Boundary measures of holomorphic functions on the imaginary domain

Shigeru Yamagami · 2025

Read Boundary measures of holomorphic functions on the imaginary domain by Shigeru Yamagami (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-09093.txt.

Introduction

Shigeru Yamagami defines boundary measures for holomorphic functions on the imaginary domain in a 2025 arXiv paper. The work ties to the Herglotz-Nevanlinna integral form for so-called Pick functions. Yamagami sets out what a boundary measure is in this frame. Basic traits of the new notion get spelled out. The note is short and sets ground for later use of the idea.

Catalog record

CREATORShigeru Yamagami
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-09093

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Boundary measures of holomorphic functions on the imaginary domain Shigeru Yamagami Abstract . In connection with the Herglotz-Nevanlinna integral representation of so-called Pick functions, we introduce the notion of boundary measure of holomorphic functions on the imaginary domain and elucidate some of basic properties. 1991 Mathematics Subject Classification MSC2020: 46G99, 31A10 Graduate School of Mathematics Nagoya University Nagoya, 464-8602, JAPAN Introduction In our previous paper [ 4 ] , we investigated supplementary results to the classical integral representation theorem of Herglotz, Riesz and Nevanlinna on holomorphic endofunctions of the upper half plane. As an integral transformation, it is closely related to the Cauchy-Stieltjes transform and we know a satisfactory characterization of the transform of a positive measure as a holomorphic function on the upper half plane. Although the transform itself is applicable to complex measures, its description as a holomorphic function seems somewhat awkward. We have however a pragmatic characterization in terms of existence of boundary measures if its support is not full ( [ 9 ] ), which was one of motivations in [ 4 ] . Here we shall extend it to boundary measures of any support under the assumption of some estimates on the boundary behaviour of holomorphic functions, which is in fact necessary for the existence of boundary measure. To deal with the case of full support, we need to work on several (at least two) local coordinates to describe boundary limits. The main problem there is the compatibility of locally defined boundary measures and we here impose localized boundary estimates to check the compatibility of measures arising from local coordinates. We then restrict ourselves to the most simple (and seemingly most interesting) choice of local coordinates, i.e., the original one z and its inversion \widetilde{z}=-1/z , and show that the associated local measures are related to each other under the inversion and the joined measure on the whole boundary turns out to fulfill the integral representation. During its formulation, it turns out that the assumption on boundary behavior enables us to capture boundary limits as distributions of mild singularity (such as Cauchy’s principal value) and the problem of coordinate transformation is soloved first within boundary distributions. Boundary measures are then worked out as a result of relevant facts on boundary distributions. The author is grateful to M. Nagisa and M. Uchiyama for fruitful conversations, which were big impetuses to the present work. 1. Representing Measures Let \overline{\text{\ym C}}=\text{\ym C}\sqcup\{\infty\} be the Riemann sphere and \overline{\text{\ym R}}=\text{\ym R}\sqcup\{\infty\} be the extended real line in \overline{\text{\ym C}} . For a holomorphic function \varphi(w) of w\in\text{\ym