{"id":"57c3bcf5-662a-4d71-83c3-597d642bdffb","arxiv_id":"1908.03866","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A generalized Neumann kernel boundary integral method computes capacities of generalized condensers with piecewise smooth or slit plates and also yields harmonic measure in multiply connected domains.","lead":"A boundary integral method computes the capacity of generalized condensers, configurations of charged plates with different potential levels and insulating boundaries. The method handles curved and slit plates and is checked against known exact and literature values.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.7's proof is the load-bearing point: it assumes a unique piecewise-constant h for arbitrary piecewise-constant γ, and the linear systems (4.11)/(4.24) stand or fall on it.","rationale":"The reader's conditional verdict is appropriate. The numerical examples (Sections 6-9) and the publicly available MATLAB code give real support for the method: agreement with the exact two-circle capacity (Example 6.1) and with [BSV] for polygonal and slit condensers (Examples 6.2, 8.1) is strong empirical evidence. However, these tests do not directly probe the branch-cut settings of Lemma 3.9 or the uniqueness of the linear systems (4.11)/(4.24) for arbitrary piecewise constant γ. The proof of Lemma 4.7 is abbreviated to a citation plus the observation that f = 0 solves the homogeneous problem; it does not spell out why the uniqueness of h from [N3,WN] applies to a piecewise constant γ, nor does it show that the branch cut chosen in (3.10) preserves the hypotheses. Since Theorem 4.9 uses Lemma 4.7 to identify the solution of (3.15) with Σ a_k f_k, any failure there would break the derivation of (4.11) and hence (2.12). The same reliance is inherited by Theorem 4.22. My own reading found no explicit counterexample, and the logic of Theorem 4.9 is sound once Lemma 4.7 is granted; so the verdict should remain CONDITIONAL, unchanged from the reader's assessment.","tokens_in":25413,"tokens_out":33980,"duration_ms":330206,"concrete_test":"Independently re-derive Lemma 4.7 from the solvability theorems in [WN] and [N3]: verify that for A in (3.14) the index conditions on each boundary component imply unique solvability of Re[A f] = γ + h with a unique piecewise constant h for every Hölder continuous γ, including the piecewise constant γ used in Lemma 4.7, and that the branch cut in Lemma 3.9 preserves the required periodicity. As a numerical cross-check, assemble the homogeneous counterpart of the linear system (4.11) for a nontrivial generalized condenser with δ = 0 and confirm that the matrix is nonsingular and well-conditioned as the discretization size n increases.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Section 4.3) is that the linear systems (4.11)/(4.24), built from solutions of (4.5), determine the flux constants a_k and hence cap(C) via (2.12). Theorem 4.9 derives this using Lemma 4.7, which states that Re[A f] = γ with piecewise constant γ forces f = 0 and γ = 0. The proof of Lemma 4.7 is one paragraph: it invokes 'a unique piecewise constant h' making the Riemann-Hilbert problem uniquely solvable, cites [N3,WN], and then sets h = -γ because f = 0 solves the homogeneous problem. This is valid only if the external theorem guarantees uniqueness of the pair (f,h) for every Hölder continuous γ, not just for the special γ_k in (3.10), and only if the winding-number/branch-cut hypotheses in Lemma 3.9 match those of [N3,WN]. Neither is shown in the paper. Theorem 4.22 is also stated with the proof omitted ('same argument'). If the external theorem does not cover piecewise-constant right-hand sides or the branch-cut-dependent γ_k, the matrix in (4.11) may be singular and the method would not determine a_k, so the capacity formula (2.12) would not follow from the numerical solution.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a boundary integral method for computing the conformal capacity of generalized condensers (B, E, δ) in the complex plane. The plates E_k are closures of simply connected domains bounded by piecewise smooth Jordan curves; rectilinear-slit plates are handled via auxiliary conformal mappings. The mixed Dirichlet–Neumann problem for the potential u is reduced to a Riemann–Hilbert problem, and the generalized Neumann kernel is used to obtain a representation of the boundary values of the auxiliary analytic function in terms of flux constants a_k. The main theoretical results, Theorems 4.9 and 4.22, assert that the a_k satisfy a small dense linear system (4.11) or (4.24); the capacity is then cap(C) = 2π Σ δ_k a_k. The paper includes a MATLAB implementation, numerical validation against exact two-circle capacities and published [BSV] tables, and examples with Cantor dust, Sierpinski carpets, slit condensers, and harmonic measure computations.","tokens_in":25715,"tokens_out":14419,"duration_ms":140191,"significance":"If the