{"id":"0fa893e9-9732-4939-86bb-389a8b768e12","arxiv_id":"2506.08697","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"On infinite weighted graphs, if the graph Laplacian of the distance grows slowly and suitable weighted volume growth holds, every very weak solution of the semilinear wave inequality is identically zero.","lead":"This paper studies a semilinear wave equation with a potential on weighted graphs and gives conditions under which no nonzero global-in-time solutions exist. It covers both nonnegative and sign-changing solutions, using a new class of test functions for the sign-changing case.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The test function φ_R in Theorem 3.4 decays like e^{-δ d/R}, not e^{-δd}; it therefore fails the hypothesis of Remark 5.3 and Proposition 5.2, leaving the whole-graph sums and integration by parts unjustified for u ∈ L1_loc(Xδ).","rationale":"The reader's weakest-assumption analysis isolates exactly the point on which the sign-changing theorem turns: admissibility of the test function φ_R in (5.3) for the very weak formulation (2.4). I re-derived the relevant estimates from the paper. Claim (iii) in the proof of Theorem 3.4 gives |Δφ_R| ≤ C R^{-(1+α)} η^s e^{-δ d/R} 1_{V \\ B_R}, and the bound for (φ_R)_tt contains the same spatial factor e^{-δ d/R}. The hypothesis of Remark 5.3 and of Proposition 5.2 instead requires e^{-δ d}. These two exponential rates are not interchangeable: for every fixed R > 1, e^{δ d(1-1/R)} → ∞, so the constructed test function is strictly slower-decaying than the admissibility condition demands. This is an internal gap in the proof, not a disagreement with external consensus: the paper's own Remark 5.3, as written, does not cover φ_R. The nonnegative Theorem 3.1 is not affected, since it uses compactly supported cutoffs; the objection targets the advertised sign-changing Theorem 3.4 and its corollaries. I note the gap is plausibly repairable, for example by choosing ψ so that ψ((d-j)/R) ≈ e^{-δd}; the volume conditions (3.9)-(3.10) would only become easier for faster spatial decay. But the submitted manuscript does not provide such a replacement, and the proof as written does not justify the passage from finite-support test functions to the whole-graph sums. Therefore the reader's REJECT verdict remains appropriate for the current version.","tokens_in":22717,"tokens_out":13518,"duration_ms":169707,"concrete_test":"Check the admissibility of the explicit test function (5.3) against the hypothesis of Remark 5.3. Fix R > 1, choose t in the support of η^s(t/R^{(1+α)/2}), and along a sequence x_n with d(x_n,x0) → ∞ compute φ_R(x_n,t) e^{δ d(x_n,x0)}. By (5.3), φ_R(x_n,t) behaves like C e^{-δ d(x_n,x0)/R}, so the product behaves like C e^{δ d(x_n,x0)(1-1/R)}, which diverges. Hence no constant C (even R-dependent) can satisfy |φ_R| ≤ C e^{-δd}. A complementary computational check: on V = Z with μ ≡ 1, δ = 1, R = 2, take u ≡ 1 on the time interval where (φ_R)_tt ≠ 0; then the absolute value of the second time-derivative term in (5.4) equals C Σ_{x∈Z} e^{-|x|/2} = ∞, confirming that the whole-graph sums used in the proof are not justified by the stated assumptions.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The sign-changing result Theorem 3.4 depends on applying Remark 5.3 to the test function φ_R defined in (5.3). Remark 5.3 requires pointwise bounds such as |φ(x,t)| ≤ C e^{-δ d(x,x0)} with the same δ as in the space Xδ, and Proposition 5.2 needs the same decay to justify the identity Σ u Δφ = Σ Δu φ. But for (5.3), ψ(r) = e^{-δr} for r ≥ 2 gives ψ((d(x,x0)-j)/R) ≈ e^{-δ(d-j)/R} = e^{δj/R} e^{-δ d/R}. Since R ≥ max{R0, 2j} ≥ 1, for large d the ratio e^{-δ d/R} / e^{-δd} = e^{δd(1-1/R)} tends to infinity. Consequently the bounds (i)–(iii) proved for φ_R in Theorem 3.4 all carry the factor e^{-δ d(x,x0)/R}, not e^{-δd(x,x0)}. The