{"id":"b2e9b1d6-02fc-495d-98fc-c26201c2a20d","arxiv_id":"2505.10712","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On regular metric trees, KPP solutions propagate if f'(0) > E0 and become extinct if f'(0) < E0 (homogeneous case), with speed estimates between 2*sqrt(f'(0)-E0) and M*rho1*b1/(b1-1).","lead":"This paper studies a reaction-diffusion equation on a branching tree-shaped network and finds that the fate of the solution, going extinct or spreading, is decided by whether the reaction growth rate exceeds the smallest eigenvalue of the network Laplacian. It also gives upper and lower bounds on the spreading speed that depend on branch length and branching number.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proofs depend on comparison theorems deferred to an unpublished preprint; Lemma 5.1's matching condition also fails as written, breaking Theorem 2.2.","rationale":"The reader's weakest assumption (unproven comparison principles) is the most load-bearing because it underpins all main theorems. I agree with that assessment. A further independent check of Lemma 5.1 shows an explicit mismatch in the construction used for Theorem 2.2; while likely a typographical sign error, it means the proof in the current arXiv version is not self-consistent. Since the fix is straightforward and does not change the mathematical architecture, the appropriate verdict stays conditional rather than rejection; the paper should either prove or cite publicly available versions of Theorems 4.2-4.4 and correct (5.5). No ad hominem; this is on the argument.","tokens_in":34934,"tokens_out":18252,"duration_ms":161107,"concrete_test":"Take a homogeneous tree with b=2, r=1, λ∈(0,E0) and α∈[α_-,1); evaluate (5.5) at ρ_1 for n=1 and n=2 and check g_2(ρ_1) - g_1(ρ_1). The difference is nonzero for b=2, confirming the matching failure; then replace b^{n/2} by b^{-n/2} and verify (5.4d) and the bound in (5.7). If the corrected formula satisfies all conditions, the issue is a typo, and the unresolved dependence on [29] remains the decisive risk.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorems 4.2, 4.3 and 4.4, which provide the comparison and monotonicity principles used in Theorems 2.1, 2.2, 2.3 and 2.5, are stated with only 'We refer the reader to [29]' and no proof. If any of these principles requires hypotheses beyond (H1)-(H2), the central dichotomy and both speed bounds lose their support. Independently of [29], the proof of Theorem 2.2 contains a concrete matching error: for the function g in Lemma 5.1, formula (5.5) gives g_n(ρ_n) = α^n b^{n/2} sin(√λ r) and g_{n+1}(ρ_n) = α^n b^{n/2+1} sin(√λ r), which violates the required continuity g_{n+1}(ρ_n) = g_n(ρ_n) for b ≥ 2. Hence z(ρ,t) := g(ρ)e^{-[λ-f'(0)]t} is not a supersolution of (1.1) as written, so the extinction conclusion of Theorem 2.2 is not established by the displayed construction. A corrected exponent b^{-n/2} restores (5.4d), so this is likely a typographical error, but it needs a public correction.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the semilinear heat equation u_t = Δu + f(u) on regular metric trees with homogeneous Neumann condition at the root, under geometric assumptions (H0) and hypotheses (H1)-(H5) on the initial datum and on the KPP-type nonlinearity f. The main claims are: if f'(0) > E0 := min σ(-Δ), then every nonzero solution converges to 1 uniformly on compact subsets of the tree (Theorem 2.1); on homogeneous trees, if f'(0) < E0 and the initial datum is suitably small, the solution converges to 0 uniformly (Theorem 2.2), with an analogous extinction result under the sub-Fujita-type assumption (H5) (Theorem 2.3); and the asymptotic speed of propagation is bounded above by ĉ = Mρ1 b1/(b1-1) (Theorem 2.4) and below by č = 2√(f'(0)-E0) (Theorem 2.5). The proofs rely on spectral analysis of the Neumann Laplacian on regular trees (including the explicit formula for homogeneous trees in Theorem 3.3), on symmetric super- and subsolutions, and on several comparison principles.","tokens_in":35206,"tokens_out":9159,"duration_ms":80492,"significance":"If the results are correct, the paper establishes a nontrivial threshold dichotomy for KPP-type equations on metric trees, parallel to the hyperbolic-space theory, and it