theoretical framework is correct, this is the first numerical method for capacities of generalized condensers with piecewise smooth boundaries, and it also computes harmonic measure in multiply connected domains. The method has near-linear complexity in the discretization size, is validated to 6–12 digits against independent benchmarks, and is demonstrated on domains with thousands of boundary components. The full MATLAB code is provided, making the computational claims reproducible. The paper's practical contribution is substantial; the main caveat is the need for a more complete statement and proof of the uniqueness result on which the linear systems rely.","major_comments":[{"comment":"The proof of Lemma 4.7 is load-bearing for Theorems 4.9 and 4.22, but as written it is not self-contained. It invokes 'a unique piecewise constant real-valued function h' making Re[A f] = γ + h uniquely solvable, whereas the only stated result of this type, Theorem 4.3, is proved or cited only for the particular functions γ_k defined in (3.10). Lemma 4.7 requires the external theorem from [N3,WN] to hold for every piecewise constant, indeed every Hölder continuous, right-hand side γ, together with uniqueness of the pair (f,h) under the normalization f(∞)=0 for unbounded G (and the corresponding normalization for bounded G). Please state the general theorem explicitly and verify that the branch-cut/winding-number hypotheses are satisfied under Lemma 3.9; this is what justifies the conclusion h = −γ and hence the invertibility of the linear systems (4.11) and (4.24).","section":"Section 4, Lemma 4.7"},{"comment":"The proof of Theorem 4.22 is omitted with the justification 'the theorem can be proved by the same argument as in the proof of Theorem 4.9.' This is not a routine omission because the Case II linear system (4.24) differs structurally from (4.11): there is no explicit Σ a_k = 0 row, a_m is recovered afterwards from (2.10), and the role of the exterior plate E_m changes the normalization argument in the bounded-G case. The nonsingularity of (4.24) and the derivation of the homogeneous solution are therefore not immediate corollaries of the proof of Theorem 4.9. Please provide a complete proof or a detailed set of modifications.","section":"Section 4, Theorem 4.22"}],"minor_comments":[{"comment":"The sentence 'The integral equation (4.5) been used for computing...' should read 'has been used for computing...'.","section":"Section 4"},{"comment":"The fifth row lists δ2 = 0.15; from the monotone sequence 0.15, 0.30, 0.45, 0.60, 0.90 this appears to be a typo, probably 0.75.","section":"Table 4"},{"comment":"Step 2 ends with 'Γ_j for j = m + 1, m + 2, ..., m'; the upper limit should be m + ℓ.","section":"Algorithm 4.26, step 2"},{"comment":"The proof contains an algebraic identity that is valid only because Σ a_k = 0; a remark at the point where Re[e^{-iθ} log(η−α)] Σ a_k is introduced would improve readability.","section":"Section 3, proof of Lemma 3.9"},{"comment":"The reference [Ku1] is listed but not cited in the text; either cite it or remove it.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong numerical contribution, and the identified proof gaps are repairable rather than fundamental. I would ask the authors to state the general uniqueness theorem for the modified Riemann–Hilbert problem and to supply a full proof of Theorem 4.22. The claim of being the first numerical method for generalized condensers should be checked against the wider literature, since the introduction only surveys regular condensers."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read of Nasser–Vuorinen. The headline is good news: they give the first numerical method for capacity of generalized condensers—multiple plates with different potential levels, mixed Dirichlet/Neumann boundaries, slit geometries—by reducing the problem to a Riemann–Hilbert problem and solving it with the generalized Neumann kernel. That isn't just a repackaging: the linear systems (4.11)/(4.24) that determine the flux constants are new, and they make the capacity calculation (2.12) direct. The code is public, and the validation on regular condensers (two circles, [BSV] tables, Cantor dust) matches exact or published values to 6–12 digits. That is real evidence the machinery works.\n\nThe theoretical chain from (2.6) to (2.12) is standard and, as far as I can tell, correct. The soft spots are real but not load-bearing. Lemma 4.7, which guarantees that the piecewise-constant right-hand side in (4.11) is uniquely determined, gets a one-paragraph proof that leans on an external uniqueness theorem from [N3, WN] stated in weaker form. The stress-test worry is whether that theorem covers arbitrary piecewise-constant γ and the branch-cut-dependent γ_k. I think it does: piecewise constants on disjoint boundary curves are Hölder continuous, and Lemma 3.9 exists precisely to make the γ_k single-valued. But the paper should say what theorem is being used and in what generality. Theorem 4.22 is explicitly unproved ('same argument'), which is fine if true but needs checking. The other soft spot is that Sections 7–8 (true generalized condensers) have no independent benchmarks, so the error is only demonstrated indirectly; a referee should ask for at least one check against a known or asymptotic case. Minor: Table 4 has a duplicated δ2 value (0.15 appears twice, presumably 0.75).