数学杂志  2018, Vol. 38 Issue (4): 713-720   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
吴秀碧
张石梅
龙见仁
石磊
关于一类高阶性线差分方程亚纯解的增长性
吴秀碧1, 张石梅1, 龙见仁1,2, 石磊1    
1. 贵州师范大学数学科学学院, 贵州 贵阳 550001;
2. 北京邮电大学计算机学院; 理学院, 北京 100876
摘要:本文研究了一类高阶线性差分方程非零亚纯解的增长性问题.利用Phragmén-Lindelöf指标函数的方法,获得了当方程拥有多个具有相同型的主导系数时,方程非零亚纯解的增长性的一个下界估计.该结果改进了前人的一些已有结果.
关键词增长级    线性差分方程    主导系数    指标函数    
ON THE GROWTH OF MEROMORPHIC SOLUTIONS OF A CLASS OF HIGHER ORDER LINEAR DIFFERENCE EQUATIONS
WU Xiu-bi1, ZHANG Shi-mei1, LONG Jian-ren1,2, SHI Lei1    
1. School of Mathematical Science, Guizhou Normal University, Guiyang 550001, China;
2. School of Computer Science; School of Science, Beijing University of Posts and Telecommunications, Beijing 100876, China
Abstract: In this paper, we study the growth of meromorphic solutions of a class of higher order linear difference equations. By using the Phragmén-Lindelöf indicator functions, we obtain an estimation of lower bound of the growth of meromorphic solutions of the difference equations if it has at least two dominating coefficients with same type, which improves the previous results.
Key words: order of growth     linear difference equation     dominating coefficients     indicator function    
1 引言及结果

近年来, 有大量的文献关注复差分方程的亚纯解的增长性问题, 例如参考文献[1]和[2]等.本文主要考察以下形式的线性差分方程

$ A_n(z)f(z+n)+\cdots+A_1(z)f(z+1)+A_0(z)f(z)=0, $ (1.1)

其中$A_j(z)~(j=0, \cdots, n)$均是整函数.

2008年, Chiang和Feng研究了方程(1.1)解的增长性, 并给出了其解增长级的一个下界估计.

定理 A[3]  设$A_j(z)~(j=0, \cdots, n)$均为多项式, 且存在$l:0\leq l\leq n$使得$\max\{\deg(P_j):0\leq j\leq n, j\neq l\}<\deg(P_l)$.如果$f(z)$是方程(1.1)的任意非零亚纯解, 则必有$\rho(f)\geq 1$.

同时, 他们还考虑了方程系数为超越整函数的情形, 也给出了方程解的增长级的下界估计.

定理 B[3]  设$A_j(z)~(j=0, \cdots, n)$均为整函数, 且存在$l:0\leq l\leq n$使得$\max\{\rho(A_j):0\leq j\leq n, j\neq l\}<\rho(A_l)$.如果$f(z)$是方程(1.1)的任意非零亚纯解, 则必有$\rho(f)\geq\rho(A_l)+1$.

当系数都是多项式时, 称次数等于$\max\{\deg P_j(z):0\leq j\leq n\}$的系数为主导系数; 当系数有超越整函数时, 称增长级等于$\max\{\rho(A_j(z):0\leq j\leq n\}$的系数为主导系数.注意到定理A和定理B的共同点就是都只有一个主导系数.如果方程拥有多个主导系数时, 是否还有相应结论?这个问题引起很多研究人员的关注. 2011年, 陈宗煊考虑了这个问题, 在弱化定理A的条件下, 得到如下定理.

定理 C[2]  设系数$P_0(z), \cdots, P_n(z)$均为多项式, 且满足$P_n(z)P_0(z)\not\equiv 0$

$ \deg (P_n+\cdots+P_0)=\max\{\deg P_j:j=0, \cdots, n\}\geq1. $ (1.2)

如果$f(z)$是方程(1.2)的任意非零解亚纯解, 则必有$\rho(f)\geq 1$.

Laine和Yang改进了定理B, 并证明了以下结果.

