{"id":"8cf8d38e-4c5f-49d1-9e88-1f8c1f395cdb","arxiv_id":"2411.13126","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors prove well-posedness for stationary nonlocal Hamilton-Jacobi equations with Kirchhoff junction conditions, for integro-differential order below 1, including a flux-limited equivalence.","lead":"This mathematics paper proves existence and uniqueness for nonlocal Hamilton-Jacobi equations on networks where the nonlocal term can reach across several edges at once. A smart generalist should care because it extends a core tool of optimal control and traffic modeling to networks with long-range interactions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"General-network extension (Prop. 7.3, Thm. 7.4) is delegated to \"minor changes\" and \"similar arguments\"; the advertised multi-junction scope is not secured by the written proof.","rationale":"The single-junction existence and comparison arguments are presented in real detail: the barriers in Lemma 4.1, the fixed-point iteration in Lemma 4.2, the Lipschitz estimates in Proposition 4.3 and Theorem 4.4, and the comparison proof in Theorem 5.1 all contain explicit computations and are internally consistent on a first reading. The reader's conditional verdict therefore seems appropriate, and no clear mathematical error emerged in Theorems 4.4 or 5.1. The most load-bearing weakness is instead in the advertised extension to general networks. Proposition 7.3 is the backbone of the multi-junction existence argument, but its proof is reduced to \"same lines ... with minor changes\"; Theorem 7.4's uniqueness is compressed into \"follows readily\" from the single-junction comparison. The single-junction Lipschitz and comparison proofs use a distinguished branch and a Kirchhoff condition at exactly one vertex; when an edge connects two interior vertices, those proofs need a genuinely local control at both endpoints, and the interaction between the two junction conditions is not written out. This is a rigor gap in a claim central to the abstract, so the paper should remain conditional pending a full verification of the multi-junction case.","tokens_in":44725,"tokens_out":46923,"duration_ms":486220,"concrete_test":"Write out the full analogue of Lemma 5.2 and Theorem 4.4 Step 2 for the minimal multi-junction network Gamma = E1-bar union E2-bar union E3-bar with two interior vertices v1,v2 joined by E2 and no boundary vertices, and check whether the trial maximum in (4.27) can be ruled out when it falls at the non-distinguished interior vertex. Concretely, choose x0 in E2 near v1, use the modified test function, and verify that the Kirchhoff conditions at both v1 and v2 are compatible with a single choice of K,L. If not, supply a modified local barrier; if none exists, Theorem 7.4's uniqueness assertion is not established.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Proposition 7.3 is asserted to follow from Lemma 4.2 and Proposition 4.3 \"with minor changes\", and Theorem 7.4 says uniqueness \"follows readily\" from Theorem 5.1. These are not cosmetic gaps: the local arguments in Sections 4-5 use a distinguished edge containing the trial point and a test function whose slopes at the junction are dictated by one edge, and the contradiction uses the single Kirchhoff flux sum_i -phi_{x_i}(O)=B. On a network with two or more interior vertices, an edge can connect two junctions, so the same trial maximum can fall at a second interior vertex; the modified test function must then be compatible with flux conditions at both vertices, and the Lipschitz estimate of Lemma 5.2 must be proved locally for every interior vertex simultaneously. The paper does not supply the required modifications, nor does it prove that the Poincare-Miranda construction of Theta* depends only on the estimates actually established. If any of these steps fails for a two-junction network, Theorem 7.4 and the abstract's general-network claim are unsupported, even if the single-junction Theorems 4.4 and 5.1 are correct.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the stationary nonlocal Hamilton-Jacobi equation λu − I_i u + H_i(x,u_{x_i}) = 0 on each edge of a network, complemented by a Kirchhoff-type condition at interior vertices and Dirichlet conditions at boundary vertices. The nonlocal operators are of Lévy type with order σ < 1 and may integrate over the whole network, including contributions from several different edges. For a single junction (Theorem 4.4 and Theorem 5.1), the authors prove existence of a continuous viscosity solution by vanishing viscosity and Perron's method, and a strong comparison principle, under coercivity and Lipschitz assumptions on the Hamiltonians. They also introduce a nonlocal analogue of flux-limited solutions and prove its equivalence with