Nonlinear free L\'evy-Khinchine formula and conformal mapping

Philippe Biane

REVIEW 2 major objections 4 minor 6 references

Nonlinear free L\'evy-Khinchine formula and conformal mapping

T0 review · 2 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read 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.

desk verdict Explicit parametrization of FAL2 functions, proved for μ0 = δ0 but stated more broadly; the math is sound but the scope claim needs correction. read the letter →

arxiv 1908.00755 v1 pith:QM6BSDIR submitted 2019-08-02 math.PR

Philippe Biane This is my paper · ORCID

classification math.PR MSC 46L5460G5130C3560E07

keywords freeprobabilityLévyprocessesNevanlinnafunctionsLévy-Khinchineformulaconformalmappingstarlikedomainsadditiveconvolutionsubordination

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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.

What carries the argument

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.

What would settle it

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.

Watch

Extended reading notes

Core claim

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.

Load-bearing premise

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.

Editorial extensions

If this is right

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.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

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.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the 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.

The 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.

The 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.

A 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.

The citation pattern is fair; the paper builds directly on [4] and the reliance there is legitimate. No self-citation inflation.

Who 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.

Referee Report

2 major / 4 minor

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.

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 (2)

[Theorem 4.1 and §2.3] 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.
[§4.2.2] 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.

minor comments (4)

[§4.1] 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.
[Theorems 5.2 and 5.5] 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.
[Proposition 3.3] 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.
[Throughout] 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.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Theorem 4.1 is a genuine conformal parametrization proved from the [4] characterization; the µ0=δ0 caveat is a scope issue, not a circular one.

full rationale

The paper's central result, Theorem 4.1, parametrizes FAL2 functions as φ = ψ∘Φ where Ψ′ = −ψ and Ψ(C+) contains C+. This is not equivalent to the definition of FAL2 by construction. The FAL2 class is recalled from the author's earlier [4] (Theorem 2.2) via the subordination/Markov characterization: φ is Nevanlinna with φ(ζ)/ζ→0 and φ∘F_t^{-1}∘F_{µ0}^{-1} admitting C+-valued analytic continuation. The forward direction in §4.2.1 starts from an arbitrary Nevanlinna ψ with Ψ(C+) containing C+, sets Φ=Ψ^{-1}, and verifies directly that φ=ψ∘Φ satisfies exactly the FAL2 conditions: the flow F_t = Ψ(Φ(z)+t) solves ∂_t F_t + φ(F_t)=0 and φ∘F_t^{-1}(z)=ψ(Φ(z)-t) is Nevanlinna; the asymptotic a=0 is derived from the geometry of Lemma 3.5, not assumed. The converse in §4.2.2 constructs Φ as a primitive of −1/φ, uses the FAL2 analytic-continuation hypothesis to extend Ψ to Ω=∪_{t≥0}(Ω0−t), and obtains the dichotomy Ω=C or a translate of C+; this is an honest parametrization argument. The only substantive caveat is that the proof is carried out for µ0=δ0 (stated in §2.3: 'we shall show that one can give a more explicit parameterisation these functions, at least in the case µ0 = δ0') while Theorem 4.1 is worded without that restriction; this is a scope gap and a correctness risk, not a circular reduction. The self-citation [4] supplies the definition and the foundational characterization theorem, but the new formula is proved from those inputs rather than being one of them; no fitted parameter is renamed as a prediction and no equation is equal to its input by definition.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The result is a pure mathematics theorem; no parameters are fitted to data. The construction uses a free parameter: an arbitrary Nevanlinna function ψ subject to the integral conditions of Proposition 3.6 (or equivalently its representation data α, β, ν), but these are the variables of the parametrization, not ad hoc constants. The proof relies on standard complex analysis (Nevanlinna representation, starlike domains) and on prior free-probability results, including the author's Theorem 2.2 from [4], which defines the class under study.

assumptions (5)

standard math Nevanlinna representation: any analytic function mapping C+ to C−∪R has the form αz+β+∫((1+uz)/(z−u))ν(du) for α≤0, β∈R, ν a finite positive measure
Invoked throughout, e.g., in §2.1.2 and in the proof of Proposition 3.6.
domain assumption Theorem 2.2 of [4]: a Nevanlinna function φ with φ(z)/z→0 is a free additive Lévy function of the second kind iff the semigroup solving ∂Ft/∂t+φ(Ft)=0 satisfies the analytic continuation condition for φ∘Ft^{-1}∘F_{µ0}^{-1}
Defines the class under study and is the criterion that the new parametrization characterizes; the proof of Theorem 4.1 verifies this criterion for µ0=δ0.
standard math Starlike-domain characterization: the image of C+ under the primitive of −ψ for a nonzero Nevanlinna function ψ is starlike at −∞, and conversely every such domain arises this way (Proposition 3.3, proof referenced to Pommerenke Ch. 2.2)
Used in §3.2-3.3 to convert the geometric condition 'Ψ(C+) contains a translate of C+' into integral conditions on ψ.
domain assumption Free convolution theory: the Voiculescu transform linearizes free additive convolution, and freely infinitely divisible measures have Voiculescu transforms of the form (14); subordination gives the Markov kernels (Theorems 2.1 and 2.2)
Background from [2] and [4] needed to connect analytic functions to Markov processes; these are prior published results, not derived in this paper.
standard math Analytic properties of Cauchy and Voiculescu transforms: existence of right inverses on domains Γα,β and Θα,β, and behavior at infinity
Used in §2.1.1 and Theorem 2.2; standard complex analysis for Cauchy transforms.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Nonlinear free L\'evy-Khinchine formula and conformal mapping." pith.science (2026). https://pith.science/paper/QM6BSDIR