hypothesis of Remark 5.3 is therefore not satisfied, and the approximation argument used there cannot pass from finite-support truncations to φ_R: the terms ∫ Σ u(φ_R)_tt and ∫ Σ u Δφ_R need not be absolutely convergent under only u ∈ L1_loc([0,∞), Xδ). For instance, on V = Z with μ ≡ 1, δ = 1, u ≡ 1 on the time support of φ_R belongs to L1_loc(Xδ), but Σ_x e^{-δ|x|/R} = ∞. Thus (5.4) is not obtained by a legitimate application of (2.4), and all subsequent Young and Hölder estimates in the proof of Theorem 3.4 rest on an unjustified step. This is load-bearing because it is precisely the step that moves from compactly supported test functions to the whole-graph sums that the sign-changing theorem requires.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the semilinear wave inequality u_tt - Δu ≥ v|u|^σ on infinite weighted graphs and claims nonexistence of nonzero global very weak solutions. For nonnegative solutions (Theorem 3.1) the proof adapts the parabolic arguments of the authors' preprint [34], using compactly supported test functions and weighted volume growth conditions. For sign-changing solutions (Theorem 3.4) the paper introduces a novel technique: test functions supported on the whole graph with exponential decay, together with an ℓ^1-type assumption u ∈ L^1_loc([0,∞), X_δ). Corollaries cover the cases v ≥ g and v ≡ 1, and a separate result treats finite graphs.","tokens_in":23140,"tokens_out":9145,"duration_ms":113595,"significance":"If the sign-changing theorem were correct, it would be a substantial extension of known Euclidean and manifold nonexistence results to discrete structures and would introduce a new test-function technique for hyperbolic inequalities on graphs. The nonnegative result is a plausible adaptation of existing parabolic techniques. However, the central proof of the sign-changing theorem contains a load-bearing gap: the globally supported test function does not satisfy the decay hypotheses under which the paper's own approximation and integration-by-parts tools are proved. Since the advertised novelty is precisely the sign-changing result, the paper cannot be accepted in its current form.","major_comments":[{"comment":"The test function φ_R defined in (5.3) satisfies |φ_R(x,t)| ≤ C e^{-δ d(x,x0)/R} (and similarly for its derivatives), not the bound |φ_R(x,t)| ≤ C e^{-δ d(x,x0)} required by Remark 5.3 and Proposition 5.2. For R > 1 the decay e^{-δ d/R} is slower than e^{-δ d}, so φ_R does not meet the hypotheses of those results. Consequently the extension of the very weak solution inequality (2.4) to φ_R, used as Eq. (5.4), and the identity Σ_x Δu φ_R = Σ_x u Δφ_R obtained from Proposition 5.2 are not justified by the stated assumptions. The summability u ∈ L^1_loc([0,∞), X_δ) does not ensure convergence of Σ_x u(x,t)(φ_R)_tt(x,t) or Σ_x u(x,t)Δφ_R(x,t): for example, on V = Z with μ ≡ 1 and δ = 1, the constant function u ≡ 1 belongs to X_δ, yet Σ_x e^{-δ|x|/R} = ∞ for every R > 1. All subsequent Young and Hölder estimates in the proof of Theorem 3.4 depend on this unjustified step.","section":"Section 5, Eq. (5.3) and Remark 5.3 / Proposition 5.2"},{"comment":"The approximation argument in Remark 5.3 is also incomplete with respect to the nonlinear term. For the truncated test functions φ_k the very weak solution inequality has right-hand side ∫_0^T Σ_{x∈B_k} v|u|^σ φ_k. Passing to the limit k → ∞ requires convergence or at least a meaningful limit of Σ_x v|u|^σ φ. The assumption u ∈ L^1_loc([0,∞), X_δ) only controls Σ_x |u| e^{-δ d}; it does not imply any global summability of v|u|^σ with a weight. The definition of very weak solution only gives v|u|^σ ∈ L^1_loc at each vertex. Thus the right-hand side of (2.4) for the whole-graph φ_R may be infinite, and the inequality in the form used to derive (5.4) and the subsequent estimates is not established.","section":"Remark 