{"id":"52a7a9fe-6adc-484c-a8ce-9b3d24cdbc56","arxiv_id":"2502.01234","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For finite-energy smooth measures, the Revuz map is a homeomorphism between the measure space with the Dirichlet-form metric and the PCAF space with the L2(P_{m+κ+ν0}) local-uniform topology.","lead":"This paper proves that convergence of Revuz measures of finite energy integrals is equivalent to convergence of their positive continuous additive functionals, once the underlying measure is augmented by the killing measure and an energy functional near the cemetery point. The result makes the Revuz correspondence a homeomorphism, which gives a practical tool for proving convergence of time-changed Markov processes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"ν0 is explicitly only subadditive, yet the proof needs polarization and bilinearity for differences of PCAFs; this is not established.","rationale":"The reader identified the regularity of ν0 as the weakest assumption; I agree that ν0 is the critical object. However, the more pointed problem is not merely finiteness of the ν0-integrals but their additivity/bilinearity on the subspace of PCAF differences. The paper itself concedes that ν0 is not additive in general, and the proof of Proposition 1.2 only addresses the diagonal case A=B. Theorem 1.1 and Lemma 4.5 require the off-diagonal identity for differences, so the polarization step is load-bearing. If the limit-of-measures definition of ν0 happens to satisfy the parallelogram law on the relevant functions, the theorem may be correct and the gap is only expository; if not, the claimed homeomorphism and the L2(P_{m+κ+ν0}) topology are not well defined. The proposed test directly checks the parallelogram and the off-diagonal identity in the paper's own nontrivial example. Because the issue is a missing verification rather than a demonstrated contradiction, I would request this check before full acceptance, hence CONDITIONAL rather than REJECT.","tokens_in":26129,"tokens_out":16413,"duration_ms":158318,"concrete_test":"Use the absorbed Brownian motion example in Section 6.6, where ν0 is nontrivial. For the stated PCAFs An and A, compute directly from the defining limit ∫ f dν0 := lim_{t↓0} (1/t)∫ f(x)(φ2(x) − e^{-2t}Ptφ2(x))dm(x) the four quantities Eν0[(gAn∞+gA∞)^2], Eν0[(gAn∞−gA∞)^2], Eν0[(gAn∞)^2], and Eν0[(gA∞)^2]. Check the parallelogram identity Q(An+A) + Q(An−A) = 2Q(An) + 2Q(A). Also compare Eν0[(gAn∞−gA∞)^2] with the ν0-contribution obtained from the RHS of Proposition 1.2 by polarization. If either identity fails, the L2(P_{m+κ+ν0}) topology and Proposition 1.2 for differences are invalid; if both hold, the concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central equivalence (Theorem 1.1) and the energy identity (Proposition 1.2) are applied not only to single PCAFs but to differences An−A. The proof of Proposition 1.2 computes Eν0[(gA∞)^2] for one PCAF and then invokes the polarization identity. This is valid only if Eν0 is a genuine quadratic form on the linear span of the functionals gA∞−gB∞. But Remark 4.2 explicitly states that ν0 is not a measure and satisfies only subadditivity, not additivity, in general. Consequently: (i) the notation L2(P_{m+κ+ν0}) is not justified as a normed space unless the limit defining ∫ f dν0 preserves the parallelogram law and triangle inequality on the relevant subspace; (ii) the step identifying Eν0[(gAn∞−gA∞)^2] with the ρ-squared distance via Proposition 1.2 is not derived from the A=B computation. Lemma 4.1(iv) proves only finiteness of the relevant ν0-integrals, not the needed additivity/bilinearity. Since ν0 is defined as a limit of measures νt = (1/t)(φ2 − e^{-2t}Ptφ2)m, the missing check is whether, after integrating against the functions Ex[(gAn∞−gA∞)^2], the limit exists and respects the parallelogram law. Without this, the topology in Theorem 1.1 may be only a quasi-norm and the 'if and only if' with ρ may fail.