C}\setminus\text{\ym R} , a complex Radon measure \lambda on \overline{\text{\ym R}} is called the representing measure of \varphi if it satisfies \varphi(w)=\frac{\varphi(i)+\varphi(-i)}{2}+\int_{\overline{\text{\ym R}}}\frac{1+sw}{s-w}\,\lambda(ds)\quad(w\in\text{\ym C}\setminus\text{\ym R}). The representing measure, if any exists, is unique because \{(1+sw)/(s-w);w\in\text{\ym C}\setminus\text{\ym R}\} is total in the space C(\overline{\text{\ym R}}) of continuous functions on \overline{\text{\ym R}} . Related to the representing measure, given a holomorphic function \varphi on \text{\ym C}\setminus(\text{\ym R}\sqcup K) with K a compact subset of \text{\ym C}\setminus\text{\ym R} , the boundary support [\varphi] of \varphi is defined to be the complement in \overline{\text{\ym R}} of an open subset \{c\in\overline{\text{\ym R}};\text{$\varphi(z)$ is continuously extended to a neighborhood of $c$ in $\overline{\text{\ym C}}$}\}. Note that [\varphi] is a support in the sense of hyperfunction, which turns out to be the support of the representing measure. The existence of representing measure is a strong condition on holomorphic functions. Example 1.1 . A holomorphic function \varphi on \text{\ym C}\setminus\text{\ym R} is extended to to an entire function on ℂ if and only if [\varphi]=\{\infty\} . As for an entire function, the representing measure exists if and only if it is an affine function. By the holomorphic change-of-variable z=\frac{i-w}{i+w},\quad w=i\frac{1-z}{1+z} of \overline{\text{\ym C}} , the upper half plane \text{\ym C}_{+}=\{w\in\text{\ym C};\text{Im}\,w>0\} is bijectively mapped onto the unit disk D=\{z\in\text{\ym C};|z|<1\} and, under the correspondence \phi(z)=-i\varphi(w) , we have the disk version of representing measure: Let \text{\ym T}=\{z\in\text{\ym C};|z|=1\} be the boundary of D . A holomorphic function \phi on \overline{\text{\ym C}}\setminus\text{\ym T} admits the representing measure if there exists a complex Radon measure \mu on 𝕋 satisfying \phi(z)=\frac{\phi(0)+\phi(\infty)}{2}+\frac{1}{2\pi}\int_{\text{\ym T}}\frac{\zeta+z}{\zeta-z}\,\mu(d\zeta). In view of the Jordan decomposition on signed measures and the Herglotz-Riesz integral representation, functions having representing measures are exactly linear combinations of positive ones (so-called Pick functions on \text{\ym C}_{+} ). To make the relevant structure clear, we introduce a *-operation on the space \mathscr{R} of holomorphic functions on \text{\ym C}\setminus\text{\ym R} by \varphi^{*}(w)=\overline{\varphi(\overline{w})} and a *-operation 1 1 1 A minus sign is put here to relate the real part of \phi to the imaginary part of \varphi . on the space \mathscr{T} of holomorphic functions on \overline{\text{\ym C}}\setminus\text{\ym T} by \phi^{*}(z)=-\overline{\phi(1/\overline{z})} . Then \mathscr{R} and \mathscr{T} are *-isomorphic by the correspondence \varphi(w)=i\phi(z) so that reality \overline{\varphi(w)}=\varphi(\overline{w}) ( w\in\text{\ym C}\setminus\text{\ym R} ) is equivalent to \varphi^{*}=\varphi for \varphi\in\mathscr{R} , while imaginarity \overline{\phi(z)}=-\phi(1/\overline{z}) ( z\in\overline{\text{\ym C}}\setminus\text{\ym T} ) is equivalent to \phi^{*}=\phi for \phi\in\mathscr{T} . The *-operations behave well with respect to (the existence of) representing measures: In the case of the extended real boundary \overline{\text{\ym R}} , the representing measure of \varphi^{*} is given by the complex conjugate \overline{\lambda} of \lambda as \varphi^{*}(w)=\frac{\varphi^{*}(i)+\varphi^{*}(-i)}{2}+\int_{\overline{\text{\ym R}}}\frac{1+sw}{s-w}\,\overline{\lambda}(ds), and, in