定理 D[4]  假设$A_j(z)~(j=0, \cdots, n)$均为有穷级整函数, $w_j~(j=0, \cdots, n)$为任意复常数, 且型最大的主导系数仅有一个.记$\rho=\max\{\rho(A_j):0\leq j\leq n\}$, 则方程

$ A_n(z)f(z+w_n)+\cdots+A_1(z)f(z+w_1)+A_0(z)f(z)=0 $ (1.3)

的任意非零解都满足$\rho(f)\geq\rho+1$.

同时, Laine和Yang还提出如下问题.

问题  如果方程型最大的主导系数不止一个, 定理B或定理D的结论是否还成立?

2015年, Heittokangas [5]等应用Phragmén-Lindelöf指标函数来研究二阶微分线性微分方程解的增长性, 得到很多结果.受到文献[5]的研究方法的启发, 本文利用指标函数研究高阶线性差分方程问题, 为此先回顾Phragmén-Lindelöf指标函数的定义.

对于一个有穷正级整函数$g(z)$, 其指标函数定义为

$ h_g(\theta)=\limsup\limits_{r\rightarrow\infty}\frac{\log |g(re^{i\theta})|}{r^{\rho(g)}}, ~~\theta\in[-\pi, ~\pi). $

如果$g(z)$的型有限, 则$h_g(\theta)$是一个连续有上界的函数.且对任意$\varepsilon>0$, 易知

$ \limsup\limits_{r\rightarrow\infty}\frac{\log |g(re^{i\theta})|}{r^{\rho(g)+\varepsilon}}=0. $ (1.4)

为了方便叙述起见, 我们引入以下符号.对于$n+1$个有穷正级的整函数$A_j$ $(j=0, \cdots, n), $$\rho=\max\{\rho(A_j):0\leq j\leq n\}$, 记主导函数指标集为$D(n)$, 即$D(n)=\{j:\rho(A_j)=\rho\}$, 并用$|D(n)|$表示$D(n)$的元素个数.

本文得到如下几个结果.

定理 1.1  设$A_j(z)(j=0, \cdots, n)$是有穷正级整函数且至多有一个为无穷型, $w_j(j=0, \cdots, n)$为任意复常数, 记$\rho=\max\{\rho(A_j):0\leq j\leq n\}$.如果方程

$ A_n(z)f(z+w_n)+\cdots+A_1(z)f(z+w_1)+A_0(z)f(z)=0 $ (1.5)

有一个非零解$f(z)$满足$\rho(f)<\rho+1$, 则对于任意的$\theta\in [-\pi, ~\pi)$, 要么$\max\{h_{A_i}(\theta):i\in D(n)\}\leq0$, 要么存在$s, k\in D(n)$使得

$ h_{A_s}(\theta)= h_{A_k}(\theta)\geq \max\{h_{A_i}(\theta):i\in D(n)\}. $

如果方程(1.5)的所有系数的增长级都相同时, 我们得到下面的结果.

定理 1.2  设$A_j(z)~(j=0, \cdots, n)$是有穷正级整函数且至多有一个为无穷型, $w_j(j=0, \cdots, n)$为任意复常数, 记$\rho=\max\{\rho(A_j):0\leq j\leq n\}$$|D(n)| =n+1$.如果方程(1.5)有一个非零解$f(z)$满足$\rho(f)<\rho+1$, 则对于任意的$\theta\in [-\pi, ~\pi)$, 必存在$s, k\in D(n)$使得

$ h_{A_s}(\theta)= h_{A_k}(\theta)\geq \max\{h_{A_i}(\theta):i\in D(n)\}. $

由以上两个定理及其证明过程, 容易得到以下推论.

推论 1.3  设$A_j(z)~(j=0, \cdots, n)$是有穷正级整函数且至多有一个为无穷型, 记$\rho=\max\{\rho(A_j):0\leq j\leq n\}$, 如果存在$\theta_0\in[-\pi, ~\pi)$和某个$s_0\in D(n)$满足下列三个条件之一

(1) $|D(n)|<n+1$$h_{A_{s_0}}(\theta_0)>\max\bigl\{0, h_{A_i}(\theta_0):i\in D(n)-\{s_0\}\bigr\}$;

(2) $|D(n)|=n+1$$h_{A_{s_0}}(\theta_0)>\max\bigl\{h_{A_i}(\theta_0):i\in D(n)-\{s_0\}\bigr\}$;

(3) $|D(n)|=n+1$$\max\{h_{A_i}(\theta_0):i\in D(n)\}<0$.