@misc{pith2026190800755,
  author       = {Pith},
  title        = {Pith review of: Nonlinear free L\'evy-Khinchine formula and conformal mapping},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QM6BSDIR}},
  note         = {Machine review of arXiv:1908.00755}
}

read the original abstract

There are two natural notions of L\'evy processes in free probability: the first one has free increments with homogeneous distributions and the other has homogeneous transition probabilities. In the two cases one can associate a Nevanlinna function to a free L\'evy process. The Nevanlinna functions appearing in the first notion were characterised by Bercovici and Voiculescu. I give an explicit parametrisation for the Nevanlinna functions associated with the second kind of free L\'evy processes. This gives a nonlinear free L\'evy-Khinchine formula.

Figures

Figures reproduced from arXiv: 1908.00755 by the authors.

**Figure 2.** shows ψ(z) = −z 1/2 where α = 0 and ν(du) = √ −udu 2π(1+u2) 1u<0, with β + R u ν(du) = −∞. One has Ψ(z) = 2 3 z 3/2 and the image Ψ(C+) is a 3/4 plane, which contains the upper half-plane [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** The image of Ψ(z) = −2z 1/2 4. Free Lévy processes with homogeneous transition probabilities 4.1. Some preliminary computations. Let us recall that, by Theorem 2.2 we are trying to characterise Nevanlinna functions ϕ such that (i) ϕ(ζ) ζ → ζ→∞ ζ∈Γα,β 0 in every domain of the form Γα,β (ii) For any t ≥ 0, ϕ ◦ F −1 t has an analytic continuation to C+, with values in C−. Here Ft(z), for t ≥ 0 is the semi-group of anal… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

[4]

Biane, Processes with free increments,Math

P. Biane, Processes with free increments,Math. Z. 227, (1998), 143–174

work page 1998
[1]

Bercovici, V

H. Bercovici, V. Pata, Stable laws and domains of attraction in free probability theory,Annals of Mathe- matics, 149 (1999), 1023–1060

work page 1999
[2]

Bercovici, D

H. Bercovici, D. Voiculescu , Lévy-Hin˘ cin type theorems for multiplicative and additive free convolution, Paciﬁc J. Math.153 (1992), 217–248

work page 1992
[3]

Bercovici, D

H. Bercovici, D. Voiculescu , Free convolution of measures with unbounded support,Indiana University Mathematics Journal 42 (1993), 733–773

work page 1993
[5]

Pommerenke , Univalent functions

C. Pommerenke , Univalent functions. With a chapter on quadratic diﬀerentials by Gerd Jensen.Studia Mathematica/Mathematische Lehrbücher, Band XXV. Vandenhoeck & Ruprecht, Göttingen, 1975

work page 1975
[6]

Voiculescu, Addition of certain non-commuting random variables,Jour

D. Voiculescu, Addition of certain non-commuting random variables,Jour. Funct. Anal.66 (1986), 323– 346. Institut Gaspard-Monge, université Paris-Est Marne-la-V allée, 5 Boulev ard Descartes, Champs- sur-Marne, 77454, Marne-la-V allée cedex 2, France

work page 1986

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [4]

Biane, Processes with free increments,Math

P. Biane, Processes with free increments,Math. Z. 227, (1998), 143–174

work page 1998

[2] [1]

Bercovici, V

H. Bercovici, V. Pata, Stable laws and domains of attraction in free probability theory,Annals of Mathe- matics, 149 (1999), 1023–1060

work page 1999

[3] [2]

Bercovici, D

H. Bercovici, D. Voiculescu , Lévy-Hin˘ cin type theorems for multiplicative and additive free convolution, Paciﬁc J. Math.153 (1992), 217–248

work page 1992

[4] [3]

Bercovici, D

H. Bercovici, D. Voiculescu , Free convolution of measures with unbounded support,Indiana University Mathematics Journal 42 (1993), 733–773

work page 1993

[5] [5]

Pommerenke , Univalent functions

C. Pommerenke , Univalent functions. With a chapter on quadratic diﬀerentials by Gerd Jensen.Studia Mathematica/Mathematische Lehrbücher, Band XXV. Vandenhoeck & Ruprecht, Göttingen, 1975

work page 1975

[6] [6]

Voiculescu, Addition of certain non-commuting random variables,Jour

D. Voiculescu, Addition of certain non-commuting random variables,Jour. Funct. Anal.66 (1986), 323– 346. Institut Gaspard-Monge, université Paris-Est Marne-la-V allée, 5 Boulev ard Descartes, Champs- sur-Marne, 77454, Marne-la-V allée cedex 2, France

work page 1986