{"id":"d99acef0-e88f-4d4e-9ff7-b52908811cfe","arxiv_id":"1908.00755","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper explicitly parametrizes all free additive Lévy functions of the second kind (for initial law δ0) as φ = ψ∘Φ, where Ψ is a conformal map whose image contains the upper half-plane and ψ is a Nevanlinna function.","lead":"This mathematics paper classifies the free Lévy processes of the second kind, those with time-homogeneous transition probabilities, by parameterizing the analytic functions that describe them. The result, called the nonlinear free Lévy-Khinchine formula, turns a hard-to-check condition into an explicit geometric parametrization using conformal maps.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main theorem is proved only for µ0=δ0 but stated without that restriction; this scope gap is the key concern.","rationale":"The reader's verdict of CONDITIONAL is appropriate. The proof of the direct direction (§4.2.1) is sound: given ψ with Ψ(C+)⊃C+, the flow Ft=Ψ(Φ+t) is well-defined and φ=ψ∘Φ satisfies the FAL2 conditions; the argument that φ(iy)/(iy)→0 by ruling out a<0 is a bit terse but correct. The converse (§4.2.2) contains the Ω dichotomy; the step 'Φ takes values with arbitrarily high imaginary parts' is asserted without proof, but it can be justified from the two cases of Proposition 3.6 and the asymptotic -1/φ(iy), so I do not see a fatal gap. The sign of Im Ψ' in §4.2.2 is stated as negative, which conflicts with the later conclusion Ψ'=-ψ with ψ Nevanlinna (positive imaginary part); this appears to be a typo and does not affect the argument. The only issue that changes the theorem's scope is the unstated µ0=δ0 restriction. The manuscript itself flags it in §2.3, so it is not hidden, but the abstract and Theorem 4.1 should carry it. This is a presentation/scope problem, not a correctness problem, hence CONDITIONAL remains the right verdict. The proposed test (e.g., with µ0=δ_a) would settle whether the restriction is removable.","tokens_in":13085,"tokens_out":48008,"duration_ms":402679,"concrete_test":"Choose a concrete FAL2 function from the parametrization, e.g., φ(z) = -√(2z) (with ψ(z) = -z, Ψ(z) = z²/2), and test the condition of Theorem 2.2 for an initial distribution µ0 = δ_a with a ≠ 0: compute F_{µ0} for δ_a as in [4] and check whether φ∘F_t^{-1}∘F_{µ0}^{-1} has an analytic continuation to C+ with values in C−. If this fails, Theorem 4.1 as stated is false for general µ0; if it holds for a range of a, the restriction may be removable and the theorem needs only a qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's core result, Theorem 4.1, is proved only under the initial-distribution restriction µ0=δ0, as acknowledged in §2.3 ('at least in the case µ0 = δ0'). The abstract and the statement of Theorem 4.1, however, present the characterization without this hypothesis. Since the definition of FAL2 functions in Theorem 2.2 explicitly involves the initial distribution through the factor F_{µ0}^{-1} in the continuation condition (ϕ∘F_t^{-1}∘F_{µ0}^{-1}), the collection of functions being parametrized may depend on µ0. Nothing in §4 addresses µ0 ≠ δ0, so the headline claim is broader than what is established. A secondary point: the converse in §4.2.2 asserts that φ(iy)/iy→0 implies Φ takes values with arbitrarily high imaginary parts, which is used for the Ω dichotomy; this step is plausible and likely correct, but it is not fully justified in the text. The µ0 restriction is the load-bearing gap because it directly affects the scope of the central theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies free Lévy processes of the second kind, i.e., families of probability measures (µ_t) and (µ_{s,t}) satisfying µ_s ⊞ µ_{s,t} = µ_t and µ_{s,t} ⊞ µ_{t,u} = µ_{s,u}, whose associated transition kernels are time-homogeneous. By a previous characterization of the author (Theorem 2.2), such processes correspond to certain Nevanlinna functions φ, called FAL2 functions, satisfying an asymptotic condition and a continuation condition. The present paper gives an explicit parametrization of these functions in terms of conformal mappings: φ = ψ ∘ Φ, where Ψ is a univalent function with derivative Ψ′ = −ψ (ψ Nevanlinna) and Ψ(C+) contains C+. This is called the nonlinear free Lévy-Khinchine formula. The proof uses primitives of Nevanlinna functions, starlike domains at −∞, and a necessary and sufficient geometric criterion (Proposition 3.6). The paper also establishes analogues for free multiplicative convolution on the circle and on the positive half-line.","tokens_in":13241,"tokens_out":11169,"duration_ms":95698,"significance":"If correct, the parametrization closes a question left open in [4] and provides a complete, geometric description of a class of free Lévy processes that is of interest in free probability. The result is original and the proof of the additive theorem is essentially constructive, with both directions established. The geometric criterion in Proposition 3.6 is