那么方程(1.3)的每个非零解$f(z)$都满足$\rho(f)\geq\rho+1$.

2 引理

为了证明上述定理, 需要如下的几个引理.

引理 1 [3]  设$\eta_1, ~\eta_2$是任给的两个复数, $f(z)$是一个有穷级的亚纯函数, 则对于任意给定的正数$\varepsilon$, 必有相应的性线测度为零的集合$E\subset[0, ~2\pi)$且当$\theta\not\in E$时, 必存在正常数$R_0=R_0(\theta)>1$, 当$z$满足$\arg z=\theta$$|z|\geq R_0$时, 有

$ \exp\{-r^{\rho(f)-1+\varepsilon}\}\leq\left|\frac{f(z+\eta_1)}{f(z+\eta_2)}\right|\leq\exp\{r^{\rho(f)-1+\varepsilon}\}. $

引理 2  设$A(z)$$B(z)$为整函数且满足下列条件之一

(1) $\rho(B)<\rho(A)\in(0, ~~+\infty)$$h_A(\theta)\geq 0$;

(2) $\rho(B)=\rho(A)\in(0, ~~+\infty)$$h_A(\theta)>0\geq h_B(\theta)$;

(3) $\rho(B)=\rho(A)\in(0, ~~+\infty)$$h_A(\theta)\geq h_B(\theta)>0$.

则有

$ \lim\limits_{r\rightarrow\infty}\sup\frac{\log[|A(re^{i\theta})|+|B(re^{i\theta})|]}{r^{\rho(A)}}\leq h_A(\theta) . $ (2.1)

  当$A(z)$$B(z)$满足条件(1)时, 取$\varepsilon_0=\frac{\rho(A)-\rho(B)}{2}$, 则由级的定义可知, 当$r$充分大时, 有

$ |B(re^{i\theta})|\leq M(r, B)\leq\exp\{r^{\rho(B)+\varepsilon_0}\}. $ (2.2)

同时由$h_A(\theta)$的定义可得, 对任意$\varepsilon>0$, 有

$ |A(re^{i\theta})|\leq\exp\{(h_A(\theta)+\varepsilon)r^{\rho(A)}\}. $ (2.3)

于是结合(2.2)和(2.3)式便有

$ \begin{align*} &\lim\limits_{r\rightarrow\infty}\sup\frac{\log[|A(re^{i\theta})|+|B(re^{i\theta})|]}{r^{\rho(A)}} \\&\leq \lim\limits_{r\rightarrow\infty}\sup\frac{\log\{\exp[(h_A(\theta)+\varepsilon)r^{\rho(A)}]+\exp[r^{\rho(B)+\varepsilon_0}]\}}{r^{\rho(A)}} \\&\leq \lim\limits_{r\rightarrow\infty}\sup\frac{(h_A(\theta)+\varepsilon)r^{\rho(A)}+r^{\rho(B)+\varepsilon_0}+\log2}{r^{\rho(A)}} \leq h_A(\theta)+\varepsilon. \end{align*} $

由于$\varepsilon$的任意性便得结论.

$A(z)$$B(z)$满足条件(2)或者条件(3)时, 类似(2.3)式有

$ |B(re^{i\theta})|\leq\exp\{(h_B(\theta)+\varepsilon)r^{\rho(A)}\}. $ (2.4)

从而由(2.3)和(2.4)式, 并注意到$ h_B(\theta)-h_A(\theta)\leq 0$, 于是有

$ \begin{align*} &\lim\limits_{r\rightarrow\infty}\sup\frac{\log[|A(re^{i\theta})|+|B(re^{i\theta})|]}{r^{\rho(A)}} \\&\leq \lim\limits_{r\rightarrow\infty}\sup\frac{\log\{\exp[(h_A(\theta)+\varepsilon)r^{\rho(A)}]+\exp[(h_B(\theta)+\varepsilon)r^{\rho(B)}]\}}{r^{\rho(A)}} \\&\leq \lim\limits_{r\rightarrow\infty}\sup\frac{(h_A(\theta)+\varepsilon)r^{\rho(A)}+\log\{1+\exp\{[h_B(\theta)-h_A(\theta)]r^{\rho(A)}\}\}}{r^{\rho(A)}} \\&\leq h_A(\theta)+\varepsilon. \end{align*} $

同理由于$\varepsilon$的任意性便得结论.