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the Revuz correspondence between the set S0 of smooth measures of finite energy integrals and the set of positive continuous additive functionals (PCAFs) of an m-symmetric Hunt process associated with a regular Dirichlet form. The main result, Theorem 1.1, asserts that the Revuz map is a homeomorphism when S0 is equipped with the metric ρ introduced by Nishimori--Tomisaki--Tsuchida--Uemura and the space of PCAFs is equipped with the topology induced by L2(P_{m+κ+ν0}) with the local uniform topology, where κ is the killing measure and ν0 is a newly introduced functional representing the energy of the part of the process continuously escaping to the cemetery point. The key identity is Proposition 1.2, which equates the L2(P_{αm+κ/2+ν0/2}) inner product of the exponentialized PCAF limits to the Dirichlet-form inner product of the corresponding potentials. The paper also proves a strengthening of a known almost-sure convergence result (Theorem 3.7) and a vague-convergence criterion under a boundedness assumption (Theorem 5.2). Examples illustrate the necessity of both κ and ν0.","tokens_in":26441,"tokens_out":12278,"duration_ms":101447,"significance":"If the main theorem is correct, it provides a complete and natural topological characterization of the Revuz correspondence for finite-energy measures, answering a natural question in Dirichlet form theory. The introduction of the functional ν0 and the explicit use of the Beurling--Deny killing measure are novel. The paper is careful to state hypotheses, gives detailed proofs based on standard Dirichlet form machinery, and supplies examples showing that both κ and ν0 are generally needed. The claimed homeomorphism, if established, would be a useful tool for proving convergence of time-changed processes and associated measures.","major_comments":[{"comment":"The reduction 'By using the polarization identity, it is enough to consider the case of A = B' is not justified. The functional ν0 is explicitly stated in Remark 4.2 to be subadditive but not additive in general, so the map Q(A) = E_{ν0}[(gA∞)^2] need not satisfy the parallelogram law, and the cross term E_{ν0}[gA∞ gB∞] cannot be obtained from the diagonal case by polarization. Since Proposition 1.2 is the load-bearing identity used in Theorem 1.1 to identify L2(P_{m+κ+ν0}) convergence of differences of PCAFs with ρ-convergence of their Revuz measures, this gap directly affects the central claim. The authors should either prove additivity/bilinearity of ν0 on the relevant subspace of functions, or prove identity (1.1) directly for distinct A and B by repeating the calculation with cross terms.","section":"Section 4, proof of Proposition 1.2"},{"comment":"The statement that the PCAF space is equipped with the topology induced by the L2(P_{m+κ+ν0})-norm is ambiguous because ν0 is not a measure and is only subadditive. The notation L2(P_{m+κ+ν0}) suggests a genuine normed space, but the paper does not prove that the expression evaluated on differences of PCAFs defines a norm (in particular, that the triangle inequality and the parallelogram law hold). The homeomorphism statement requires a well-defined topology on the PCAF side; otherwise the 'if and only if' in Theorem 1.1 is not meaningful. The authors should define the topology explicitly and establish that it is a norm topology on the relevant set of differences.","section":"Theorem 1.1 and Remark 4.8"}],"minor_comments":[{"comment":"The formula displayed as 'Eν[(gAn∞)2] = 2 R E U 2 An U 1 An 1dν' contains garbled notation and should be typeset and explained more clearly.","section":"Proof of Theorem 3.7"},{"comment":"The statement says the right-hand side of (4.1) is well-defined for an α-excessive function f, but it would be clearer to specify that the limit is an increasing limit for such f, as is standard for energy functionals.","section":"Lemma 4.1(iii)"},{"comment":"In the displayed computation of lim_{n→∞} Ex[A^n_t], the expression 't(1 + sin(nx) sinh t / e^t)' appears to contain a typographical error and should be checked for consistency with the preceding line.","section":"Example 6.1"}],"recommendation":"major_revision","confidential_remarks":"The central idea is promising and the result is likely true, but the proof of the main identity (Proposition 1.2) has a genuine gap: the polarization argument is invalid given the stated subadditivity of ν0. This is not a cosmetic issue; it is the hinge on which the homeomorphism theorem rests. The fix may be straightforward—a direct proof of the cross identity or a careful proof of additivity/bilinearity of ν0 on the relevant functions—so I recommend major revision rather than rejection. I would also ask the authors to tighten the definition of the topology in Theorem 1.1, since the paper currently uses L2 notation for an object that is not an ordinary L2 space."