Integrals of the difference of subharmonic functions over discs and planar small sets

B. N. Khabibullin

arXiv: 2012.06334v2 · PDF · 3 January 2021
Categories: math.CV
Licence: CC BY 4.0 The site shows the text.

Abstract

The maximum of the modulus of a meromorphic function cannot be restricted from above by the Nevanlinna characteristic of this meromorphic function. But integrals from the logarithm of the module of a meromorphic function allow similar restrictions from above. This is illustrated by one of the important theorems of Rolf Nevanlinna in the classical monograph by A.A. Goldberg and I.V. Ostrovskii on meromorphic functions, as well as by the Edrei-Fuchs Lemma on small arcs and its versions for small intervals in articles by A.F. Grishin, M.L. Sodin, T.I. Malyutina. Similar results for integrals of differences of subharmonic functions even with weights were recently obtained by B.N. Khabiblullin, L.A. Gabdrakhmanova. All these results are on integrals over subsets on a ray. In this article, we establish such results for integrals of the logarithm of the modulus of a meromorphic function and the difference of subharmonic functions over discs and planar small sets. Our estimates are uniform in the sense that the constants in these estimates are explicitly written out and do not depend on meromorphic functions and the difference of subharmonic functions provided that these functions has an integral normalization near zero.

Items 15

Theorem

Theorem 1

BWPN

Let 0< r_0< r<+\infty, 1<k\in {\mathbb{R}}^+, E\subset [0,r] be {\rm mes}-measurable, g\in L^p(E), where 1<p\leq \infty and q\in [1,+\infty) is defined by \dfrac{1}{p}+\dfrac{1}{q}=1. If U\not\equiv \pm\infty is a non-trivial \delta-subharmonic functions on {\mathbb{C}}, and u\not\equiv -\infty is a subharmonic function on {\mathbb{C}}, then

\begin{flalign} \int_{E} {\sf M}_{U^+}(t)g(t){\,{\rm d}} t&\leq \frac{4qk}{k-1} \bigl({\sf T}_{U}(r_0,kr)+{\sf C}_{U^+}(r_0)\bigr)\|g\|_{L^p(E)} \sqrt[q]{{\rm mes} \,E} \ln\frac{4kr}{{\rm mes} \,E}, \tag{\ref{1l}T} \\ \int_{E} {\sf M}_{|u|}(t)g(t){\,{\rm d}} t& \leq \frac{5qk}{k-1} \bigl({\sf M}_{u^+}(kr)+{\sf C}_{u^-}(r_0)\bigr) \sqrt[q]{{\rm mes} \,E} \ln\frac{4kr}{{\rm mes} \,E}. \tag{\ref{1l}M} \end{flalign}

Used by Corollary 1, Proof

Corollary

Corollary 1

IY1H

Let, under the conditions of Theorem 1, F\not\equiv 0 be a meromorphic function, and f\not\equiv 0 be an entire function. Then, in the traditional notation (?)–(?), we have

\begin{flalign} \int_{E} \bigl(\ln^+ M(t,F)\bigr)g(t){\,{\rm d}} t&\leq \frac{4qk}{k-1} \bigl( T(kr,F)-N(r_0,F)\bigr)\|g\|_{L^p(E)} \sqrt[q]{{\rm mes}\,E} \ln\frac{4kr}{{\rm mes} \,E}, \tag{\ref{1l}T} \\ \int_{E} \bigl|\ln M(t,f)\bigr| g(t){\,{\rm d}} t& \leq \frac{5qk}{k-1} \bigl(\ln^+ M(kr,f)+m(r_0,1/f ) \bigr) \sqrt[q]{{\rm mes} \,E} \ln\frac{4kr}{{\rm mes} \,E}. \tag{\ref{1l}M} \end{flalign}

Uses Theorem 1

Proof of Corollary 1

By (?)–(?) we have for U:=\ln |F|

{\sf T}_{U}(r_0,kr)+{\sf C}_{U^+}(r_0)\overset{\eqref{{cs}T},\eqref{{cs}m}}{=}T(kr,F)-T(r_0,F)+ m(r_0,f)\overset{\eqref{{TN}T}}{=}T(kr,F)-N(r_0,F).

We can replace the bracket on the right-hand side of (1) with the right side of these equalities. Then we obtain (1) by (?) for the integrand in the left-hand side of (1).