\n\nThis is a paper for computational complex analysis and potential theory. It deserves a serious referee; the right recommendation is likely 'major revision' with the requested expansions, not rejection. I'd take it to the reading group.","headline":"A solid, genuinely useful computational paper—its theoretical scaffolding has a couple of soft spots that a referee should ask to be shored up, but the method and validation hold up.","tokens_in":26226,"tokens_out":6150,"would_cite":true,"duration_ms":63863,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65R20","65E05","30C85","31A15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that conformal capacity of generalized condensers with piecewise-smooth or slit plates can be computed numerically by solving generalized Neumann kernel integral equations and one small linear system.","keywords":["conformal capacity","generalized condenser","boundary integral equation","generalized Neumann kernel","Riemann-Hilbert problem","harmonic measure","multiply connected domain","numerical conformal mapping"],"falsifier":"Compute the capacity of the two-circle condenser in Section 6.1 with increasing $n$ and compare it with the exact value $2\\pi/\\log(1/q)$: if the relative error does not converge toward machine precision, the derived equations are not producing the constants $a_k$. A sharper test targets the branch-cut settings of Lemma 3.9: move the auxiliary points across different branch choices and check whether the computed capacity changes.","tokens_in":25212,"feed_emoji":"⚡","tokens_out":8339,"duration_ms":76438,"temperature":0.7,"pith_summary":"The paper establishes a numerical method for the conformal capacity of generalized condensers: domains with several disjoint plates, each held at its own constant potential, with Neumann conditions on the outer boundary of the field. It shows that the mixed boundary-value problem for the potential can be reformulated as a Riemann-Hilbert problem and solved through the generalized Neumann kernel, reducing the capacity to a small linear system once a set of integral equations is solved. The paper presents this as the first numerical method for generalized-condenser capacity. If the method is right, capacities for many-plate geometries such as Cantor-dust and Sierpinski-carpet condensers, as well as rectilinear-slit plates after a conformal mapping, can be computed quickly and accurately. The same computation also yields the potential function and the harmonic measure in multiply connected domains.","feed_headline":"Integral equations compute capacities of multi-plate condensers","feed_subtitle":"Handles curved, square, and slit plates, and also yields harmonic measure in multiply connected domains.","key_machinery":"The load-bearing object is the generalized Neumann kernel $N(s,t)=\\frac1\\pi\\operatorname{Im}\\left(\\frac{A(s)}{A(t)}\\frac{\\dot\\eta(t)}{\\eta(t)-\\eta(s)}\\right)$ and its companion kernel $M$, built from the boundary parametrization $\\eta$ and the Riemann-Hilbert coefficient $A$, which encodes Dirichlet levels on plate boundaries and the Neumann condition on outer boundaries. The key reduction is Theorem 4.3: for each data function $\\gamma_k$ there is a unique $\\mu_k$ solving $(I-N)\\mu_k=-M\\gamma_k$ and a unique piecewise constant $h_k$, so the desired constants $a_k$ satisfy the small linear systems (4.11) or (4.24). Once those constants are known, formula (2.12) converts them directly into the capacity, and the same basis representation of the auxiliary function $f$ gives the potential everywhere.","core_discovery":"The central claim is that for a generalized condenser $C=(B,E,\\delta)$ whose plates are bordered by piecewise smooth Jordan curves or rectilinear slits, the capacity $\\operatorname{cap}(C)$ is obtained numerically from the formula $\\operatorname{cap}(C)=2\\pi\\sum_{k=1}^m \\delta_k a_k$, where the constants $a_k$ are found by solving the generalized Neumann kernel equations $(I-N)\\mu_k=-M\\gamma_k$ and then one linear system whose entries come from the piecewise-constant functions $h_k$. The argument reduces the mixed Dirichlet-Neumann problem for the potential to a Riemann-Hilbert problem for an auxiliary analytic function, uses the unique solvability of the generalized Neumann kernel equation to construct basis functions, and proves that the derived linear systems are non-singular. On this basis the same algorithm computes the potential $u$ in the field by Cauchy's integral formula and the harmonic measure for multiply