provides explicit, parameter-free bounds on the propagation speed that depend only on the spectral threshold and on the geometry of the tree. The spectral computation in Theorem 3.3 is self-contained and convincing, and Theorem 4.5 is proved in detail. The constructions of super- and subsolutions are explicit and give falsifiable predictions. However, the central results are not fully established within the manuscript because the comparison principles on which they rely are deferred to an unpublished preprint, and because the matching condition in Lemma 5.1 fails as written.","major_comments":[{"comment":"The comparison principles stated as Theorems 4.2, 4.3 and 4.4 are not proved in this paper; the text refers only to the unpublished preprint [29]. These results are load-bearing: Theorem 4.2 is used in the proofs of Theorems 2.2 and 2.5, Theorem 4.3 in the proof of Theorem 2.1, and Theorem 4.4 (via Remark 4.4) in the proofs of Theorems 2.3 and 2.4. Since [29] is not publicly available, the central dichotomy and the speed bounds are not verifiable from the manuscript alone. The authors should either include the proofs of these comparison theorems or clearly state that the main results are conditional on [29] and make that preprint accessible.","section":"Section 4.2, Theorems 4.2-4.4"},{"comment":"The application of Theorem 4.3 after (5.2) uses the constant function 1 as the upper stationary supersolution. However, q ≡ 1 is not a stationary supersolution in the sense of Definition 4.4, because it does not belong to H1(T); it is only a weak stationary solution (Remark 4.2). The statement of Theorem 4.3 does not cover weak stationary supersolutions, so the proof as written needs either an extended version of Theorem 4.3 or a separate limiting argument to justify the upper bound in (5.2).","section":"Section 5, proof of Theorem 2.1"},{"comment":"The claimed matching condition (5.4d) fails as written. Substituting (5.5) gives g_n(ρ_n) = α^n b^{n/2} sin(√λ r) and g_{n+1}(ρ_n) = α^n b^{n/2+1} sin(√λ r), which are unequal for b ≥ 2. Consequently the function z(ρ,t) := g(ρ)e^{-[λ-f'(0)]t} used in the proof of Theorem 2.2 is not continuous at the vertices and is not a supersolution of problem (1.1) as defined. Replacing b^{n/2} by b^{-n/2} in (5.5) restores (5.4d); this correction and the resulting changes in the estimates (5.6)-(5.7) should be made explicit before the extinction conclusion of Theorem 2.2 is accepted.","section":"Lemma 5.1, equation (5.5)"}],"minor_comments":[{"comment":"The phrase \"T is is a regular metric tree\" contains a duplicated \"is\" and should read \"T is a regular metric tree\".","section":"Assumption (H0), Section 1.2"},{"comment":"The sentence \"Moreover, there holds ψϵ ≤ 0 in R+\" should state ψϵ′ ≤ 0 in R+, since the intended assertion is the nonpositivity of the derivative used later in the proof of Theorem 2.5.","section":"Lemma 6.4, after (6.17)"},{"comment":"The notation \"θ(0) = π/2\" should be \"θ(b) → π/2 as b → ∞\"; also the numerical value θ(2) ≈ π/12 is inaccurate, since arccos(2√2/3) ≈ 0.34 rad ≈ π/9.","section":"Proposition 2.6, proof"},{"comment":"The sentence \"A stationary subsolution q is defined by reversing inequalities in (4.13b)-(4.13c)\" refers to equations that are introduced later and concern symmetric stationary supersolutions; it should refer to (4.8b)-(4.8c) of the present definition.","section":"Definition 4.4"}],"recommendation":"major_revision","confidential_remarks":"The core issue is the dependence on the unpublished preprint [29] for the comparison principles, together with the concrete matching error in Lemma 5.1. If the authors can supply or make available the proofs in [29] and correct the exponent in (5.5), the results appear plausible and the paper would be a solid contribution. The editor may wish to verify that [29] is indeed available and that the hypotheses of the comparison theorems match (H1)-(H2) as used here."