a useful contribution in its own right, and the multiplicative results (in particular the fact that all FUL2 functions are constant) are striking. However, the main theorem is proved only for the initial distribution µ0 = δ0, and this restriction is not stated in the abstract or in Theorem 4.1; because the defining condition for FAL2 functions involves µ0, the scope of the headline claim is broader than what is proved. The paper is otherwise sound in its central derivation.","major_comments":[{"comment":"The main theorem is stated for an arbitrary 'FAL2 function', but the proof is carried out only under the restriction µ0 = δ0, which is acknowledged in §2.3 ('at least in the case µ0 = δ0'). The defining condition of FAL2 functions in Theorem 2.2 explicitly involves the initial distribution through the factor F_{µ0}^{-1} in the continuation condition (φ∘F_t^{-1}∘F_{µ0}^{-1}), and the simplified condition (ii) used in §4.1 is the µ0 = δ0 case, since F_{δ0}^{-1} is the identity. The set of functions satisfying the condition may therefore depend on µ0, and no argument is given for µ0 ≠ δ0. Please either add the hypothesis 'µ0 = δ0' to the statements of Theorem 4.1 and the abstract, or extend the proof to general µ0.","section":"Theorem 4.1 and §2.3"},{"comment":"The step 'Since φ(iy)/iy→0 as y→∞ the function Φ takes values with arbitrarily high imaginary parts, therefore Ω is either the whole complex plane, or a translate of the upper half plane' is load-bearing for the exhaustiveness of the parametrization in Theorem 4.1, but it is not justified in the text. It should be proved that Φ(iy) has unbounded imaginary part using the Nevanlinna representation of 1/φ and the asymptotic condition φ(iy)=o(y), and that this, together with the real-translation invariance of Ω, forces Ω to be C or a translate of C+. Please expand this argument.","section":"§4.2.2"}],"minor_comments":[{"comment":"Condition (ii) in §4.1 is stated as 'φ∘F_t^{-1} has an analytic continuation', whereas Theorem 2.2 requires 'φ∘F_t^{-1}∘F_{µ0}^{-1}' for general µ0. Since the paper works under µ0=δ0, this should be stated explicitly at this point to avoid confusion.","section":"§4.1"},{"comment":"In equations (24) and (26), the second relation is written with ⊞ instead of ⊠, which appears to be a typo; the operation should be consistent with the multiplicative convolution.","section":"Theorems 5.2 and 5.5"},{"comment":"The proof of the converse part of Proposition 3.3 is delegated to [5] without even a sketch; since this proposition is one of the pillars of the geometric criterion, a brief indication of the argument (or an appendix) would make the paper more self-contained.","section":"Proposition 3.3"},{"comment":"There are several typographical errors, e.g., 'Nevanlina' in Theorem 4.1, 'revover' in §2.1.2, and 'configuration' in the introduction; these should be corrected in the final version.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is a valuable contribution and the core idea is sound, but the mismatch between the statement and the proof of the main theorem needs to be resolved before publication. If the author can restrict the theorem and abstract to µ0 = δ0 (or prove the general case), I would support acceptance. The secondary issue about the dichotomy in §4.2.2 also needs a more detailed justification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Biane proves the nonlinear free Lévy-Khinchine formula for second-kind free Lévy functions: an explicit parametrization of FAL2 functions via conformal mappings, where the difficult continuation condition from Theorem 2.2 is replaced by the geometric condition that Ψ(C+) contains C+. Both directions of the additive theorem are proved, and the multiplicative analogues are worked out, including the notable corollary that all unitary FUL2 functions are constant. This is genuine progress on an open problem from his earlier paper, and the parametrization is much more usable than the original characterization.\n\nThe proof is careful. Proposition 3.6 gives clean necessary and sufficient conditions on the Nevanlinna data for the image to contain a translate of the upper half-plane, and the flow argument in §4.2.1 is transparent. The delegation of Proposition 3.3 to Pommerenke is acceptable for a specialized audience. The reciprocal statement in §4.2.2 is a bit compressed, but the structure is sound.\n\nThe main soft spot is the scope gap flagged in §2.3: the whole parametrization is carried out under μ0 = δ0, but Theorem 4.1 and the abstract state it without that restriction. This is not just an omission of a technical hypothesis. The FAL2 class itself is defined relative to μ0 through the continuation condition of Theorem 2.2, so the set of functions being classified may depend on μ0. The paper proves the exhaustive parametrization only for μ0 = δ0. The result may well extend to general μ0, but that is not shown. This should be fixed either by proving the general case or by stating the theorem with the restriction and explicitly saying the general case is open.