{"id":"7bfc0097-f205-4190-92c7-6cb751cba0f9","arxiv_id":"1908.04815","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In dimensions n≥62, there is a smooth metric on the spherical cap with positive Yamabe constant for which lower-energy positive solutions to the constant scalar curvature and constant boundary mean curvature equation exist but fail to be compact.","lead":"This paper constructs a metric on a sphere with boundary such that solutions to a central geometric PDE, the constant scalar curvature with constant boundary mean curvature equation, blow up in dimensions 62 and higher. It also gives a simple example where the equation admits more than one solution.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.2 applies the single-bump Proposition 6.1 to the summed metric without proving that cut-off tails and neighboring bumps are negligible in Corollary 4.6; the stated overlap radius is miscomputed, but the missing interaction estimate is a genuine gap.","rationale":"The reader correctly identifies that Theorem 6.2 invokes Proposition 6.1 for a metric that is not literally the single-bump metric of that proposition. However, the quantitative statement about overlapping supports is mistaken: the cut-off χ(4N^2|x-x_N|) gives support radius 1/(2N^2), not 0.7/N, so the overlap of neighboring supports is only a very thin annulus of width O(N^{-4}). Thus the reader's specific numerical claim does not land. The underlying concern does land in a weaker form: the proof does not contain an estimate showing that the transition tails of the N-th bump and the contributions of all other bumps to the Lyapunov-Schmidt functional F_g are smaller than the leading term λ^4 μ^2 F(ξ,ε). Because the bubble width λ=2^{-N/2} is exponentially smaller than the spacing 1/N^2 between centers, such an estimate is very plausible and likely obtainable by repeating the estimates of Section 4 with an extra error term. But as written, Theorem 6.2 is not a direct logical consequence of Proposition 6.1, so the paper requires a small but real amendment. The same gap does not affect the non-uniqueness example in Section 2, which is independent. Therefore the appropriate verdict remains CONDITIONAL: the main construction is probably correct, but the proof needs an additional interaction estimate or a revised Proposition 6.1 that covers the summed metric.","tokens_in":30084,"tokens_out":16294,"duration_ms":160027,"concrete_test":"Compute, for n=62, λ=2^{-N/2}, ρ=(2N)^{-2}, μ=2^{-N}, the difference |F_g(λξ,λε) - λ^4 μ^2 F(ξ,ε)| for the summed metric of Theorem 6.2, separating the contribution of the cut-off tail (where χ'≠0 and r>ρ) and of the neighboring bumps h_M, M≠N. Verify that this difference is bounded by the right-hand side of Corollary 4.6 (with d=1) plus an explicitly negligible term, e.g. o(μ^2 λ^4) as N→∞. If the extra term is instead comparable to λ^4 μ^2, then Theorem 6.2 is unsupported; if it is negligible, the gap is confirmed fillable.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim (Theorem 1.1) rests on Theorem 6.2, whose metric is h_ab = sum_N χ(4N^2|x-x_N|) 2^{-N} f(2^N|x'-x_N|^2) H_ab(x-x_N). Proposition 6.1, the only existence result used, assumes h is a single bump with h=0 outside B_ρ and parameters λ, ρ, μ. In Theorem 6.2 this proposition is invoked 'with λ=2^{-N/2}, ρ=(2N)^{-2}, μ=2^{-N}' although the actual h is an infinite sum. The reader's overlap estimate is off: the cut-off χ(4N^2|x-x_N|) has support radius 1/(2N^2), not 0.7/N, so neighboring supports overlap only in a sliver of width O(N^{-4}). But the logical gap remains: even if the overlap is tiny, the paper does not estimate the contribution of the transition annulus and of the other terms h_M (M≠N) to the functional F_g in Corollary 4.6, nor does it check that the full h satisfies the hypotheses of Proposition 6.1. Since the bubble width λ=2^{-N/2} is much smaller than the inter-center distance ~1/N^2, such an estimate should hold, but it is not written. Without it, Theorem 6.2 is not a direct consequence of Proposition 6.1.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper has two main results. First, Proposition 2.1 constructs a warped product manifold with boundary on which the conformal equation (1.1) with constant positive scalar and boundary mean curvature terms has at least two positive solutions. Second, for n≥62, Theorem 1.1 claims existence of a smooth metric on S^n_+ with positive Yamabe constant, non-conformally-flat, umbilic boundary, and a sequence of positive solutions to (1.1) with c1,c>0, energies below the single-bubble threshold S_c, and boundary sup-norms blowing up. The proof follows the Brendle-Almaraz strategy: Lyapunov-Schmidt reduction on R^n_+ (Section 3), expansion of the reduced energy F_g (Section 4), construction of an auxiliary function F with a strict local minimum at (0,1) via a carefully chosen quadratic polynomial (Section 5), and a perturbation argument (Section 6). The non-compactness conclusion is obtained by applying a single-bubble existence result (Proposition 6.1) to each of infinitely many local bumps and summing them into one metric h.","tokens_in":30378,"tokens_out":30187,"duration_ms":261083,"significance":"If the proof were complete, Theorem 1.1 would be a substantial result: it gives the first non-compactness for the positive-constant scalar curvature and constant boundary mean curvature equation in dimensions n≥62, complementing the compactness results of Han-Li and the boundary blow-up results of Almaraz for the scalar-flat case. The explicit construction in Section 5—choosing the polynomial coefficient a0 via the discriminant of a quadratic and verifying I(1)>0, I'(1)=0, I''(1)<0, J(1)<0—is concrete and checkable, and the energy expansions in Section 4 are detailed. The main theorem, however, is not established by the current argument because Theorem 6.2 applies a single-bump proposition to an infinite-sum metric without the required interaction estimates.","major_comments":[{"comment":"The proof invokes Proposition 6.1 for the infinite-sum metric h(x)=Σ_N χ(4N^2|x-x_N|)2^{-N}f(2^N|x'-x_N|^2)H(x-x_N). Proposition 6.1 requires h to vanish identically outside B_ρ^+ and to equal μλ^2 f(λ^{-2}|x'|^2)H(x) inside B_ρ^+. The summed h fails this: the N-th bump has support radius about (2N^2)^{-1}, while neighboring centers are separated by about N^{-2}, so supports overlap, and the other summands as well as the cutoff annulus are never estimated in the expansion of F_g(λξ,λε) from Corollary 4.6. Without such estimates, the existence of solutions for the summed metric does not follow from Proposition 6.1, so Theorem 1.1 is not proven by the stated argument.","section":"Section 6 (Theorem 6.2)"},{"comment":"The parameter choice in the invocation of Proposition 6.1 is inconsistent. With λ=2^{-N/2}, the N-th summand equals λ^2 f(λ^{-2}|y'|^2)H(y) up to the cutoff, whereas Proposition 6.1 defines h_ij=μλ^2 f(λ^{-2}|x'|^2)H_ij(x). Matching the two forms forces μ=1, not μ=2^{-N} as stated. With the stated μ=2^{-N}, the single-bump metric in Proposition 6.1 is smaller by a factor 2^N than the actual N-th bump in the summed metric, so the leading-order and error terms in Corollary 4.6 are evaluated for the wrong perturbation size. This is readily corrected by taking μ=1, but the correction does not address the interaction gap.","section":"Section 6 (Theorem 6.2)"}],"minor_comments":[{"comment":"The title contains a typo: 'con stant' should be 'constant'; Proposition 2.1 states 'c1,c2>0' but the equation and proof use c, not c2.","section":"Title page"},{"comment":"The citation in the proof is incomplete: '[?, Theorems 8.25 and 8.26]e have' should refer to [11] (Gilbarg-Trudinger) and should be written correctly.","section":"Section 4 (proof of Proposition 4.1)"},{"comment":"The notation ω_n is used in the computation of Sc(∞) without a definition; the authors should state explicitly that ω_n denotes the volume of the unit n-sphere.","section":"Section 2"},{"comment":"The Euler beta function B(·,·) appears in equations (5.4)-(5.6) without definition or reference; it should be identified for the reader.","section":"Section 5"},{"comment":"The infinite sum defining h is asserted to be smooth; the proof should justify uniform convergence of the sum and of its derivatives after choosing N0 large, given the bounds on χ and f.","section":"Section 6 (Theorem 6.2)"},{"comment":"The passage from the half-space metric of Theorem 6.2 to the closed hemisphere S^n_+ in Theorem 1.1 is not spelled out; this is standard since the metric is Euclidean outside a compact set and the boundary is totally geodesic, but it should be stated explicitly.","section":"Theorem 1.1"}],"recommendation":"major_revision","confidential_remarks":"The core analytic work in Sections 3-5 appears sound and the gap in Section 6 is likely repairable with additional interaction estimates, so I do not recommend rejection. However, the paper relies on the authors' own preprint [5] for the mountain-pass critical point used in Proposition 2.1; if that preprint is not yet accepted, its status should be clarified. The mismatch in the parameter choice of μ in Theorem 6.2 should also be corrected, not merely glossed over."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real content here is the Lyapunov-Schmidt machinery in Sections 4-5. The paper adapts Brendle's closed-manifold counterexample and Almaraz's boundary work to the full equation (1.1) with c1,c>0, and the algebraic estimates that force the dimension threshold n≥62 look correct. The warped-product non-uniqueness example in Section 2 is a nice bonus. I checked the discriminant calculation in Proposition 5.9 and the q(62)>0 boundary: that part holds up. If the main theorem were fully justified, this would be the first non-compactness result for lower-energy solutions with both constants positive, which is exactly the direction Han-Li asked about.