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nBottom line: this is a real advance, not a repackaging. The authors prove that on regular metric trees the extinction/propagation threshold is f'(0) vs E0 = min σ(-Δ), the analog of the hyperbolic-space result, with the interesting twist that E0 depends on branching and edge length. The speed bounds are new, and the explicit spectral computation for homogeneous trees (Theorem 3.3) is solid and self-contained. The observation that travelling waves don't exist on regular trees (Section 1.5) explains why the sub/supersolution construction has to be different, and the construction itself is clever. Credit where due: the spectral part and the symmetric-supersolution machinery are genuinely useful.\n\nThe soft spots are real but not fatal. The biggest is that Theorems 4.2–4.4—the comparison principles underpinning Theorems 2.1, 2.2, 2.3 and 2.4—are not proved here; they're cited from the authors' unpublished preprint [29]. So the central dichotomy rests on an external, not-yet-verified result. That's a citation gap, not circularity, but it means the paper is not self-contained as it stands. I'd want to see those proofs or at least have [29] posted before accepting the main theorems as fully verified.\n\nThe second issue is more concrete. In Lemma 5.1, the function g defined by (5.5) does not satisfy the continuity condition (5.4d): at ρ_n, g_n(ρ_n)=α^n b^{n/2} sin(√λ r) while g_{n+1}(ρ_n)=α^n b^{n/2+1} sin(√λ r), so equality fails for b≥2. The correct exponent is b^{-n/2} in (5.5); that restores (5.4d) and makes the rest of the lemma work. This looks like a typo, not a deep flaw, but it must be fixed in print because Theorem 2.2 relies on it.\n\nMinor: Remark 2.1 contains a corrupted formula (garbled symbols), and the 'suitably small' in the theorem statements is vague until you read the definitions right after, which actually quantify it via g, h-tilde, and m—so that's fine.\n\nWho's this for? Anyone working on reaction-diffusion on networks, or on nonlinear PDE on noncompact spaces. It earns a serious referee. I'd send it to review with a request that the authors address the [29] dependence and correct the Lemma 5.1 exponent. If the comparison principles check out, this will be a useful citable result.\n\nRecommendation: send to peer review, conditional on the authors supplying the deferred proofs or a public version of [29] and fixing (5.5).","headline":"New threshold result for KPP on metric trees, with a clean spectral constant and explicit speed bounds; two fixable issues—an unpublished comparison-principle citation and a matching typo in Lemma 5.1—need attention before the main theorems are fully supported.","tokens_in":35736,"tokens_out":3679,"would_cite":true,"duration_ms":31642,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35K55","35C07","35R02"],"pacs":[],"model":"deepseek-v4-flash","headline":"A spectral threshold $f'(0)=E_0$ decides whether KPP-type waves invade or die out on regular metric trees.","keywords":["metric trees","semilinear parabolic equations","KPP-type reaction terms","propagation versus extinction","spectral threshold","front speed","comparison principles","Neumann Laplacian"],"falsifier":"Take the homogeneous tree with $b=2$ and edge length $r=1$, source $f(u)=\\lambda u(1-u)$, and let $\\lambda$ exceed $E_0=(\\arccos(2\\sqrt{2}/3))^2$. Simulate the Cauchy-Neumann problem on exhausting finite truncated trees with a small compact initial bump near the root; if the solution does not converge to 1 uniformly on compact subsets, Theorem 2.1 is refuted. Alternatively, a counterexample to the order-preserving comparison principle for sub- and supersolutions on a non-compact tree would invalidate both threshold theorems.","tokens_in":34727,"feed_emoji":"🌳","tokens_out":8115,"duration_ms":78138,"temperature":0.7,"pith_summary":"The paper studies the semilinear heat equation $u_t=\\Delta u+f(u)$ on an infinite regular metric tree, with a Neumann condition at the root and initial data between 0 and 1, for KPP-type sources $f$. Its central claim is that the long-time behaviour is controlled by the sign of $f'(0)-E_0$, where $E_0$ is the bottom of the spectrum of the Neumann Laplacian on the tree. Under the tree's structural assumptions $E_0$ is positive; if $f'(0)>E_0$, every nontrivial solution converges to 1 uniformly on compact subsets, while on homogeneous trees with $f'(0)<E_0$ every sufficiently small solution converges to 0 uniformly. The paper also bounds the asymptotic speed of propagation from below by $2\\sqrt{f'(0)-E_0}$ and from above by $M\\rho_1 