperature":1.0,"reasoning_tokens":280,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T15:47:56.030991+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute B_δ on the pair its dyadic step is designed to make sharp: g1,g2 with Fourier transforms on unit-thick cone pieces |τ|=|ξ| at scale |ξ|∼δ^{-1}, τ-supports just touching. If the true d=3 bound is δ^{1/2}‖g1‖‖g2‖ (the earlier geometric method's rate) rather than δ‖g1‖‖g2‖, Proposition 2.3 is false and Theorem 1.4 does not follow. A separate algebraic check settles the range issue: does the claimed decay still hold at δ=1/2, the value used in (3.12), which lies outside the stated range δ<2^{-2}?","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the duality principle in the L^q_xL^r_t form used here, together with the orthonormal Strichartz framework the remark in Section 2 extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the density-operator framework and the L^1_xL^∞_t form of the maximal estimate, fixing the norm used in the main theorems."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Formulates the pointwise convergence problem for orthonormal systems and the maximal estimate (1.8) that the paper targets."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Cowling's single-particle wave maximal estimate at q=2, s>1/2 — the β=1 interpolation endpoint."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the duality principle (Proposition 2.1) that turns the orthonormal maximal estimate into a Schatten-norm bound — the backbone of the proof."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The prior work this paper improves: it proved the necessity of s≥s_d(2β) by counterexample and gave non-sharp sufficiency in d=2,3,4 with weaker δ^{1/2}-type bounds."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Walther's sharpness of s>1/2 and his L^p(L^∞) oscillatory estimates — the β=1 endpoint and its optimality."}],"review_version":1}