5.3"},{"comment":"The proof of the nonnegative result is not self-contained. The key estimate (4.2) is obtained 'following the same procedure of [34, Theorem 2.6]', and the final conclusion is said to follow 'by an application of Hölder inequality, in a similar way as [34, Theorem 2.6]' without presenting those details. Since [34] is an unpublished preprint of the same authors and is not included in the manuscript, the referee cannot verify the decisive final step of Theorem 3.1 from the material provided. The authors should either reproduce the relevant argument or cite a published reference that contains it.","section":"Section 4, proof of Theorem 3.1"}],"minor_comments":[{"comment":"The abstract contains missing spaces ('Weinvestigatethe semilinear wave equation') which should be corrected.","section":"Abstract"},{"comment":"The sentence 'is enough to observe that it' should read 'is enough to observe that'.","section":"Corollary 3.5, proof"},{"comment":"The phrase 'we would need to construct a a suitable compactly supported nonnegative cut-off function' contains a duplicated article 'a'.","section":"Section 5, Proposition 5.1 discussion"},{"comment":"The condition Σ_x u_1(x)μ(x) ≥ 0 is used together with u_1 ∈ X_δ to justify that the sums of the positive and negative parts of u_1 over V are finite; this implication should be stated explicitly, since the condition alone involves an infinite sum that need not converge in general.","section":"Theorem 3.4, condition (3.8)"}],"recommendation":"reject","confidential_remarks":"The sign-changing theorem is the advertised novel contribution and it is not valid as written: the globally supported test function fails the decay hypotheses of the paper's own approximation and integration-by-parts lemmas. This is not a local typo but a gap in the main derivation. The nonnegative part also leans heavily on the authors' unpublished preprint [34] for its essential final step. Unless the authors can supply a different admissible test-function construction or substantially stronger summability assumptions, the central claim of the paper remains unproved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper has a genuine new idea and a load-bearing gap. The nonnegative result (Theorem 3.1) is a reasonable graph analogue of the authors' earlier parabolic work [34], adapted to the wave equation; it is not fully self-contained, but the adaptation looks plausible. The sign-changing result (Theorem 3.4) is the actual contribution: the use of non-compactly supported, exponentially decaying test functions is new in this context, and Proposition 5.1, showing why compact support cannot work for sign-changing solutions, is a nice observation. The examples on Z^N, trees, and product graphs are also useful and clearly presented.\n\nBut Theorem 3.4 has a serious gap. The test function is φ_R(x,t) = η^s(t/R^{(1+α)/2}) ψ((d(x,x0)-j)/R), with ψ(r)=e^{-δr} for r≥2. So its spatial decay is e^{-δ d(x,x0)/R}, not e^{-δ d(x,x0)}. Remark 5.3 and Proposition 5.2, which are used to pass from the very weak solution definition to the whole-graph sums in (5.4), require the same δ that appears in Xδ. The paper never justifies that the sums and integration by parts converge under only u ∈ L1_loc(Xδ) when the test function decays at the slower rate δ/R. The stress-test note's concrete counterexample (u≡1 on Z) does not work because u≡1 is not in Xδ, but the structural concern is real: one can take u(x)=e^{δd}/d^2 on Z, which is in Xδ, while Σ |u|e^{-δ d/R} diverges for R>1. This means the Young and Hölder estimates in the proof of Theorem 3.4 are applied to expressions that may not be finite. The gap is load-bearing, since this is exactly the step that moves from compactly supported test functions to whole-graph sums.