引理 3  设$A(z)$为有穷正级且型有穷的整函数, $f(z)$是亚纯函数且满足$\rho(f)<\rho(A)+1$.则对任意给定$\eta_1, ~\eta_2$和充分小的$\varepsilon>0$, 必有相应的性线测度为零的集合$E\subset[0, ~2\pi)$且当$\theta\not\in E$, 存在正常数$R_0=R_0(\theta)>1$, 当$z$满足$\arg z=\theta$$|z|=r$充分大时, 有

$ \left|\frac{f(z+\eta_1)}{f(z+\eta_2)}\right||A(re^{i\theta})|\leq\exp\{[h_A(\theta)+\varepsilon]r^{\rho(A)}\}. $

  取$\varepsilon$满足$0<\varepsilon<\frac{\rho(A)+1-\rho(f)}{3}$, 由引理1可知存在线性测度为零的集合$E\subset[0, ~2\pi)$且当$\theta\not\in E$时, 存在正常数$R_0=R_0(\theta)>1$, 当$z$满足$\arg z=\theta$$|z|\geq R_0$, 有

$ \left|\frac{f(z+\eta_1)}{f(z+\eta_2)}\right|\leq\exp\{r^{\rho(f)-1+\varepsilon}\}. $ (2.5)

同时由$h_A(\theta_0)$的定义可得有

$ |A(re^{i\theta})|\leq\exp\{(h_A(\theta)+\frac{\varepsilon}{2})r^{\rho(A)}\}. $ (2.6)

由(2.5)和(2.6)式便知

$ \begin{align*} \left|\frac{f(z+\eta_1)}{f(z+\eta_2)}\right||A(re^{i\theta})| &\leq \exp\{r^{\rho(f)-1+\varepsilon}\}\exp\{(h_A(\theta)+\frac{\varepsilon}{2})r^{\rho(A)}\} \\&=\exp\{r^{\rho(A)}[(h_A(\theta)+\frac{\varepsilon}{2})+\frac{{r^{\rho(f)-1+\varepsilon}}}{r^{\rho(A)}}]\}. \end{align*} $

注意到$\rho(A)>\rho(f)-1+\varepsilon$, 于是当$r$充分大时就得引理结论.

3 定理的证明

定理1.1的证明  根据$D(n)$的定义可知$D(n)$非空且元素多于一个, 否则由定理C便得方程(1.3)的每一个非零解都满足$\rho(f)\geq\rho+1$.易知对于任意$i\in D(n)$, $A_i(z)$都为有限型.若不然, 假设存在$i_0\in D(n)$使得$A_{i_0}(z)$为无穷型, 但由定理条件可知, 无穷型的系数至多只有一个, 于是由定理D可得, 方程(1.3)的每一个非零解都满足$\rho(f)\geq\rho+1$, 这都得到了矛盾.

现假设定理的结论不成立, 即存在$\theta_0\in [-\pi, ~\pi)$, 有$\max\{h_{A_i}(\theta_0):i\in D(n)\}>0$且对于任意$s, k\in D(n)$, 要么

$ h_{A_s}(\theta_0)\neq h_{A_k}(\theta_0), $ (3.1)

要么

$ h_{A_s}(\theta_0)= h_{A_k}(\theta_0)<\max\{h_{A_i}(\theta_0):i\in D(n)\}. $ (3.2)

注意到不管是(3.1)式成立还是(3.2)式成立, 都必定存在$s_0\in D(n)$使得下式成立

$ h_{A_{s_0}}(\theta_0)>\max\bigl\{h_{A_i}(\theta_0):i\in D(n)-\{s_0\}\bigr\}. $ (3.3)