\n\nThe soft spot is in Theorem 6.2. Proposition 6.1 assumes the metric perturbation h is a single bump supported in B_ρ and zero outside. Theorem 6.2 defines h as an infinite sum of such bumps and invokes Proposition 6.1 with parameters for each bump. The stress-test note's specific overlap objection is actually mistargeted: the cut-off χ(4N^2|x-x_N|) has support radius about 1/(2N^2), which is smaller than the inter-center spacing ~1/N^2, so the supports do not overlap; there is even a positive gap. But the broader logical gap remains. Even with disjoint supports, the summed h is not equal to a single bump globally. Proposition 6.1 requires h=0 outside B_ρ, and Theorem 6.2 does not estimate the effect of the other bumps on the reduced functional F_g. Because the bubble width λ=2^{-N/2} is exponentially small compared with the inter-center distance ~1/N^2, the missing interaction estimate should be obtainable, and I suspect the theorem is true. But as written, Theorem 6.2 does not follow from Proposition 6.1.\n\nI also note the paper leans on the authors' earlier existence results [5,6] for the non-uniqueness example. That is legitimate, though a referee should ask whether those results cover the warped product case without extra hypotheses. The citation pattern is otherwise standard for this literature.\n\nWho is this for? Geometric analysts working on Yamabe-type problems with boundary. It is a technical paper, not a survey, and the construction is the point. It deserves a serious referee: the gap is fillable and the claimed result is significant. I would send it out, with the referee asked to focus on whether the summed metric can be handled by a perturbation/tail estimate or whether Proposition 6.1 needs to be reworked. If that gap is closed, this is publishable in a strong journal.","headline":"A solid Brendle-style blow-up construction for the boundary Yamabe equation with both constants positive, dimensions n≥62, but Theorem 6.2 has a genuine gap: Proposition 6.1 is applied to a summed metric it does not cover.","tokens_in":30934,"tokens_out":4842,"would_cite":true,"duration_ms":53489,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C21","35J20","35B33","34B18"],"pacs":[],"model":"deepseek-v4-flash","headline":"In dimensions at least 62, the boundary Yamabe-type equation has noncompact sets of lower-energy solutions.","keywords":["manifold with boundary","constant scalar curvature","boundary mean curvature","blow-up","noncompactness","Lyapunov-Schmidt reduction","Yamabe problem","umbilic boundary"],"falsifier":"Compute the reduced functional $F_g(\\lambda\\xi,\\lambda\\epsilon)$ directly for the summed-bump metric of Theorem 6.2 and check whether cross terms between adjacent bumps violate the bound used in Proposition 6.1. If the cross-term error exceeds the strict-minimum margin $F(0,1)-\\inf_{\\partial\\Omega'}F$, the perturbation argument produces no critical point and Theorem 1.1 collapses.","tokens_in":29851,"feed_emoji":"💥","tokens_out":10189,"duration_ms":94343,"temperature":0.7,"pith_summary":"The paper constructs a smooth counterexample to compactness for the constant scalar curvature / constant boundary mean curvature equation (1.1) on the upper hemisphere $S^n_+$ when $n\\ge 62$. It produces a metric with positive Yamabe constant, non-conformally flat interior, and umbilic boundary, together with a sequence of positive smooth solutions whose energy is below the single-bubble threshold $S_c$ but whose boundary maximum blows up. This shows that the lower-energy solution set is noncompact, settling the boundary analogue in high dimensions without identifying the sharp dimension. The paper also gives a warped-product example, valid for $n\\ge5$, where the equation admits at least two distinct positive solutions.","feed_headline":"Boundary Yamabe blow-up occurs in dimensions n ≥ 62","feed_subtitle":"Even below the single-bubble energy threshold, positive solutions concentrate on the boundary and escape compactness.","key_machinery":"The central mechanism is a Lyapunov-Schmidt reduction to a family of half-space bubbles $u_{(\\xi,\\epsilon)}$ solving the flat model equation, with the linearized operator inverted on the orthogonal complement $E_{(\\xi,\\epsilon)}$. The reduced energy is a function $F_g(\\xi,\\epsilon)$ whose critical points yield solutions of (1.1); $F_g$ is approximated by $\\lambda^4\\mu^2F(\\xi,\\epsilon)$, where the auxiliary function $F$ is built from a trace-free, divergence-free tensor $H_{ij}(x)=f(|x'|^2)(W_{ikjl}x_kx_l)$ encoding the Weyl curvature of the perturbation. Choosing $f(s)=a_0-s$ with $a_0$ fixed by the discriminant condition forces the integral identities $I(1)>0$, $I'(1)=0$, $I''(1)<0$, and $J(1)<0$, so $F$ has a strict local minimum at $(0,1)$. Scaling the bubble parameters by $\\lambda$ and the metric bump by $\\mu\\lambda^2$ converts this local minimum into a genuine solution of the nonlinear equation.","core_discovery":"For every $n\\ge62$, there is a smooth Riemannian metric $g$ on $S^n_+$ with $Y(S^n_+,\\partial S^n_+,[g])>0$, $g$ not conformally flat, and $\\partial S^n_+$ umbilic, such that equation (1.1) has positive smooth solutions $v_\\nu$ with $I(S^n_+,g)[v_\\nu]