b_1/(b_1-1)$. These results matter because they show that tree geometry, through branching and edge length, can suppress or accelerate invasion fronts in a way that parallels the hyperbolic space and is absent in Euclidean space.","feed_headline":"On metric trees, a spectral gap decides invasion versus die-out","feed_subtitle":"Sources above the Laplacian's bottom eigenvalue always win; below it, small populations die out.","key_machinery":"The central objects are the spectral bottom $E_0=\\min\\sigma(-\\Delta)$ of the Neumann Laplacian on the regular tree and symmetric stationary sub- and supersolutions built on a weighted half-line reduction. Symmetric functions on the tree reduce to functions on $\\mathbb R^+$ with the branching weight $\\beta(\\rho)$; the restricted Laplacian becomes an operator $A$ on $L^2(\\mathbb R^+)$ with jump conditions at the radii $\\rho_n$, and for homogeneous trees $E_0=\\theta^2/r^2$ with $\\theta=\\arccos(2\\sqrt b/(b+1))$. The proof of propagation uses eigenfunctions on exhausting domains to build small stationary subsolutions; the proof of extinction uses explicit radial profiles $g$ satisfying $g''+\\lambda g=0$ with $\\lambda<E_0$ as stationary supersolutions. Comparison principles then trap the solution between these barriers. The speed bounds come from traveling-wave-like symmetric profiles $m(\\rho-ct)$ adapted to the Kirchhoff jump conditions at the vertices.","core_discovery":"The paper establishes a sharp dichotomy for the Cauchy-Neumann problem on an infinite regular metric tree. With a KPP source (so $f(0)=f(1)=0$, $f>0$ in $(0,1)$, and $f(u)/u\\le f'(0)$), the bottom of the $L^2$ spectrum $E_0=\\min\\sigma(-\\Delta)$ is positive under the structural assumptions on the tree. If $f'(0)>E_0$, Theorem 2.1 asserts that every solution with $u_0\\not\\equiv 0$ satisfies $u(x,t)\\to 1$ uniformly on compact subsets of $T$; if $T$ is homogeneous and $f'(0)<E_0$, Theorem 2.2 asserts that every suitably small solution satisfies $u(x,t)\\to 0$ uniformly on $T$. The propagation speed is bracketed: no front moves faster than $\\hat c=M\\rho_1 b_1/(b_1-1)$, and when propagation occurs no front moves slower than $\\check c=2\\sqrt{f'(0)-E_0}$. The mechanism is that the positivity of $E_0$ strengthens diffusion on trees relative to Euclidean space, so a weak source can be suppressed by spreading while a strong source overwhelms it.","pith_inferences":["If the driving mechanism is really the positivity of $E_0$ rather than exponential volume growth, then regular trees whose branching and edge lengths give polynomial volume growth, where the paper conjectures $E_0=0$, should behave like Euclidean space: every nonzero KPP solution propagates to 1, with speed near $2\\sqrt{f'(0)}$. This is the paper's conjecture, not one of its theorems.","A testable extension is to compute the first Dirichlet eigenvalue on truncations of a homogeneous tree as the truncation grows; the paper's Proposition A.6 says it decreases to $E_0$, so finite-network simulations can empirically probe the threshold $f'(0)=E_0$.","A natural sharper target left open is the exact front speed $c_0$; one could seek it as the minimal speed of a traveling-wave-like solution of the half-line problem with Kirchhoff jump conditions, analogous to the Euclidean ODE $q''+cq'+f(q)=0$.","Should the deferred comparison principles, cited to an unpublished companion paper, turn out to require additional hypotheses on the tree or the nonlinearity, the dichotomy and speed bounds would inherit those hypotheses; until that companion proof is public, the theorem's full range remains conditional."],"forward_implications":["If $f'(0)>E_0$, any nonzero initial datum, however small in $L^2(T)$, leads to full invasion of every compact subset of the tree.","On a homogeneous tree with $f'(0)<E_0$, all initial data lying below an explicit radial profile are driven to zero uniformly in the whole tree.","The asymptotic front speed is pinned between $2\\sqrt{f'(0)-E_0}$ and $M\\rho_1 b_1/(b_1-1)$, so the Euclidean KPP speed $2\\sqrt{f'(0)}$ is not the tree speed.","Lengthening the first edges or lowering the branching number raises the upper speed bound, so the geometry of the tree directly changes invasion rates.","A polynomial growth