于是由引理1便知存在至多$n$个零测集$E_i~(0\leq i\leq n, i\neq s_0)$, 对于任意

$ \theta\in [-\pi, ~\pi)-\bigcup\limits_{i=0, i\neq s_0}^nE_i $

都能找到$R_0=R_0(\theta)>1$, 使得当$|z|=r\geq R_0$时, 有

$ \exp\{-r^{\rho(f)-1+\varepsilon}\}\leq\left|\frac{f(re^{i\theta}+w_i)}{f(re^{i\theta}+w_{s_0})}\right|\leq\exp\{r^{\rho(f)-1+\varepsilon}\}~~(0\leq i\leq n, i\neq s_0), $ (3.4)

其中$w_0=0$.由于$A_i(z)~(i\in D(n))$都是有限型, 故$h_{A_i}(\theta)~(i\in D(n))$都是连续函数.注意到$\bigcup\limits_{i=0, i\neq s_0}^nE_i$是一个零测度集, 结合(3.3)式便知存在$\theta_1\in[-\pi, ~\pi)-\bigcup\limits_{i=0, i\neq s_0}^nE_i$使得下列不等式成立

$ h_{A_{s_0}}(\theta_1)>\max\bigl\{h_{A_i}(\theta_1):i\in D(n)-\{s_0\}\bigr\}. $ (3.5)

同时, 改写方程可以得到

$ \begin{align*} |A_{s_0}(re^{i\theta_1})| \leq& |A_n(re^{i\theta_1})|\left|\frac{f(re^{i\theta_1}+w_n)}{f(re^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_{s_0+1}(re^{i\theta_1})|\left|\frac{f(re^{i\theta_1}+w_{s_0+1})}{f(re^{i\theta_1}+w_{s_0})}\right| \\&+|A_{s_0-1}(re^{i\theta_1})|\left|\frac{f(re^{i\theta_1}+w_{s_0-1})}{f(re^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_0(re^{i\theta_1})|\left|\frac{f(re^{i\theta_1})}{f(re^{i\theta_1}+w_{s_0})}\right|. \end{align*} $

现设$\delta=\max\bigl\{h_{A_i}(\theta_1):i\in D(n)-\{s_0\}\bigr\}$再分类讨论.如果$\delta>0$, 则存在$k_0\in D(n)-\{s_0\}$使得

$ h_{A_{k_0}}(\theta_1)=\max\bigl\{h_{A_i}(\theta_1):i\in D(n)-\{s_0\}\bigr\}>0. $

$\varepsilon=\frac{\rho+1-\rho(f)}{2}$, 则由(3.4)式可以得到

$ \begin{align*} &|A_{s_0}(re^{i\theta_1})| \\\leq& \exp\{r^{\rho(f)-1+\varepsilon}\}\{|A_n(re^{i\theta_1})|+\cdots+|A_{s_0+1}(re^{i\theta_1})|+|A_{s_0-1}(re^{i\theta_1})|+\cdots+|A_0(re^{i\theta_1})|\}. \end{align*} $

于是由引理2可得

$ \begin{align*} h_{A_{s_0}}(\theta_1)&=\lim\limits_{r\rightarrow\infty}\sup\frac{\log|A_{s_0}(re^{i\theta_1})|}{r^{\rho}} \\&\leq\lim\limits_{r\rightarrow\infty}\sup\frac{r^{\rho(f)-1+\varepsilon}+\log\{|A_n(re^{i\theta_1})|+\cdots+|A_0(re^{i\theta_1})|\}}{r^{\rho}} \\&\leq\lim\limits_{r\rightarrow\infty}\sup\frac{r^{\rho(f)-1+\varepsilon}}{r^{\rho}}+\lim\limits_{r\rightarrow\infty}\sup\frac{\log\{|A_n(re^{i\theta_1})|+\cdots+|A_0(re^{i\theta_1})|\}}{r^{\rho}} \\&\leq h_{A_{k_0}}(\theta_1) =\max\{h_{A_i}(\theta_1):i\in D(n)-\{s_0\}\}. \end{align*} $

这与(3.5)式矛盾.