Kirchhoff-type solutions (Section 8). The final section claims an extension of the well-posedness results to general networks (Proposition 7.3 and Theorem 7.4).","tokens_in":44906,"tokens_out":13905,"duration_ms":134894,"significance":"The single-junction results are a genuinely new contribution: they are, to my knowledge, the first to treat nonlocal Hamilton-Jacobi equations on networks where the nonlocal operator couples different edges, and the comparison proof avoids the usual doubling-of-variables procedure by exploiting the Lipschitz regularity of subsolutions. The flux-limited equivalence in Section 8 is also new in the nonlocal setting. The paper is clearly organized, provides detailed estimates for the core theorems, and includes a useful self-contained comparison principle for censored problems in Appendix B. However, the advertised general-network extension is only sketched via 'minor changes' and 'readily follows', and the specific multi-vertex issues are not addressed in the written proof; as a result the paper's central claim in the abstract and Section 7 is not fully supported.","major_comments":[{"comment":"Proposition 7.3 is asserted to follow from Lemma 4.2 and Proposition 4.3 'with minor changes', but the multi-vertex setting introduces genuinely new structure. In Lemma 4.2 the iterative construction and the contraction estimate (4.12) use a single scalar parameter θ at the unique interior vertex and split indices into censored and non-censored edges; for a general network the Dirichlet data are a vector Θ=(θ_\\bar v)_{\\bar v\\in V_i}, an edge can have interior vertices at both endpoints, and the zero-order coefficient Λ^η_i in the counterpart of (4.10) must account for contributions from all edges incident at both endpoints. In addition, the uniqueness of the solution to (7.7)-(7.8) requires the network version of the comparison principle from Lemma B.1, which is only covered by a remark. As written, the well-posedness of the viscous Dirichlet problem on a general network is not established.","section":"Section 7, Proposition 7.3"},{"comment":"The existence proof for a general network relies on the Poincaré-Miranda theorem applied to the map Θ ↦ F_j(Θ) = ∑_{E∈Inc(v_j)} −∂_E u^{ε,Θ}(v_j) − B_{v_j}. For this application one needs the continuity of Θ ↦ u^{ε,Θ} with respect to the whole vector Θ, and one needs uniform (in ε) Lipschitz estimates of u^{ε,Θ*_ε} in a neighborhood of every interior vertex in order to pass to the limit ε→0. The first property is only sketched for the scalar case in Proposition 4.3(v), and the second is proved in the star case using the special test function (4.27) whose kink satisfies the single junction condition K−(N−1)L≤B. The paper does not prove the network analogues of these two ingredients, so the existence half of Theorem 7.4 is not secured by the written argument.","section":"Section 7, Theorem 7.4 (existence part)"},{"comment":"The uniqueness assertion says it 'follows readily' from Theorem 5.1 'arguing locally as in the junction case.' This is a nontrivial reduction. Theorem 5.1 relies on Lemma 5.2, whose proof uses the star-shaped geometry: the maximum of u(x)−u(x_0)−φ(x) is considered with φ defined as in (4.27), and the possibility of the maximum at O is excluded using the single Kirchhoff inequality. On a network, after localizing at an interior vertex v, the auxiliary function must be extended over the whole graph, and the maximum can be attained at a different interior vertex w; the slope constants must then simultaneously satisfy flux inequalities at v and w. The required modification is not provided, and the network version of the local Lipschitz estimate for subsolutions is not proved. Consequently the comparison principle for general networks is not established.","section":"Section 7, Theorem 7.4 (uniqueness part)"}],"minor_comments":[{"comment":"In the displayed equality in the converse part of the proof, the subscript of φ in the first term is written as φ_{x_{i0}}(O); this should presumably be φ_{x_i}(O). The intended identity G_i^δ = G_i + I_i u(O) − I_i[B_c^δ(O)]u(O) − I_i[B_δ(O)]φ(O) is otherwise clear.","section":"Section 3, proof of Lemma 3.3"},{"comment":"The theorem states that u solves the Dirichlet problem (2.12)-(2.14)-(2.13), with (2.14) written as u_i(a_i)=h_i pointwise. The proof only establishes the relaxed Dirichlet condition for the supersolution side, as acknowledged in Remark 4.5(ii). The statement should clarify that the Dirichlet condition is understood in the viscosity sense of Definition 3.1, not necessarily pointwise.","section":"Section 4, Theorem 4.4 and Remark 4.5(ii)"},{"comment":"The final sentence says 'there exists a unique continuous viscosity solution u ∈ C^{0,1}(Γ_δ)', but Γ_δ is defined in (4.26) and is a set that