\n\nA smaller issue: the step in §4.2.2 where φ(iy)/iy→0 is used to infer that Φ takes values with arbitrarily high imaginary parts is plausible but not fully justified in the text. This is minor; it can be filled without changing the argument.\n\nThe citation pattern is fair; the paper builds directly on [4] and the reliance there is legitimate. No self-citation inflation.\n\nWho is this for? Specialists in free probability and operator algebras, and anyone interested in conformal mapping methods for semigroups of analytic functions. It deserves serious refereeing; the main theorem is likely correct and the parametrization is a useful tool. I would send it to a competent referee, with the instruction to make sure the μ0 restriction is stated honestly.","headline":"Explicit parametrization of FAL2 functions, proved for μ0 = δ0 but stated more broadly; the math is sound but the scope claim needs correction.","tokens_in":13770,"tokens_out":2804,"would_cite":true,"duration_ms":26427,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L54","60G51","30C35","60E07"],"pacs":[],"model":"deepseek-v4-flash","headline":"A nonlinear free Lévy-Khinchine formula represents every free additive Lévy function of the second kind as ψ∘Φ, where Ψ is a univalent map whose image contains the upper half-plane.","keywords":["free probability","free Lévy processes","Nevanlinna functions","free Lévy-Khinchine formula","conformal mapping","starlike domains","free additive convolution","subordination"],"falsifier":"Find a FAL2 function $\\varphi$ whose associated extended domain $\\Omega=\\bigcup_{t\\geq0}(\\Phi(\\mathbb{C}_+)-t)$, with $\\Phi$ a primitive of $-1/\\varphi$, is neither the whole plane nor a translate of the upper half-plane. Such an example would break the dichotomy used in the converse direction of the proof and show the parametrisation is not exhaustive.","tokens_in":12836,"feed_emoji":"📐","tokens_out":12839,"duration_ms":112747,"temperature":0.7,"pith_summary":"The paper solves a classification problem from the theory of free Lévy processes: which processes with homogeneous transition probabilities are possible? It proves that every non-constant such process, starting from a fixed point, is generated by a conformal pair: a Nevanlinna function $\\psi$ and a univalent primitive $\\Psi$ of $-\\psi$ whose image $\\Psi(\\mathbb{C}_+)$ contains $\\mathbb{C}_+$. The transition semigroup is then the conjugate of horizontal translation, given by $F_t(z)=\\Psi(\\Phi(z)+t)$ with $\\Phi=\\Psi^{-1}$, and the process's Nevanlinna function is the composition $\\varphi=\\psi\\circ\\Phi$. Because the admissible $\\Psi$ form a convex set of starlike domains, this is a Lévy-Khinchine-type parametrisation, though one that is nonlinear in the parameter.","feed_headline":"A conformal recipe classifies homogeneous free Lévy processes","feed_subtitle":"All time-homogeneous free Lévy transition laws take the form ψ∘Φ for a conformal pair Ψ,Φ.","key_machinery":"The engine is the primitive of a Nevanlinna function. For a Nevanlinna $\\psi$, the function $\\Psi=-\\int \\psi\\,dz$ is univalent and maps $\\mathbb{C}_+$ onto a domain starlike at $-\\infty$, and Proposition 3.6 gives exactly when that domain contains a translate of $\\mathbb{C}_+$. The differential equation $\\partial F_t/\\partial t+\\varphi(F_t)=0$ is solved by $F_t=\\Psi(\\Phi(z)+t)$, so the whole transition semigroup is a conformal conjugate of horizontal translation. The condition that $\\Psi(\\mathbb{C}_+)$ contains $\\mathbb{C}_+$ is what makes $\\varphi=\\psi\\circ\\Phi$ satisfy the analytic-continuation requirement in the earlier Theorem 2.2.","core_discovery":"The central claim is Theorem 4.1. A free additive Lévy function of the second kind (FAL2), i.e. the Nevanlinna generator of a semigroup of analytic maps of $\\mathbb{C}_+$ arising from a free Lévy process with time-homogeneous transition probabilities, is either constant or has the form $\\varphi=\\psi\\circ\\Phi$. Here $\\Psi$ is a univalent function with inverse $\\Phi$, the derivative is $\\Psi'=-\\psi$ with $\\psi$ a Nevanlinna function, and $\\Psi(\\mathbb{C}_+)$ contains $\\mathbb{C}_+$. Conversely, any such data produce a FAL2 function, so the parametrisation is exhaustive for the stated class. This is called the nonlinear free Lévy-Khinchine formula.","pith_inferences":["The flow is a horizontal translation seen through a conformal lens, so the free Lévy process inherits a geometric picture that may connect to evolution families and slit-mapping constructions in univalent function theory beyond the starlike case.","If the parametrisation can be lifted from the point-mass initial distribution, the likely form is the same theorem with the image containing a translate of the upper half-plane rather than the upper half-plane itself; the dichotomy in the converse proof is the first thing to test.","The explicit description makes it possible to generate new examples by drawing