如果$\delta\leq0.$对于$\forall i\in D(n)-\{s_0\}, ~~j \in\{0, 1, 2, \cdots n\}-D(n)$, 设

$ \varepsilon=\min\bigl\{\frac{h_{A_{s_0}}(\theta_1)- h_{A_i}(\theta_1)}{n+1}, ~\frac{\rho+1-\rho(f)}{2}, ~\frac{h_{A_{s_0}}(\theta_1)}{2}, ~\frac{\rho-\rho(A_j)}{3}\bigr\}, $

则存在点列$r_m$使得

$ \begin{align*} &\exp\{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\} \leq |A_{s_0}(r_me^{i\theta_1})| \\\leq&|A_n(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_n)}{f(r_me^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_{s_0+1}(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_{s_0+1})}{f(r_me^{i\theta_1}+w_{s_0})}\right| \\&+|A_{s_0-1}(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_{s_0-1})}{f(r_me^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_0(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1})}{f(r_me^{i\theta_1}+w_{s_0})}\right|. \end{align*} $

根据引理3可得

$ \begin{align*} &\exp\{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\} \\\leq& \exp\biggl\{r_m^{\rho}\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]\biggr\}+\exp\{r_m^{\rho(f)-1+\varepsilon}\}\sum\limits_{i\not\in D(n)}|A_i(r_me^{i\theta_1})| \\\leq& \exp\biggl\{r_m^{\rho}\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]\biggr\}+\exp\{r_m^{\rho(f)-1+\varepsilon}+r_m^{\rho-\varepsilon}\}. \end{align*} $

不等式右边第二项移到左边, 整理后得

$ \begin{align*} &\exp \{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\}\bigl\{1-\exp\{r_m^{\rho(f)-1+\varepsilon}+r_m^{\rho-\varepsilon}-[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\}\bigr\} \\\leq&\exp \biggl\{r_m^{\rho}\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]\biggr\}. \end{align*} $

由于$h_{A_{s_0}}(\theta_1)-\varepsilon>0$$\rho>\rho(f)-1+\varepsilon$, 于是当$r_m$充分大时, 有

$ \begin{align*} \exp\{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\}\{1-\circ(1)\} \leq \exp\biggl\{r_m^{\rho}\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]\biggr\}. \end{align*} $

两边取对数并注意到$\delta\leq0$, 故有

$ \begin{align*} h_{A_{s_0}}(\theta_1)\leq&\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]+\varepsilon \leq\sum\limits_{i\in D(n)-\{s_0\}}h_{A_i}(\theta_1)+(n+1)\varepsilon\\ \leq& h_{A_i}(\theta_1)+(n+1)\varepsilon. \end{align*} $

$\varepsilon\geq\frac{h_{A_{s_0}}(\theta_1)-h_{A_i}(\theta_1)}{n+1}$.这与$\varepsilon$的设法矛盾.

定理1.2的证明  现假设定理的结论不成立, 即存在$\theta_0\in [-\pi, ~\pi)$, 对于任意$s, k\in D(n)$, 要么

$ h_{A_s}(\theta_0)\neq h_{A_k}(\theta_0); $ (3.6)

要么

$ h_{A_s}(\theta_0)= h_{A_k}(\theta_0)<\max\{h_{A_i}(\theta_0):i\in D(n)\}. $ (3.7)

注意到不管是(3.6)式成立还是(3.7)式成立, 都必定存在$s_0\in D(n)$使得下式成立

$ h_{A_{s_0}}(\theta_0)>\max\bigl\{h_{A_i}(\theta_0):i\in D(n)-\{s_0\}\bigr\}. $

完全类似于定理1.1的证明, 即存在$\theta_1\in[-\pi, \pi)-\bigcup\limits_{i=0, i\neq s_0}^n{E_i}$, 有

$ \exp\{-r^{\rho(f)-1+\varepsilon}\}\leq\left|\frac{f(re^{i\theta_1}+w_i)}{f(re^{i\theta_1}+w_{s_0})}\right|\leq\exp\{r^{\rho(f)-1+\varepsilon}\}~(0\leq i\leq n, i\neq s_0) $ (3.8)

$ h_{A_{s_0}}(\theta_1)>\max\bigl\{h_{A_i}(\theta_1):i\in D(n)-\{s_0\}\bigr\} $ (3.9)

成立.设$\delta=\max\{h_{A_i}(\theta_1):i\in D(n)\}$, 若$\delta>0$, 证明方法完全类似于定理(1.1).