condition near zero, in the spirit of the Fujita condition, also forces extinction on homogeneous trees, extending the threshold phenomenon to a broader class of nonlinearities."],"supporting_citations":[{"why":"Supplies the spectral criterion $E_0>0$ for regular trees and the explicit homogeneous-tree formula used in Theorem 3.3.","marker":"[38]"},{"why":"Provides the spectrum-in-gaps computation for the homogeneous-tree Laplacian that the proof of Theorem 3.3 follows.","marker":"[36]"},{"why":"Establishes the hyperbolic-space analogue, threshold $f'(0)>\\lambda_1$ for propagation, which the tree results extend.","marker":"[21]"},{"why":"Gives the Euclidean KPP theory, including $c_0=2\\sqrt{f'(0)}$, which the tree speed bounds modify.","marker":"[2]"},{"why":"Contains the comparison principles cited for Theorems 4.2-4.4, on which the proofs of the main extinction and propagation theorems rest.","marker":"[29]"},{"why":"Supplies the semigroup well-posedness result used as Theorem 4.1 for existence and uniqueness of solutions.","marker":"[43]"}],"fun_headline_variants":["Tree spectral gap decides between spread and die-out","Invasion occurs if source exceeds tree's spectral gap","Spectral gap on trees controls propagation and extinction","Tree geometry picks survival: source above gap wins","Tree Laplacian gap separates spreading from dying out"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole argument leans on a set of comparison results that are cited to a not-yet-published companion paper rather than proved here; if any of those results needs extra hypotheses or fails on non-compact trees, the extinction-propagation dichotomy and both speed bounds would lose their proof.","fun_headline_variants_meta":{"raw":{"variants":["Tree spectral gap decides between spread and die-out","Invasion occurs if source exceeds tree's spectral gap","Spectral gap on trees controls propagation and extinction","Tree geometry picks survival: source above gap wins","Tree Laplacian gap separates spreading from dying out"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001952,"raw_usage":{"total_tokens":7575,"prompt_tokens":829,"completion_tokens":6746,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":445,"completion_tokens_details":{"reasoning_tokens":6673}},"tokens_in":445,"tokens_out":6746,"duration_ms":41677,"temperature":1.0,"reasoning_tokens":6673,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:04:46.522562+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the homogeneous tree with $b=2$ and edge length $r=1$, source $f(u)=\\lambda u(1-u)$, and let $\\lambda$ exceed $E_0=(\\arccos(2\\sqrt{2}/3))^2$. Simulate the Cauchy-Neumann problem on exhausting finite truncated trees with a small compact initial bump near the root; if the solution does not converge to 1 uniformly on compact subsets, Theorem 2.1 is refuted. Alternatively, a counterexample to the order-preserving comparison principle for sub- and supersolutions on a non-compact tree would invalidate both threshold theorems.","supporting_citations":[{"cited_title":"Solomyak, On the spectrum of the Laplacian on regular metric trees , Waves Random Complex Media 14 (2004), 155-171","cited_arxiv_id":null,"evidence_quote":"Supplies the spectral criterion $E_0>0$ for regular trees and the explicit homogeneous-tree formula used in Theorem 3.3."},{"cited_title":"Sobolev & M","cited_arxiv_id":null,"evidence_quote":"Provides the spectrum-in-gaps computation for the homogeneous-tree Laplacian that the proof of Theorem 3.3 follows."},{"cited_title":"Matano, F","cited_arxiv_id":null,"evidence_quote":"Establishes the hyperbolic-space analogue, threshold $f'(0)>\\lambda_1$ for propagation, which the tree results extend."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Euclidean KPP theory, including $c_0=2\\sqrt{f'(0)}$, which the tree speed bounds modify."},{"cited_title":"Punzo & A","cited_arxiv_id":null,"evidence_quote":"Contains the comparison principles cited for Theorems 4.2-4.4, on which the proofs of the main extinction and propagation theorems rest."},{"cited_title":"G. Castelnuovo","cited_arxiv_id":null,"evidence_quote":"Supplies the semigroup well-posedness result used as Theorem 4.1 for existence and uniqueness of solutions."}],"review_version":1}