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the main theorem has a load-bearing gap. The result is natural and worth taking seriously, but the proof of Proposition 1.2, stated for pairs of PCAFs, computes only the diagonal case and invokes polarization. Polarization requires the map f to int f d nu_0 to be additive on relevant nonnegative functions. Remark 4.2 says nu_0 is not a measure and satisfies only subadditivity. So the identity E_{nu_0}[gA_infty gB_infty] is not established for distinct A and B. Theorem 1.1 then uses that identity for differences A_n - A to get from L^2(P_{m+kappa+nu_0}) convergence back to rho-convergence. Without a valid bilinear form, that step does not go through. Lemma 4.3 gives one-sided upper bounds, not the two-sided control needed for equivalence. The stress-test note is on target; the reader's clean bill missed this. What is genuinely good: the setup is natural, the counterexamples in Section 6 show that both kappa and nu_0 are needed, and the Section 3 L^1(P_x) convergence theorem is a real strengthening of [24]. The paper is clearly written, the preliminaries are careful, and the construction of nu_0 as a limit of entrance laws is interesting even if the notation int . d nu_0 is doing heavy lifting. The gap is not a typo; it is the central pillar. It might be repairable, e.g., by proving (1.1) for pairs directly via the Revuz formula and Levy system, or by replacing the equivalence with two-sided inequalities plus compactness. But as written, the main theorem is unproven. The paper deserves a serious referee, but not acceptance in this form. I would send it to a specialist with a clear request to verify whether the polarization step can be replaced or repaired. The examples and Section 5 may survive independently, and if the gap is fixed this becomes a solid contribution.","headline":"The paper's main theorem is not proved as written: the key bilinear identity relies on polarization for a functional the paper itself says is only subadditive.","tokens_in":658,"tokens_out":2624,"would_cite":false,"duration_ms":667910,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60J45","31C25","60J55"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the Revuz map is a homeomorphism from the space of finite-energy smooth measures with the metric $\\rho$ to the PCAF space with the $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ local uniform topology.","keywords":["Revuz correspondence","positive continuous additive functionals","finite energy integrals","smooth measures","Dirichlet forms","killing measure","energy functional","homeomorphism"],"falsifier":"Take the absorbed Brownian motion on $(0,\\infty)$ of Example 6.6 with $\\mu_n = n^{3/2}1_{(0,1/n)}dx$ and $\\mu = 0$. Theorem 1.1 predicts that $\\rho(\\mu_n, 0)$ fails to converge to 0 and, equivalently, that $\\mathbb{E}_{m+\\kappa+\\nu_0}[\\sup_{0\\le t\\le T}|A^n_t|^2]$ fails to vanish; the paper proves the first divergence. A direct computation of the second quantity for this explicit process—or for the perturbed Dirichlet form of Example 6.5—would check the equivalence claim from the PCAF side.","tokens_in":25923,"feed_emoji":"🔁","tokens_out":11644,"duration_ms":97751,"temperature":0.7,"pith_summary":"Classical Dirichlet form theory pairs each smooth measure—the Revuz measure—with a positive continuous additive functional (PCAF) of the associated Hunt process. This paper asks whether that pairing respects convergence, and answers yes: restricted to smooth measures of finite energy integrals, the Revuz map is a homeomorphism when measures carry the metric $\\rho$ defined through the Dirichlet form and PCAFs carry the $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ local uniform topology. The topology on PCAFs uses the underlying measure $m$, the killing measure $\\kappa$, and a functional $\\nu_0$ that measures the energy of the part of the process continuously escaping to the cemetery point. The result matters because convergence of PCAFs is what one needs when constructing or taking limits of time-changed processes, while Revuz measures are often the objects one can compute with.","feed_headline":"Revuz map is a homeomorphism for finite energy integrals","feed_subtitle":"Smooth-measure convergence is equivalent to L2 convergence of additive functionals under m, κ, and ν0.","key_machinery":"The load-bearing object is the energy functional $\\nu_0$, defined for non-negative Borel $f$ by $\\int f\\,d\\nu_0 = \\lim_{t\\searrow 0} \\frac{1}{t}\\int f(x)\\,\\mathbb{E}_x[e^{-2\\zeta}1_{\\{\\partial\\}}(X_{\\zeta-})1_{\\{\\zeta\\le t\\}}]\\,dm(x)$; it is not a measure but a monotone, subadditive positive functional representing the energy of the part of the process that continuously escapes to the cemetery point. Along with the