depends on δ; the sentence appears to refer to the solution from Theorem 4.4. For readability, the notation should distinguish the Lipschitz-regularity statement from the comparison conclusion.","section":"Section 5, Theorem 5.1"},{"comment":"The general network is introduced in the introduction as made of curves, but (7.1) defines every edge as a straight segment. If curved edges are intended, the parametrization and the definition of the inward derivative (7.2) need to be adapted; otherwise the introductory wording should be corrected to 'straight segments'.","section":"Section 7.1, equation (7.1)"}],"recommendation":"major_revision","confidential_remarks":"The single-junction core of the paper is mathematically sound and valuable, and I would be glad to see it published. The main issue is that the general-network extension, which appears in the title and abstract, is not actually proved: Proposition 7.3, the Poincaré-Miranda existence step, and the network uniqueness argument are all delegated to sketches that omit the multi-vertex complications. If the authors are willing to either add a full proof of the network case (perhaps in an appendix) or restrict the advertised results to the single-junction case, the paper would be suitable. I would not recommend rejection conditional on the core results, but the current manuscript's scope claim is not met."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the single-junction well-posedness for nonlocal Hamilton-Jacobi equations whose kernels see several edges at once. Theorem 4.4 and Theorem 5.1 are real theorems with real estimates: the barriers, the vanishing-viscosity construction, the flux-limited equivalence in Section 8, and the censored-problem appendix all hang together. The order-σ<1 restriction is explicit and sensible. This is a solid piece of analysis and a legitimate step beyond the local network literature.\n\nWhere it gets soft is the general-network section. Proposition 7.3 is dismissed with \"same lines ... with minor changes,\" and Theorem 7.4 says uniqueness \"follows readily\" from the single-junction comparison. The stress-test note is on target: the local arguments use a distinguished edge and the single Kirchhoff sum at one vertex, and on a network with two interior vertices connected by an edge, a trial maximum could land at a second junction. The test-function construction and the Lipschitz lemma would need to be genuinely adapted, not just relabeled. The authors do not write those modifications. I suspect the claims are true and can be repaired, but as written the abstract overstates what is proven.\n\nThe other honest limitation is the asymmetry in Lemma 5.3: part (ii), the stronger junction inequality, is proved for subsolutions (which are Lipschitz by Lemma 5.2) but not for supersolutions, so the comparison proof only uses part (i) for supersolutions. The authors flag this clearly, and the comparison still works, but it is a real restriction worth remembering.\n\nI don't see circularity or fitting. The coercivity assumption (2.6)(iii) is load-bearing, but it is stated up front and is standard in this control-motivated setting. The citations, including the authors' own prior work with Barles-Chasseigne, are legitimate support rather than padding.\n\nBottom line: the single-junction theorems deserve a serious referee and probably publication after reasonable revision. The general-network theorems need either a completed proof or a humbler statement. I'd send it to review, but I'd tell the editor to make the authors address Section 7 honestly.","headline":"Solid single-junction theory for nonlocal Hamilton-Jacobi on networks; the advertised general-network extension is sketched, not fully proved.","tokens_in":45485,"tokens_out":1920,"would_cite":true,"duration_ms":21861,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35K55","35R09","47G20","35B40","33B20"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that nonlocal Hamilton-Jacobi equations on networks with Kirchhoff junction conditions have a unique continuous viscosity solution when the nonlocal operators have order below 1 and the Hamiltonian grows at least linearly.","keywords":["Nonlocal Hamilton-Jacobi equations","Networks","Kirchhoff conditions","Viscosity solutions","Comparison principle","Flux-limited solutions","Integro-differential operators","Junction conditions"],"falsifier":"On a two-edge junction with $H_i(x,p)=p^2/2$, kernels $\\nu_{ij}(r)=\\Lambda r^{-1-\\sigma}$ with $\\sigma=1/2$, $B=0$, and $h_1=h_2=0$, compute the vanishing-viscosity limit for several choices of the junction value $\\theta$ and check that the limit is the same function and satisfies the Kirchhoff condition; any dependence on the approximation path, or any discrepancy between this limit and the Perron