\n\nA repair would require either a stronger decay assumption on u, a different test function, or a separate argument proving that assumptions (3.9)–(3.10) plus u∈L1_loc(Xδ) imply the needed convergence. The nonnegative theorem also outsources key estimates and the final Hölder step to the authors' own preprint [34], which is a soft spot, but that is a presentation issue, not a mathematical flaw.\n\nThis paper deserves a serious referee: it targets the right subfield, the ideas are worth engaging with, and the gap is repairable. I would not cite it in its current form, but I would recommend sending it to review and asking for major revision.","headline":"The sign-changing theorem is a real novelty, but it currently rests on a test-function admissibility gap: φ_R decays like e^{-δ d/R}, not e^{-δ d}, so the invoked integration-by-parts and approximation results do not apply as written.","tokens_in":23670,"tokens_out":5415,"would_cite":false,"duration_ms":62562,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35A01","35A02","35B44","35K05","35K58","35R02"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that, on infinite weighted graphs, the semilinear wave inequality $u_{tt}-\\Delta u \\ge v|u|^\\sigma$ has no nonzero global very weak solutions under weighted volume-growth conditions.","keywords":["semilinear wave equation","weighted graphs","graph Laplacian","nonexistence of global solutions","sign-changing solutions","very weak solutions","weighted volume growth","test function method"],"falsifier":"To test the proof, compute the truncated sums in Remark 5.3 explicitly for $\\phi_R$ from (5.3) on the lattice $\\mathbb{Z}$ with $v\\equiv 1$ and $\\sigma=2$: if the sums and time integrals are finite whenever $u\\in L^1_{\\rm loc}([0,\\infty),X_\\delta)$ and (3.9)-(3.10) hold, the admissibility step is justified; if they diverge under those hypotheses, the proof has a gap. To refute the theorem itself, one would need a nonzero very weak solution of the inequality on a graph satisfying (2.6), with $\\sum_x u_1\\mu\\ge 0$, $u_0,u_1\\in X_\\delta$, $u\\in L^1_{\\rm loc}([0,\\infty),X_\\delta)$, and (3.9)-(3.10).","tokens_in":22517,"feed_emoji":"🌊","tokens_out":12604,"duration_ms":127626,"temperature":0.7,"pith_summary":"This paper proves that, on infinite weighted graphs, the semilinear wave inequality $u_{tt}-\\Delta u \\ge v|u|^\\sigma$ has no nonzero global-in-time very weak solutions, provided the weighted volume of suitable space-time regions grows slowly enough and the initial data satisfy a sign or decay condition. The statement covers both nonnegative solutions and, under stronger hypotheses, solutions that may change sign; for sign-changing solutions the authors develop a new technique using test functions supported on the whole graph with exponential decay, because they show compactly supported cut-offs cannot work. These are sufficient conditions, not necessary ones: they identify growth rates (involving the distance-function parameter $\\alpha$ and the exponent $\\sigma$) beyond which global existence is impossible on the graph. If correct, the results provide discrete counterparts of classical nonexistence theorems for wave equations on Euclidean space and Riemannian manifolds, and they give explicit examples on lattices, trees, and product graphs.","feed_headline":"Slow volume growth forces graph waves to vanish","feed_subtitle":"Both nonnegative and sign-changing global solutions are ruled out on graphs with weighted volume-growth bounds.","key_machinery":"The machinery is a family of cut-off test functions engineered so that the nonlinear term absorbs the linear terms. For nonnegative solutions the test function is compactly supported, $\\phi_R(x,t)=\\varphi((d(x,x_0)^{\\theta_1}+t^{\\theta_2})/R^{\\theta_1})$, and the control comes from the estimate $-\\Delta\\phi_R\\le C R^{-(1+\\alpha)}\\mathbf{1}_{F_R}$ together with Young's inequality and the volume