connected domains.","pith_inferences":["Because the capacity formula is linear in the potential levels $\\delta_k$, one set of constants $a_k$ can be reused for many level assignments; the paper does not advertise this, but it follows directly from (2.12).","The separation into an integral-equation stage and a tiny linear stage suggests the constants $a_k$, which are essentially boundary flux integrals, could serve as building blocks for other conformal invariants that are Dirichlet integrals over the same field.","For slit plates the accuracy of the final capacity is inherited from the separately computed conformal map; a direct integral formulation on slit boundaries would remove that intermediate step, and the boundary conditions derived here give a starting point for such a formulation.","The uniqueness lemma for piecewise-constant right-hand sides, if it extends to other coefficient functions $A$, could yield numerical methods for a wider class of mixed boundary-value problems than the constant-level case treated in this paper."],"forward_implications":["Capacities of generalized condensers with thousands of boundary components, such as Cantor-dust and Sierpinski-carpet approximations, become computable in seconds to minutes.","The same solver computes harmonic measure in multiply connected domains, including piecewise-smooth boundary components, by taking the level vector $\\delta$ to be 1 on one boundary component and 0 on the others.","Rectilinear-slit condensers are covered by first mapping the slit domain conformally to a smooth-bordered domain, since capacity is conformally invariant, so the scope extends beyond Jordan-curve boundaries.","The cost estimate $O(m'(m+\\ell)n\\log n)$ for the integral-equation step, with only a small linear system afterwards, makes large many-plate problems feasible on a laptop."],"supporting_citations":[{"why":"Supplies the fast solver for the generalized Neumann kernel integral equation (4.5) and for the piecewise-constant functions $h_k$.","marker":"[N3]"},{"why":"Provides the Riemann-Hilbert theory and generalized Neumann kernel foundations for the unique solvability used in Theorem 4.3.","marker":"[WN]"},{"why":"Defines generalized condenser capacity and supplies the Dirichlet integral and Green formula behind (2.12).","marker":"[D1]"},{"why":"Gives the reference capacity values for regular condensers against which the method is checked in Section 6.","marker":"[BSV]"},{"why":"Earlier mixed boundary-value method whose structure is simplified here for constant-level boundary conditions.","marker":"[AMN]"},{"why":"Iterative conformal mapping method used to turn slit-bordered domains into smooth-bordered ones before applying the capacity solver.","marker":"[NG]"},{"why":"Earlier application of the generalized Neumann-kernel integral equation to canonical slit conformal mappings, supporting the framework of Theorem 4.3.","marker":"[N2]"}],"fun_headline_variants":["Integral method computes condenser capacity in complex shapes","New algorithm calculates capacity for slit and curved plates","Boundary integrals solve generalized condenser capacity problems","Compute capacity and harmonic measure via integral equations","Generalized condenser capacity from boundary integral method"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The method stands on the unique-solvability theory of the generalized Neumann kernel equations and of the derived linear systems: if those systems can fail to have exactly one solution for the piecewise-constant data used here, the constants $a_k$, and with them the capacity formula, would not be determined by the numerical solution.","fun_headline_variants_meta":{"raw":{"variants":["Integral method computes condenser capacity in complex shapes","New algorithm calculates capacity for slit and curved plates","Boundary integrals solve generalized condenser capacity problems","Compute capacity and harmonic measure via integral equations","Generalized condenser capacity from boundary integral method"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000258,"raw_usage":{"total_tokens":1498,"prompt_tokens":777,"completion_tokens":721,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":393,"completion_tokens_details":{"reasoning_tokens":653}},"tokens_in":393,"tokens_out":721,"duration_ms":6447,"temperature":1.0,"reasoning_tokens":653,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:00:16.734544+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the capacity of the two-circle condenser in Section 6.1 with increasing $n$ and compare it with the exact value $2\\pi/\\log(1/q)$: if the relative error does not converge toward machine precision, the derived equations are not producing the constants $a_k$. A sharper test targets the branch-cut settings of Lemma 3.9: move the auxiliary points across different branch choices and check whether the computed capacity changes.","supporting_citations":[],"review_version":1}