For u:=\ln |f|, we have

{\sf M}_{u^+}(kr)+{\sf C}_{u^-}(r_0)\overset{\eqref{{cs}M}}{=}\ln^+ M(kr,f)+ {\sf C}_{\ln^-|f|}(r_0)\overset{\eqref{{cs}m}}{=}\ln^+ M(kr,f)+m(r_0,1/f ).

We can replace the bracket on the right-hand side of (1) with the right side of these equalities. Then we obtain (1) by (?) for the integrand in the left-hand side of (1).

Uses Theorem 1

Remark

Remark 1

JR6Z

Our elementary example [KhaMst, 1.1, (3)] shows that it is impossible to discard the terms -N(r_0,F) and m(r_0,1/f ) in parentheses on the right-hand sides of inequality (1) and (1), respectively. In particular, the classical Rolf Nevanlinna Theorem [GOe, Ch. 1, Theorem 7.2] can be formulated in the following correct form: for each meromorphic function F\not\equiv 0 and 1<k\in {\mathbb{R}}^+

\frac{1}{r}\int_0^r \ln^+ M(t,F){\,{\rm d}} t\leq \frac{4k\ln 4k}{k-1} \bigl( T(kr,F)-N(r_0,F)\bigr)\quad \text{\it for all\/ $0<r_0\leq r\in {\mathbb{R}}^+$.}

Indeed, it is sufficient to choose E:=[0,r], g\equiv 1 with {\rm mes} \,E=r, and p:=\infty with q=1 in (1).

Theorem

Theorem 2

63UT

Let 0< r_0< r<+\infty, 1<k\in {\mathbb{R}}^+, E\subset \overline D(r) be a \uplambda-measurable subset. If U\not\equiv \pm\infty is a \delta-subharmonic functions on {\mathbb{C}}, and u\not\equiv -\infty is a subharmonic function on {\mathbb{C}}, then

\begin{flalign} \int_{E} U^+{\,{\rm d}} \uplambda \leq \frac{2k}{k-1} \bigl({\sf T}_{U}(r_0,kr)+{\sf C}_{U^+}(r_0)\bigr)\uplambda(E) \ln\frac{100kr^2}{\uplambda (E)}, \tag{\ref{1}T} \\ \int_{E} |u|{\,{\rm d}} \uplambda \leq \frac{3k}{k-1} \bigl({\sf M}_{u^+}(kr)+{\sf C}_{u^-}(r_0)\bigr) \uplambda(E) \ln\frac{100kr^2}{\uplambda (E)}. \tag{\ref{1}M} \end{flalign}

Used by Proof, Corollary 3, Proof

Proof (Proof of Theorem 2) of Theorem 2

We can assume that U=u-v is the canonical representation of U. Consider a number b>0 such that (1+b)^2=k. By Lemma 3, we have

\begin{gathered} \int_{E}U^+{\,{\rm d}} \uplambda \leq \frac12\frac{\sqrt{k}+1}{\sqrt{k}-1} \Bigl( {\sf C}_{U^+}(\sqrt{k}r) +{\sf N}_{\varDelta_v}(\sqrt{k}r, kr)\Bigr) \uplambda(E)\ln\frac{\bigl(10\sqrt{k}r\bigr)^2}{\uplambda(E)} \\ \leq \frac{k}{k-1} \Bigl( \underset{{\sf T}_{U}(r_0, \sqrt{k}r)}{\underbrace{{\sf C}_{U^+}(r_0, \sqrt{k}r)+{\sf N}_{\varDelta_v}(r_0, \sqrt{k}r)}}+{\sf C}_{U^+}(r_0) +\underset{{\sf T}_{U}(r_0, kr)}{\underbrace{{\sf C}_{U^+}(r_0, kr)+{\sf N}_{\varDelta_v}(r_0, kr)}}\Bigr)\uplambda(E)\ln\frac{\bigl(10r\bigr)^2k}{\uplambda(E)}\\ \leq \frac{k}{k-1} \Bigl( 2{\sf T}_{U}(r_0, kr)+{\sf C}_{U^+}(r_0)\Bigr)\uplambda(E)\ln\frac{\bigl(10r\bigr)^2k}{\uplambda(E)} \leq \frac{2k}{k-1} \Bigl( {\sf T}_{U}(r_0, kr)+{\sf C}_{U^+}(r_0)\Bigr)\uplambda(E)\ln\frac{\bigl(100kr^2}{\uplambda(E)} \end{gathered}

and obtain (2).