bound (3.1). For sign-changing solutions, Proposition 5.1 shows that a nonzero cut-off satisfying $|\\Delta\\psi^\\gamma|\\le C\\psi^\\beta F$ with $\\psi(x_1)=0$ must vanish everywhere, so compact support is impossible; the authors instead use $\\phi_R(x,t)=\\eta^s(t/R^{(1+\\alpha)/2})\\psi((d(x,x_0)-j)/R)$ with $\\psi(r)=e^{-\\delta r}$ for large $r$, and the bound $|\\Delta\\phi_R|\\le C R^{-(1+\\alpha)}\\eta^s e^{-\\delta d/R}\\mathbf{1}_{V\\setminus B_R(x_0)}$. The supporting identity is Proposition 5.2, an integration-by-parts formula for the graph Laplacian valid when one factor lies in the weighted $\\ell^1$ space $X_\\delta$ and the other decays like $e^{-\\delta d}$, which lets the Laplacian act on the test function instead of on $u$.","core_discovery":"The central claim is that global nonexistence for the wave inequality on a weighted graph is governed by weighted volume growth, with a qualitative difference between the nonnegative and sign-changing cases. Theorem 3.1 says that if $\\Delta d(\\cdot,x_0)\\le C/d(\\cdot,x_0)^\\alpha$ and the space-time volume condition (3.1) holds, then any nonnegative very weak solution with the initial-velocity condition (3.3) is identically zero. Theorem 3.4 removes the sign assumption but requires the two-sided Laplacian bound $|\\Delta d|\\le C/d^\\alpha$, the summability of $u$, $u_0$, $u_1$ in the weighted $\\ell^1$ space $X_\\delta$, and the stronger weighted volume conditions (3.9)-(3.10); then $u\\equiv 0$. The proof for sign-changing solutions is the paper's main technical contribution: after Proposition 5.1 rules out compactly supported cut-offs, the authors use the whole-graph test function (5.3) and justify the resulting infinite sums through Proposition 5.2 and Remark 5.3.","pith_inferences":["The whole-graph test-function technique is not tied to the scalar inequality; it should extend to systems of wave inequalities or higher-order hyperbolic inequalities on graphs, with the exponential rate in $\\psi$ adjusted to the order of the operator.","The decay $e^{-\\delta d(x,x_0)/R}$ in the volume conditions suggests that the effective weight varies with the scale $R$; the sharp boundary for nonexistence may be expressible as a large-deviation rate for the graph's volume measure rather than as a polynomial exponent.","A testable extension would be to weaken $u\\in L^1_{\\rm loc}([0,\\infty),X_\\delta)$ to a moment condition on $u$ that still makes the truncated sums in Proposition 5.2 converge; that would separate the essential summability assumption from the technical admissibility of the test function."],"forward_implications":["On the integer lattice $\\mathbb{Z}^N$, the theorems reproduce the classical critical range: for $v\\equiv 1$ they cover $1<\\sigma\\le (N+1)/(N-1)$ when $N\\ge 2$, and all $\\sigma>1$ when $N=1$ (Example 6.1).","On homogeneous trees, the results allow exponentially growing potentials, such as $g(x)=C\\,d(x,x_0)^{(\\sigma-3)/2}N^{(\\sigma-1)d(x,x_0)}$, and still conclude that the only global very weak solution is zero (Example 6.2).","For finite weighted graphs, Corollary 7.2 gives a clean statement: if the potential is time-independent and $\\sum_x u_1(x)\\mu(x)\\ge 0$, then no nonzero global very weak solution exists, with no further volume-growth condition.","The sign-changing result adds a new obstruction: even solutions that are not nonnegative are forced to vanish, provided they are summable in $X_\\delta$ and the weighted volume conditions (3.9)-(3.10) hold."],"supporting_citations":[{"why":"supplies the classical blow-up result for the wave equation on $\\mathbb{R}^n$ that sets the critical-exponent benchmark for the graph results.","marker":"[24]"},{"why":"provides the broader nonexistence framework for weak solutions of hyperbolic inequalities that the authors adapt.","marker":"[32]"},{"why":"is the parabolic-on-graphs