solution, would refute Theorem 4.4.","tokens_in":44483,"feed_emoji":"🔀","tokens_out":5977,"duration_ms":54041,"temperature":0.7,"pith_summary":"The paper proves that a stationary nonlocal Hamilton-Jacobi equation on a network, with a Kirchhoff flux condition at interior vertices and Dirichlet data at boundary vertices, has a unique viscosity solution. The key advance is that the nonlocal operator may receive contributions from several edges at once, not just the edge where the equation is written, as long as the kernels have order strictly below 1. The authors construct the solution in two ways, by vanishing viscosity and by Perron's method, prove a strong comparison principle, and show that the Kirchhoff solution is exactly a flux-limited solution for a suitable flux limiter. If correct, this gives a well-posedness theory for a large class of nonlocal junction problems and connects it to control-theoretic solution notions.","feed_headline":"Nonlocal Kirchhoff junctions admit unique viscosity solutions","feed_subtitle":"For integro-differential order below 1, solutions exist, compare, and match flux-limited ones.","key_machinery":"The central object is the junction viscosity solution defined through test functions $\\varphi \\in C^1(\\Gamma)$ and the truncated operator $G^\\delta_i$; at the junction $O$ the subsolution inequality is a minimum of the PDE inequalities on all incident edges and the Kirchhoff flux condition. The nonlocal operator $I_i$ with kernels $\\nu_{ij}$ satisfying (2.7) is the device that carries information across edges, and its Levy integrability condition $\\sigma<1$ makes $I_i u(O)$ finite for Holder functions $u$. The coercivity inequality $C_H^{-1}|p|-C_H \\le H_i(x,p)$ is the workhorse: it builds barriers, gives Lipschitz estimates, and rules out large slopes in the comparison proof. The vanishing viscosity family $u^\\varepsilon$ solves a Dirichlet problem at $O$ with value $\\theta$, and the Kirchhoff data $B$ is recovered by choosing $\\theta$ through continuity, or by the Poincare-Miranda theorem on general networks, after which $\\varepsilon \\to 0$ yields the solution.","core_discovery":"The central claim is that, under assumptions (2.15), the Kirchhoff-Dirichlet problem (2.12)-(2.14)-(2.13) has a viscosity solution $u \\in C(\\Gamma)$, Holder continuous on $\\Gamma$ and locally Lipschitz away from the boundary vertices, and that any viscosity subsolution lies below any viscosity supersolution, so the solution is unique. The proof's core is that order-$\\sigma<1$ nonlocal terms evaluated at the junction remain finite and continuous for Holder functions, so the junction condition can be stated in the Lions-Souganidis junction-viscosity sense; coercivity of the Hamiltonian then substitutes for ellipticity and forces Lipschitz behaviour near the junction. Section 8 closes the circle by proving that Kirchhoff solutions are flux-limited solutions for a flux limiter built from the nonincreasing part of the Hamiltonian, and vice versa.","pith_inferences":["The comparison proof only needs Lipschitz regularity of subsolutions, so the reliance on coercivity suggests that a non-coercive Hamiltonian with, say, superlinear growth away from the junction might still admit comparison if slope control comes from another source; this is an extension, not a paper claim.","The flux-limited equivalence is proved for Hamiltonians that are convex with a unique minimum; a natural test is whether the same equivalence holds for merely quasiconvex Hamiltonians, as in the local Imbert-Monneau setting.","Because the junction arguments are local, the same framework should apply to time-dependent problems obtained by adding a time derivative, although the paper treats only the stationary equation.","A numerical implementation on a two-edge junction with $\\sigma=1/2$ and $H_i(x,p)=p^2/2$ would provide a concrete check of the predicted uniqueness and of the flux-limited characterization."],"forward_implications":["For each set of data satisfying (2.15), the Kirchhoff-Dirichlet problem has a unique Holder-continuous viscosity solution on any finite connected network.","The vanishing viscosity limit and Perron's construction give the same solution, with explicit Lipschitz control near interior vertices.","The Kirchhoff solution coincides with the flux-limited solution for the flux limiter defined in Definition 8.1, extending the local equivalence to nonlocal equations.","Strictly elliptic viscous versions of the problem have unique classical $C^{2,1-\\sigma}(\\Gamma)$ solutions satisfying the vertex conditions pointwise.","Other boundary conditions, including unbounded edges, exterior data, state constraints, and Neumann conditions, can be handled by the same junction-local arguments."],"supporting_citations":[{"why":"Supplies