If u\in {\sf sbh}_*({\mathbb{C}}), then the function {\sf M}_u^+ is increasing, and

\int_{E} u^+{\,{\rm d}} \uplambda \overset{\eqref{{Cu}M}}{\leq} {\sf M}_{u^+}(r) \uplambda (E) \quad \text{for $E\subset \overline D(r)$.}

For U_u:=0-u, the difference 0-u is the canonical representation of \delta-subharmonic non-trivial function U_u and we have

{\sf T}_{U_u}(r,R)\overset{\eqref{T}}{=} {\sf C}_{\sup\{0,u\}}(r,R)={\sf C}_{u^+}(r,R)\leq {\sf C}_{u^+}(R) \leq {\sf M}_{u^+}(R).

Hence, by Theorem 2 in part (2) for U_u in the role of U, we obtain

\begin{gathered} \int_{E} (-u)^+{\,{\rm d}} \uplambda \overset{\eqref{inDl+}}{\leq} \frac{2k}{k-1} \Bigl({\sf T}_{U_u}(r_0,kr)+{\sf C}_{U_u^+}(r_0)\Bigr) \uplambda(E) \ln\frac{100kr^2}{\uplambda (E)} \\ \overset{\eqref{TCM}}{\leq} \frac{2k}{k-1} \Bigl({\sf M}_{u^+}(kr)+{\sf C}_{(-u)^+}(r_0)\Bigr)\uplambda(E) \ln\frac{100kr^2}{\uplambda (E)}. \end{gathered}

The latter together with (Proof) gives (2).

Uses Lemma 3

Corollary

Corollary 2

LUKW

If U\not\equiv \pm\infty be a \delta-subharmonic functions on {\mathbb{C}}, and u\not\equiv -\infty be a subharmonic function on {\mathbb{C}}, then

\begin{flalign} {\sf B}_{U^+}(r) &\leq \frac{7k\ln (ek)}{k-1} \bigl({\sf T}_{U}(r_0,kr)+{\sf C}_{U^+}(r_0)\bigr) \quad\text{for all\/ $0<r_0<r\in \mathbb R^+$ and $1<k\in {\mathbb{R}}^+$}, \tag{\ref{10}T} \\ {\sf B}_{|u|}(r) &\leq \frac{11k\ln (ek)}{k-1} \bigl({\sf M}_{u^+}(kr)+{\sf C}_{u^-}(r_0)\bigr) \quad\text{for all\/ $0<r_0<r\in \mathbb R^+$ and $1<k\in {\mathbb{R}}^+$}. \tag{\ref{10}M} \end{flalign}

Used by Proof

Proof of Corollary 2

Let E:=\overline D(r) in (2). Then \uplambda(E)=\pi r^2,

\ln\frac{100kr^2}{\uplambda (E)}= \ln\frac{100k}{\pi}\leq \frac{7}{2}\ln ek, \quad {\sf B}_{U^+}(r)\overset{\eqref{Bu}}{=}\frac{1}{\uplambda(E)}\int_{E} U^+{\,{\rm d}} \uplambda, \quad {\sf B}_{|u|}(r) \overset{\eqref{Bu}}{=}\frac{1}{\uplambda(E)}\int_{E} |u|{\,{\rm d}} \uplambda,

and by (2) we obtain (2).

Uses Theorem 2

Corollary

Corollary 3

3NXX

Let, under the conditions of Theorem 2, F\not\equiv 0 be a meromorphic function, and f\not\equiv 0 be an entire function. Then, in the traditional notation (?)–(?) and (?), we have

\begin{flalign} \int_{E} \ln^+ \bigl|F(z)\bigr|{\,{\rm d}} \uplambda(z)&\leq \frac{2k}{k-1} \bigl( T(kr,F)-N(r_0,F)\bigr) \uplambda(E) \ln\frac{100kr^2}{\uplambda (E)}, \tag{\ref{1lfF}T} \\ \int_{E} \Bigl|\ln \bigl|f(z)\bigr|\Bigr| {\,{\rm d}} \uplambda(z)& \leq \frac{3k}{k-1} \bigl(\ln^+ M(kr,f)+m(r_0,1/f ) \bigr) \ln\frac{100kr^2}{\uplambda (E)}. \tag{\ref{1lfF}M} \\ {m^{[2]}}(r,F)&\leq \frac{7k\ln (ek)}{k-1}\bigl( T(kr,F)-N(r_0,F)\bigr), \tag{\ref{1lfF}F} \\ {m^{[2]}}(r,f)+{m^{[2]}}(r,1/f) &\leq \frac{11k\ln (ek)}{k-1} \bigl({\sf M}_{u^+}(kr)+{\sf C}_{u^-}(r_0)\bigr) \tag{\ref{1lfF}f} \end{flalign}

for all 0<r_0<r\in \mathbb R^+ and 1<k\in {\mathbb{R}}^+.