the case of the circle boundary 𝕋 , as \phi^{*}(z)=\frac{\phi^{*}(0)+\phi^{*}(\infty)}{2}+\frac{1}{2\pi}\int_{\text{\ym T}}\frac{\zeta+z}{\zeta-z}\,\overline{\mu}(d\zeta). Thus the existence of representing measure is reduced to the case of signed measures by extracting (\varphi+\varphi^{*})/2 and (\varphi-\varphi^{*})/2i (resp. (\phi+\phi^{*})/2 and (\phi-\phi^{*})/2i ) with the corresponding signed measures (\lambda+\overline{\lambda})/2 and (\lambda-\overline{\lambda})/2i (resp. (\mu+\overline{\mu})/2 and (\mu-\overline{\mu})/2i ). Recall here that the Poisson integral formula of disk version \phi(z)=i\text{Im}\,\phi(0)+\frac{1}{2\pi}\oint_{0}^{2\pi}\frac{e^{it}+z}{e^{it}-z}\,\text{Re}\,\phi(e^{it})\,dt holds for |z|<1 if a holomorphic function \phi on D is continuously extended to \overline{D}=D\sqcup\text{\ym T} . Then the imaginary extension of \phi , i.e., \phi(z)=-\overline{\phi(1/\overline{z})} for z\in\overline{\text{\ym C}}\setminus\overline{D} , satisfies \phi^{*}(z)=\phi(z) ( z\in\overline{\text{\ym C}}\setminus\text{\ym T} ) and i\text{Im}\,\phi(0)=\frac{\phi(0)-\overline{\phi(0)}}{2}=\frac{\phi(0)+\phi^{*}(\infty)}{2}=\frac{\phi(0)+\phi(\infty)}{2}. Moreover, if z\in\overline{\text{\ym C}}\setminus\overline{D} , we have \phi(z)=\phi^{*}(z)=-\overline{\phi(1/\overline{z})}=i\text{Im}\,\phi(0)+\frac{1}{2\pi}\oint_{0}^{2\pi}\frac{e^{it}+z}{e^{it}-z}\,\text{Re}\,\phi(e^{it})\,dt. Thus, for the imaginary extension of \phi in this case, the representing measure exists and takes the form \displaystyle\mu(dt)=\text{Re}\,\phi(e^{it})\,dt=\lim_{r\uparrow 1}\text{Re}\,\phi(re^{it})\,dt relative to the parametrization \zeta=e^{it} of 𝕋 . Notice here that, the boundary values exist both for \lim_{r\uparrow 1}\phi(re^{i\theta}) and \lim_{r\downarrow 1}\phi(re^{i\theta}) but their difference amounts to \lim_{r\uparrow 1}\phi(re^{i\theta})-\lim_{r\downarrow 1}\phi(re^{i\theta})=\lim_{r\uparrow 1}\phi(re^{i\theta})+\lim_{r\downarrow 1}\overline{\phi(e^{i\theta}/r)}=2\text{Re}\,\phi(e^{i\theta}). Returning to the initial situation without boundary continuity, for the existence problem of representing measure \mu of \phi=\phi^{*} , we may further assume that \text{Im}\,\phi(0)=0\iff\phi(0)+\phi(\infty)=0 without modifying \mu by adjusting imaginary additive constants. In that case (i.e., \phi=\phi^{*} and \text{Im}\,\phi(0)=0 ), if a representing measure \mu exists, its Jordan decomposition \mu=\mu_{+}-\mu_{-} with |\mu|=\mu_{+}+\mu_{-} and \mu_{+}\perp\mu_{-} gives rise to the decomposition \phi=\phi_{+}-\phi_{-} , where \phi_{\pm}(z)=\frac{1}{2\pi}\oint_{0}^{2\pi}\frac{\zeta+z}{\zeta-z}\,\mu_{\pm}(dt) are holomorphic functions of z\in\overline{\text{\ym C}}\setminus\text{\ym T} satisfying \phi_{\pm}^{*}=\phi_{\pm} , \text{Im}\,\phi_{\pm}(0)=0 and \text{Re}\,\phi(z)\geq 0 ( |z|<1 ) (i.e., \text{Re}\,\phi(z)\leq 0 for |z|>1 ). Moreover, the representing measure \mu_{\pm} of \phi_{\pm} is expressed by \oint f(t)\,\mu_{\pm}(dt)=\lim_{r\uparrow 1}\oint f(t)\text{Re}\,\phi_{\pm}(re^{it})\,dt for each f\in C(\text{\ym T}) (Herglotz-Riesz-Nevanlinna) and hence the signed measure on 𝕋 \mu_{r}(dt)=\Bigl(\text{Re}\,\phi_{+}(re^{it})-\text{Re}\,\phi_{-}(re^{it})\Bigr)\,dt=\frac{1}{2}\Bigl(\phi(re^{it})-\phi(r^{-1}e^{it})\Bigr)\,dt parametrized by 0<r<1 converges weakly to \mu as r\uparrow 1 . 2. Boundary Limits by Circles Now reverse the process and suppose that the signed measure \mu_{r} on 𝕋 associated with \phi=\phi^{*} converges as a linear functional on C(\text{\ym T}) to a signed Radon measure \mu when r\uparrow 1 . Then \mu is the representing measure of \phi ; letting

Read the complete text