paper whose test-function estimates and volume conditions are the template for Theorem 3.1.","marker":"[34]"},{"why":"gives the Riemannian-manifold hyperbolic result with potential-dependent weighted volume growth, which the graph theorem mirrors.","marker":"[35]"},{"why":"introduces the graph setting and distance-function assumptions used throughout, including the examples.","marker":"[33]"},{"why":"supplies the semilinear heat-equation nonexistence result on graphs that motivates the nonnegative case.","marker":"[41]"}],"fun_headline_variants":["Graph wave solutions fail to exist under volume growth bounds","No global-in-time solutions for graph wave equations","Graph wave nonexistence from weighted volume growth","Slow volume growth rules out graph wave solutions","Weighted graphs: no global waves, sign-changing too"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the whole-graph test function $\\phi_R$, with spatial decay only $e^{-\\delta d(x,x_0)/R}$, is admissible in the very-weak-solution inequality and in the integration-by-parts formula even though those tools are stated for decay $e^{-\\delta d(x,x_0)}$; without a proof that the truncated sums converge, the passage $R\\to\\infty$ is not justified.","fun_headline_variants_meta":{"raw":{"variants":["Graph wave solutions fail to exist under volume growth bounds","No global-in-time solutions for graph wave equations","Graph wave nonexistence from weighted volume growth","Slow volume growth rules out graph wave solutions","Weighted graphs: no global waves, sign-changing too"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001186,"raw_usage":{"total_tokens":4825,"prompt_tokens":805,"completion_tokens":4020,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":421,"completion_tokens_details":{"reasoning_tokens":3958}},"tokens_in":421,"tokens_out":4020,"duration_ms":35951,"temperature":1.0,"reasoning_tokens":3958,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:07:43.514368+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To test the proof, compute the truncated sums in Remark 5.3 explicitly for $\\phi_R$ from (5.3) on the lattice $\\mathbb{Z}$ with $v\\equiv 1$ and $\\sigma=2$: if the sums and time integrals are finite whenever $u\\in L^1_{\\rm loc}([0,\\infty),X_\\delta)$ and (3.9)-(3.10) hold, the admissibility step is justified; if they diverge under those hypotheses, the proof has a gap. To refute the theorem itself, one would need a nonzero very weak solution of the inequality on a graph satisfying (2.6), with $\\sum_x u_1\\mu\\ge 0$, $u_0,u_1\\in X_\\delta$, $u\\in L^1_{\\rm loc}([0,\\infty),X_\\delta)$, and (3.9)-(3.10).","supporting_citations":[{"cited_title":"Kato, Blow-up of solutions of some nonlinear hyperbolic equations, Comm","cited_arxiv_id":null,"evidence_quote":"supplies the classical blow-up result for the wave equation on $\\mathbb{R}^n$ that sets the critical-exponent benchmark for the graph results."},{"cited_title":"Mitidieri and S","cited_arxiv_id":null,"evidence_quote":"provides the broader nonexistence framework for weak solutions of hyperbolic inequalities that the authors adapt."},{"cited_title":"Nonexistence of solutions to parabolic problems with a potential on weighted graphs","cited_arxiv_id":"2404.12058","evidence_quote":"is the parabolic-on-graphs paper whose test-function estimates and volume conditions are the template for Theorem 3.1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"gives the Riemannian-manifold hyperbolic result with potential-dependent weighted volume growth, which the graph theorem mirrors."},{"cited_title":"Wu,On nonexistence of global solutions for a semilinear heat equation on graphs, Nonlinear Anal","cited_arxiv_id":null,"evidence_quote":"supplies the semilinear heat-equation nonexistence result on graphs that motivates the nonnegative case."}],"review_version":1}