Uses Theorem 2

Proof of Corollary 3

For U:=\ln |F|, we obtain (3) by (2), (Proof), (?), and also (3) by (2), (Proof), (?).

For u:=\ln |f| we obtain (3) by (2), (Proof), (?), and also (3) by (2), (Proof), (?) since {m^{[2]}}(r,f)+{m^{[2]}}(r,1/f)\overset{\eqref{BF}}{=}{\sf B}_{|\ln |f||}(r) for all r\in {\mathbb{R}}^+.

Uses Theorem 2, Proof, Corollary 2

Lemma

Lemma 1

IXN2

Let 0\leq r<R<+\infty, E\subset \overline D(r) be \uplambda-measurable, U=u-v be a difference of subharmonic functions u, v\in {\sf sbh}_* \bigl(\overline D(R)\bigr), \varDelta_v be the Riesz measure of v. Then

\int_{E} U^+{\,{\rm d}} \uplambda\leq \frac{1}{2}\Bigl(\frac{R+r}{R-r} {\sf C}_{U^+}(R)+\varDelta_v^{{\text{\tiny\rm rad}}}(R)\Bigr) \uplambda(E)\ln\frac{(10R)^2}{\uplambda(E)}.

Used by Proof

Proof of Lemma 1

For w\in E\subset \overline D(r), by the Poisson – Jensen formula [Rans, 4.5], we have

\begin{gathered} U(w)=\frac{1}{2\pi}\int_0^{2\pi}U(Re^{i\varphi}){\rm Re} \frac{Re^{i\varphi}+w}{Re^{i\varphi}-w}{\,{\rm d}} \varphi -\int_{D(R)}\ln \Bigl|\frac{R^2-z\bar w}{R(w-z)}\Bigr|{\,{\rm d}} \varDelta_u(z) \\ +\int_{D(R)}\ln \Bigl|\frac{R^2-z\bar w}{R(w-z)}\Bigr|{\,{\rm d}} \varDelta_v(z) \leq \frac{R+r}{R-r}{\sf C}_{U^+}(R) +\int_{D(R)}\ln \frac{2R}{|w-z|}{\,{\rm d}} \varDelta_v(z) \end{gathered}

where the right-hand side of the inequality is positive. Hence, by integrating, we get

\begin{gathered} \int_{E} U^+{\,{\rm d}} \uplambda \leq \frac{R+r}{R-r}{\sf C}_{U^+}(R)\uplambda(E)+ \int_{D(R)}\int_E\ln\frac{2R}{|w-z|}{\,{\rm d}} \uplambda (w){\,{\rm d}} \varDelta_v(z)\\ \leq \frac{R+r}{R-r}{\sf C}_{U^+}(R)\uplambda(E)+ \varDelta_v\bigl(\overline D(R)\bigr) \int_E\ln\frac{2R}{|w-z|}{\,{\rm d}} \uplambda (w). \end{gathered}

Denote by \uplambda_E the restriction of the Lebesgue measure \uplambda to \uplambda-measurable set E\subset \overline D(r). Obviously,

\begin{align} {\rm supp\,} \uplambda_E&\subset \overline D(r)\subset D(R), \tag{\ref{lE1}s}\label{{lE}s}\\ \uplambda_E(w,t)&\overset{\eqref{{murad}z}}{\leq} \pi t^2\quad\text{for each $w\in {\mathbb{C}}$ and $t\in {\mathbb{R}}^+$}, \tag{\ref{lE1}t}\label{{lE}t}\\ \uplambda_E ({\mathbb{C}})&=\uplambda (E)\leq \pi r^2\leq \pi R^2. \tag{\ref{lE1}E}\label{{lE}E} \end{align}

Consider the last integral in (Proof):

\int_E\ln\frac{2R}{|w-z|}{\,{\rm d}} \uplambda (w) =\int_{{\mathbb{C}}} \ln\frac{2R}{|w-z|}{\,{\rm d}} \uplambda_E (w) \overset{\eqref{{lE}s}}{=}\int_0^{2R} \ln \frac{2R}{t}{\,{\rm d}} \uplambda_E (z;t) \overset{\eqref{{lE}t}}{=} \int_0^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t.