starlike domains directly, which could feed into examples and stability questions for free Lévy processes."],"forward_implications":["Every FAL2 function is determined by a single univalent map, so checking whether a free Lévy process has homogeneous transitions reduces to checking that the map's image contains the upper half-plane.","The admissible data form a convex set of starlike domains, so the class of FAL2 functions is a nonlinear image of a convex set; choosing a Nevanlinna function with finite second moment and the right sign condition gives a concrete parametrisation.","For free multiplicative convolution on the unit circle, no nontrivial processes of this kind exist: all FUL2 functions are constant.","For free multiplicative convolution on the positive half-line, the same conformal-flow method gives a parametrisation by univalent maps on symmetric horizontal strips.","The representation gives a constructive way to produce free Lévy processes with time-homogeneous transitions, in addition to the existence criterion of Theorem 2.2."],"supporting_citations":[{"why":"Provides the earlier free Lévy-Khinchine representation for the first kind of free Lévy functions, the result this paper extends.","marker":"[2]"},{"why":"Defines FAL2 functions, supplies Theorem 2.2 with the differential equation and analytic-continuation criterion, and establishes the Markov/subordination framework used throughout.","marker":"[4]"},{"why":"Supplies the theory of starlike domains and univalent functions used to prove that primitives of Nevanlinna functions map the upper half-plane onto starlike domains.","marker":"[5]"},{"why":"Gives the free convolution of measures with unbounded support and the analytic transform that linearises free additive convolution, needed for the process construction.","marker":"[3]"},{"why":"Introduces the free additive convolution of non-commuting random variables on which the whole notion of free Lévy process rests.","marker":"[6]"}],"fun_headline_variants":["Nonlinear free Lévy-Khinchine via conformal pairs","Conformal maps parametrize homogeneous free Lévy processes","Explicit conformal recipe for free Lévy generators","All time-homogeneous free Lévy laws have ψ∘Φ form","Free Lévy second kind: explicit conformal parametrization"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof works only when the process starts at a single fixed point (the point mass $\\delta_0$), but the main theorem does not state that restriction; the classification of all FAL2 functions is therefore established only for that starting point.","fun_headline_variants_meta":{"raw":{"variants":["Nonlinear free Lévy-Khinchine via conformal pairs","Conformal maps parametrize homogeneous free Lévy processes","Explicit conformal recipe for free Lévy generators","All time-homogeneous free Lévy laws have ψ∘Φ form","Free Lévy second kind: explicit conformal parametrization"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001031,"raw_usage":{"total_tokens":4270,"prompt_tokens":799,"completion_tokens":3471,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":415,"completion_tokens_details":{"reasoning_tokens":3386}},"tokens_in":415,"tokens_out":3471,"duration_ms":27086,"temperature":1.0,"reasoning_tokens":3386,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:34:07.104870+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a FAL2 function $\\varphi$ whose associated extended domain $\\Omega=\\bigcup_{t\\geq0}(\\Phi(\\mathbb{C}_+)-t)$, with $\\Phi$ a primitive of $-1/\\varphi$, is neither the whole plane nor a translate of the upper half-plane. Such an example would break the dichotomy used in the converse direction of the proof and show the parametrisation is not exhaustive.","supporting_citations":[{"cited_title":"Bercovici, D","cited_arxiv_id":null,"evidence_quote":"Provides the earlier free Lévy-Khinchine representation for the first kind of free Lévy functions, the result this paper extends."},{"cited_title":"Biane, Processes with free increments,Math","cited_arxiv_id":null,"evidence_quote":"Defines FAL2 functions, supplies Theorem 2.2 with the differential equation and analytic-continuation criterion, and establishes the Markov/subordination framework used throughout."},{"cited_title":"Pommerenke , Univalent functions","cited_arxiv_id":null,"evidence_quote":"Supplies the theory of starlike domains and univalent functions used to prove that primitives of Nevanlinna functions map the upper half-plane onto starlike domains."},{"cited_title":"Bercovici, D","cited_arxiv_id":null,"evidence_quote":"Gives the free convolution of measures with unbounded support and the analytic transform that linearises free additive convolution, needed for the process construction."},{"cited_title":"Voiculescu, Addition of certain non-commuting random variables,Jour","cited_arxiv_id":null,"evidence_quote":"Introduces the free additive convolution of non-commuting random variables on which the whole notion of free Lévy process rests."}],"review_version":1}