the junction viscosity solution framework and the well-posedness strategy for Kirchhoff conditions.","marker":"[31]"},{"why":"Provides the comparison arguments at the junction that the proof adapts to nonlocal operators.","marker":"[32]"},{"why":"Introduces flux-limited solutions whose nonlocal analogue the paper defines and compares.","marker":"[27]"},{"why":"Contributes the idea of solving with a Dirichlet value at the junction and tuning it by the Intermediate Value Theorem to recover the Kirchhoff condition.","marker":"[34]"},{"why":"Offers the technical comparison lemmas and the flux-limiter characterization used in Sections 5 and 8.","marker":"[11]"},{"why":"Gives the Holder regularity theorem for coercive integro-differential equations used to extend the limit solution to the boundary.","marker":"[15]"},{"why":"Supplies the Poincare-Miranda theorem used to choose Kirchhoff data at several interior vertices.","marker":"[30]"}],"fun_headline_variants":["Nonlocal Kirchhoff junction solutions exist and are unique","Sub-1-order nonlocal H-J on networks: Kirchhoff uniqueness","Kirchhoff nonlocal junction: existence + uniqueness","Nonlocal H-J on networks: Kirchhoff uniqueness beyond order 1","Flux-limited solutions match Kirchhoff nonlocal problem"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proofs need the Hamiltonian to grow at least linearly in the slope on every edge, namely $C_H^{-1}|p|-C_H \\le H_i(x,p)$, because that growth is what keeps subsolutions Lipschitz near the junction and rules out slopes that would violate the flux; if coercivity is dropped, the existence and comparison arguments no longer go through.","fun_headline_variants_meta":{"raw":{"variants":["Nonlocal Kirchhoff junction solutions exist and are unique","Sub-1-order nonlocal H-J on networks: Kirchhoff uniqueness","Kirchhoff nonlocal junction: existence + uniqueness","Nonlocal H-J on networks: Kirchhoff uniqueness beyond order 1","Flux-limited solutions match Kirchhoff nonlocal problem"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000635,"raw_usage":{"total_tokens":2901,"prompt_tokens":891,"completion_tokens":2010,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":507,"completion_tokens_details":{"reasoning_tokens":1926}},"tokens_in":507,"tokens_out":2010,"duration_ms":39522,"temperature":1.0,"reasoning_tokens":1926,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T16:48:20.368572+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a two-edge junction with $H_i(x,p)=p^2/2$, kernels $\\nu_{ij}(r)=\\Lambda r^{-1-\\sigma}$ with $\\sigma=1/2$, $B=0$, and $h_1=h_2=0$, compute the vanishing-viscosity limit for several choices of the junction value $\\theta$ and check that the limit is the same function and satisfies the Kirchhoff condition; any dependence on the approximation path, or any discrepancy between this limit and the Perron solution, would refute Theorem 4.4.","supporting_citations":[{"cited_title":"Viscos ity solutions for junctions: well posedness and stability","cited_arxiv_id":null,"evidence_quote":"Supplies the junction viscosity solution framework and the well-posedness strategy for Kirchhoff conditions."},{"cited_title":"Well-p osedness for multi-dimensional junction problems with Kirchoff-type conditions","cited_arxiv_id":null,"evidence_quote":"Provides the comparison arguments at the junction that the proof adapts to nonlocal operators."},{"cited_title":"Flux-limited solutio ns for quasi-convex Hamilton- Jacobi equations on networks","cited_arxiv_id":null,"evidence_quote":"Introduces flux-limited solutions whose nonlocal analogue the paper defines and compares."},{"cited_title":"Quasi linear parabolic pde posed on a netwo rk with non linear Neumann boundary condition at vertices","cited_arxiv_id":null,"evidence_quote":"Contributes the idea of solving with a Dirichlet value at the junction and tuning it by the Intermediate Value Theorem to recover the Kirchhoff condition."},{"cited_title":"On Modern Approaches of Hamilton-Jacobi Equations and Control Problems with Discontinuities , volume 104 of Progress in Nonlinear Differential Equations and their Applications","cited_arxiv_id":null,"evidence_quote":"Offers the technical comparison lemmas and the flux-limiter characterization used in Sections 5 and 8."},{"cited_title":"Regularity results and large time behavior for integro-differential equations with coer cive Hamiltonians","cited_arxiv_id":null,"evidence_quote":"Gives the Holder regularity theorem for coercive integro-differential equations used to extend the limit solution to the boundary."},{"cited_title":"The Poincar´ e-Miranda theorem","cited_arxiv_id":null,"evidence_quote":"Supplies the Poincare-Miranda theorem used to choose Kirchhoff data at several interior vertices."}],"review_version":1}