By (?) we have \sqrt{\uplambda (E)}\overset{\eqref{{lE}E}}{\leq} \sqrt{\pi}r< 2r \leq 2R. From here we can split the last integral in (Proof) into the sum of two positive integrals:

\int_0^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t= \int_0^{\sqrt{\uplambda (E)}} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t+\int_{\sqrt{\uplambda (E)}}^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t

Using (?), we have for first integral on the right-hand side of this equality (Proof) the estimate

\int_0^{\sqrt{\uplambda (E)}} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t\overset{\eqref{{lE}t}}{\leq} \int_0^{\sqrt{\uplambda (E)}} \frac{\pi t^2}{t}{\,{\rm d}} t=\frac{\pi}{2}\uplambda (E).

Using (?), we have for second integral on the right-hand side of (Proof) the estimate

\int_{\sqrt{\uplambda (E)}}^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t\leq \uplambda({\mathbb{C}})\int_{\sqrt{\uplambda (E)}}^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t \overset{\eqref{{lE}E}}{=} \frac12\uplambda(E) \ln \frac{4R^2}{\uplambda (E)}.

Thus, it follows from (Proof)–(Proof) and (Proof) that

\int_0^{2R} \frac{\uplambda_E (z;t)}{t}{\,{\rm d}} t\leq \frac{\pi}{2}\uplambda (E)+\frac12\uplambda(E) \ln \frac{4R^2}{\uplambda (E)} = \frac{1}{2}\uplambda (E)\ln \frac{4e^{\pi}R^2}{\uplambda(E)}\leq \frac{1}{2}\uplambda (E)\ln \frac{(10R)^2}{\uplambda(E)},

where

\ln \frac{(10R)^2}{\uplambda(E)}\geq 2, \text{ since } \uplambda(E)\overset{\eqref{{lE}E}}{\leq} \pi R^2,

and estimate (Proof) together with (Proof) and (Proof) gives (1).

Lemma

Lemma 2

UYXQ

Let \mu\in {\sf Meas\,}^+\bigl(\overline D(R)\bigr). Then

\mu^{{\text{\tiny\rm rad}}}(r)\leq \frac{R}{R-r}{\sf N}_{\mu}(r,R)\quad \text{for each $0\leq r\leq R$.}

Used by Proof

Lemma

Lemma 3

ERHQ

Let 0< r<+\infty, 0<b\in {\mathbb{R}}^+, E\subset \overline D(r) be \uplambda-measurable, U=u-v be a difference of subharmonic functions u, v\in {\sf sbh}_* \Bigl(\overline D\bigl((1+b)^2r\bigr)\Bigr). Then

\int_{E} U^+{\,{\rm d}} \uplambda \leq \frac{2+b}{2b} \Bigl( {\sf C}_{U^+}\bigl((1+b)r\bigr) +{\sf N}_{\varDelta_v}\bigl((1+b)r, (1+b)^2r\bigr)\Bigr) \uplambda(E)\ln\frac{\bigl(10(1+b)r\bigr)^2}{\uplambda(E)}.

Used by Proof

Proof (Proof) of Lemma 3

By Lemma 1 with R:=(1+b)r we have

\int_{E} U^+{\,{\rm d}} \uplambda \leq \frac{1}{2}\Bigl(\frac{2+b}{b} {\sf C}_{U^+}(R)+\varDelta_v^{{\text{\tiny\rm rad}}}\bigl((1+b)r\bigr)\Bigr) \uplambda(E)\ln\frac{\bigl(10(1+b)r\bigr)^2}{\uplambda(E)}.

Hence, by Lemma 2 with (1+b)^2r instead of R and (1+b)r instead of r, we obtain

\int_{E} U^+{\,{\rm d}} \uplambda \leq \frac{1}{2}\frac{2+b}{b} \Bigl({\sf C}_{U^+}\bigl((1+b)r\bigr) +{\sf N}_{\varDelta_v}\bigl((1+b)r,(1+b)^2r\bigr) \Bigr) \uplambda(E)\ln\frac{\bigl(10(1+b)r\bigr)^2}{\uplambda(E)},

which gives (3).

Uses Lemma 1, Lemma 2

Source and licence

The text on this page is from arXiv:2012.06334v2 by B. N. Khabibullin, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2012/2012.06334.jsonl.