underlying measure $m$ and the killing measure $\\kappa$ of the Beurling–Deny decomposition, $\\nu_0$ enters the norm $\\mathbb{E}_{m+\\kappa+\\nu_0}$ and the identity $\\mathbb{E}_{\\alpha m+\\kappa/2+\\nu_0/2}[\\tilde{A}_\\infty \\tilde{B}_\\infty] = E_\\alpha(U_\\alpha\\mu, U_\\alpha\\nu)$. That identity is what carries the argument: it makes convergence of PCAFs in $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ equivalent to convergence of 1-potentials in the $E_1$-norm, i.e., to the metric $\\rho$. The proofs of the continuity statements additionally use the Fukushima decomposition of potentials and the Beurling–Deny decomposition to handle the $\\kappa$ term.","core_discovery":"On the paper's own terms, the central claim is Theorem 1.1: for $\\mu_n, \\mu \\in S_0$ with corresponding PCAFs $A_n, A$, convergence $\\mu_n \\to \\mu$ in the metric $\\rho$ (the $E_1$-norm of the difference of 1-potentials) holds if and only if, for every $T > 0$, $\\mathbb{E}_{m+\\kappa+\\nu_0}[\\sup_{0\\le t\\le T}|A^n_t - A_t|^2] \\to 0$. Equivalently, the Revuz map is a homeomorphism from $(S_0, \\rho)$ onto $A_c^+ \\cap L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ equipped with the local uniform $L^2$ topology. The proof rests on the energy identity (1.1), $\\mathbb{E}_{\\alpha m+\\kappa/2+\\nu_0/2}[\\tilde{A}_\\infty \\tilde{B}_\\infty] = E_\\alpha(U_\\alpha\\mu, U_\\alpha\\nu)$, where $\\tilde{A}_t = \\int_0^t e^{-\\alpha s}\\,dA_s$ and $U_\\alpha\\mu$ is the $\\alpha$-potential; this identity converts $L^2$ convergence of discounted PCAFs into convergence of potentials in the Dirichlet norm, giving both directions of the bicontinuity. The paper also proves a quantitative $L^1(\\mathbb{P}_x)$ continuity theorem for quasi-every $x$ and, under an assumption preventing immediate killing, a vague-convergence theorem going from almost sure PCAF convergence to measure convergence.","pith_inferences":["The energy identity suggests viewing the Revuz map as an isometry: if $\\langle A, B\\rangle = \\mathbb{E}_{m+\\kappa/2+\\nu_0/2}[\\tilde{A}_\\infty \\tilde{B}_\\infty]$ is positive definite on the relevant PCAF space, the homeomorphism is actually an isometry between metric spaces.","Since $\\nu_0$ arises as a limit of entrance laws, one can test whether it coincides with the trace of a genuine measure on the one-point compactification; if so, the theorem would extend to settings where escape-to-cemetery energy is a true measure.","For time-changed processes such as Liouville Brownian motion, the homeomorphism gives a practical route to convergence: verify the Revuz measures converge in $\\rho$ by potential computations instead of constructing the limiting PCAF explicitly.","Tracking the constants in Lemma 4.3 should yield an explicit quantitative version of Theorem 1.1, bounding the PCAF error directly by $E_1(U_1\\mu_n - U_1\\mu)$; the paper's estimates already contain the main terms."],"forward_implications":["If $\\rho(\\mu_n, \\mu) \\to 0$, then for every $T > 0$ the centred $L^2$ sup-norm $\\mathbb{E}_{m+\\kappa+\\nu_0}[\\sup_{0\\le t\\le T}|A^n_t - A_t|^2] \\to 0$, and conversely; measure convergence and PCAF convergence are the same fact.","In the conservative case (no killing, no escape to the cemetery), the criterion collapses to $L^2(\\mathbb{P}_m)$ local uniform convergence, and the energy identity becomes $\\mathbb{E}_{\\alpha m}[\\tilde{A}_\\infty \\tilde{B}_\\infty] = E_\\alpha(U_\\alpha\\mu, U_\\alpha\\nu)$.","The Revuz map restricted to $S_0$ becomes a homeomorphism, so the completeness and separability of $(S_0, \\rho)$ transfer to the corresponding space of PCAFs with the $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ local uniform topology.","Under Assumption 5.1, almost sure local uniform convergence of PCAFs plus boundedness of their second moments forces vague convergence of the Revuz measures, and under the first part of the assumption the potentials $U_1\\mu_n$ converge weakly in the Dirichlet space $E_1$.","The quantitative estimate in Remark 3.9 gives a rate: $\\mathbb{P}_\\nu(\\sup_{t\\le T}|A^n_t - A_t| \\ge \\delta) \\le C_\\nu \\delta^{-1} (E_1(U_1\\mu_n - U_1\\mu))^{1/2}$ for $\\nu \\in S_{00}$."],"supporting_citations":[{"why":"Introduces the metric $\\rho$ on $S_0$ and proves completeness, separability, and subsequential almost sure PCAF convergence from $\\rho$-convergence; Theorem 1.1 upgrades this to bicontinuity with the $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ topology.","marker":"[24]"},{"why":"Supplies the standing framework: the Revuz correspondence for PCAFs and smooth measures, the Fukushima decomposition, the Beurling–Deny decomposition with the killing measure $\\kappa$, and the energy-functional and entrance-law material in Section 5.4.","marker":"[11]"},{"why":"Provides the potential-theoretic lemmas used throughout, including the Fukushima decomposition for potentials (Lemma 5.4.1), the identity $U_\\alpha^A f = U_\\alpha(f\\mu)$, and the $L^2$ estimates from p. 245.","marker":"[17]"},{"why":"Background for excessive measures and energy functionals that underlies the definition and properties of $\\nu_0$.","marker":"[19]"},{"why":"Supplies the local Dirichlet space property of $\\phi_\\alpha$ and the vanishing of $\\nu_0$ on $F \\cap C_c$, identifying $\\nu_0$ as the energy of continuous escape to the cemetery point.","marker":"[32]"},{"why":"Provides the lemma used in Theorem 5.2 to pass from almost sure local uniform convergence of PCAFs to convergence of stochastic integrals against them.","marker":"[27]"}],"fun_headline_variants":["Revuz map homeomorphism for finite energy integrals","Finite energy: Revuz map is a homeomorphism","Measure convergence iff additive functional convergence","Revuz correspondence: bicontinuous for finite energy","Smooth measures and additive functionals: homeomorphic"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof stands on $\\nu_0$ being a well-behaved positive functional—finite, monotone, subadditive, and satisfying the integration identity $\\int U_2\\nu\\,d\\nu_0 = \\int \\mathbb{E}_x[e^{-2\\zeta}1_{\\{\\partial\\}}(X_{\\zeta-})]\\,d\\nu(x)$—so that the $L^2(\\mathbb{P}_{m+\\kappa+\\nu_0})$ norm and the energy identity (1.1) are genuinely defined on the relevant PCAFs; if $\\nu_0$ fails on any approximating sequence, both the topology and the homeomorphism collapse.","fun_headline_variants_meta":{"raw":{"variants":["Revuz map homeomorphism for finite energy integrals","Finite energy: Revuz map is a homeomorphism","Measure convergence iff additive functional convergence","Revuz correspondence: bicontinuous for finite energy","Smooth measures and additive functionals: homeomorphic"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000205,"raw_usage":{"total_tokens":1432,"prompt_tokens":1021,"completion_tokens":411,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":339}},"tokens_in":637,"tokens_out":411,"duration_ms":4241,"temperature":1.0,"reasoning_tokens":339,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T15:57:49.491078+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the absorbed Brownian motion on $(0,\\infty)$ of Example 6.6 with $\\mu_n = n^{3/2}1_{(0,1/n)}dx$ and $\\mu = 0$. Theorem 1.1 predicts that $\\rho(\\mu_n, 0)$ fails to converge to 0 and, equivalently, that $\\mathbb{E}_{m+\\kappa+\\nu_0}[\\sup_{0\\le t\\le T}|A^n_t|^2]$ fails to vanish; the paper proves the first divergence. A direct computation of the second quantity for this explicit process—or for the perturbed Dirichlet form of Example 6.5—would check the equivalence claim from the PCAF side.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the standing framework: the Revuz correspondence for PCAFs and smooth measures, the Fukushima decomposition, the Beurling–Deny decomposition with the killing measure $\\kappa$, and the energy-functional and entrance-law material in Section 5.4."},{"cited_title":"Fukushima, Y","cited_arxiv_id":null,"evidence_quote":"Provides the potential-theoretic lemmas used throughout, including the Fukushima decomposition for potentials (Lemma 5.4.1), the identity $U_\\alpha^A f = U_\\alpha(f\\mu)$, and the $L^2$ estimates from p. 245."},{"cited_title":"Getoor, Excessive Measures","cited_arxiv_id":null,"evidence_quote":"Background for excessive measures and energy functionals that underlies the definition and properties of $\\nu_0$."},{"cited_title":"Explosion by Killing and Maximum Principle in Symmetric Markov Processes","cited_arxiv_id":"2406.15974","evidence_quote":"Supplies the local Dirichlet space property of $\\phi_\\alpha$ and the vanishing of $\\nu_0$ on $F \\cap C_c$, identifying $\\nu_0$ as the energy of continuous escape to the cemetery point."},{"cited_title":"Ooi, Convergence of processes time-changed by Gaussian multiplicative chaos, Potential Analysis (Online first), https://doi.org/10.1007/s11118-025-10206-3","cited_arxiv_id":null,"evidence_quote":"Provides the lemma used in Theorem 5.2 to pass from almost sure local uniform convergence of PCAFs to convergence of stochastic integrals against them."}],"review_version":1}