$\delta\leq0$, $\forall i\in D(n)-\{s_0\}$, 设$\varepsilon=\min\biggl\{\frac{h_{A_{s_0}}(\theta_1)- h_{A_i}(\theta_1)}{n+1}, ~\frac{\rho+1-\rho(f)}{2}\biggr\}$, 则存在点列$r_m$使得

$ \begin{align*} &\exp\{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\} \leq |A_{s_0}(r_me^{i\theta_1})| \\\leq&|A_n(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_n)}{f(r_me^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_{s_0+1}(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_{s_0+1})}{f(r_me^{i\theta_1}+w_{s_0})}\right| \\&+|A_{s_0-1}(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1}+w_{s_0-1})}{f(r_me^{i\theta_1}+w_{s_0})}\right|+\cdots+|A_0(r_me^{i\theta_1})|\left|\frac{f(r_me^{i\theta_1})}{f(r_me^{i\theta_1}+w_{s_0})}\right|. \end{align*} $

根据引理3可得

$ \begin{align*} \exp\{[h_{A_{s_0}}(\theta_1)-\varepsilon]r_m^{\rho}\} \leq \exp\biggl\{r_m^{\rho}\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]\biggr\}. \end{align*} $

注意到$\delta\leq0$, 结合(3.9)式便有

$ \begin{align*} h_{A_{s_0}}(\theta_1)\leq\sum\limits_{i\in D(n)-\{s_0\}}[h_{A_i}(\theta_1)+\varepsilon]+\varepsilon \leq\sum\limits_{i\in D(n)-\{s_0\}}h_{A_i}(\theta_1)+(n+1)\varepsilon <h_{A_i}(\theta_1)+(n+1)\varepsilon. \end{align*} $

$\varepsilon>\frac{h_{A_{s_0}}(\theta_1)-h_{A_i}(\theta_1)}{n+1}$.这与$\varepsilon$的设法矛盾.

参考文献
[1] Chen Z X, Shon K H. Value distribution of meromorphic solutions of certain difference Painlevé equations[J]. J. Math. Anal. Appl., 2010, 364: 556–566. DOI:10.1016/j.jmaa.2009.10.021
[2] Chen Z X. Growth and zeros of meromorphic solutions of some linear difference equations[J]. J. Math. Anal. Appl., 2011, 373: 235–241. DOI:10.1016/j.jmaa.2010.06.049
[3] Chiang Y M, Feng S J. On the Nevanlinna characteristic of f(z+η) and difference equations in the complex plane[J]. Ramanujan J., 2008, 16: 105–129. DOI:10.1007/s11139-007-9101-1
[4] Laine I, Yang C C. Clunie theorem for difference and q-difference polynomials[J]. J. London Math. Soc., 2007, 76(3): 556–566. DOI:10.1112/jlms/jdm073
[5] Heittokangas J, Laine I, Toghe K, Wen Z T. Completely regular growth solutions of second order complex linear differential equations[J]. Ann. Acad. Sci. Fenn. Math., 2015, 40: 1–19.
[6] Ishizaki K, Yanagihara N. Wiman-Valiron method for difference equaiotns[J]. Nagoya Math. J., 2004, 175: 75–102. DOI:10.1017/S0027763000008916
[7] Wu X B, Long J R, Heittokangas J, Qiu K E. Second-order complex linear differential equations with special functions or extremal functions as coefficients[J]. Electron. J. Diff. Equa., 2015, 2015(143): 1–15.
[8] Long J R. Growth of solutions of higher order complex linear differential equaitons in an angular domain[J]. J. Math., 2015, 35(6): 1533–1540.
[9] Yang L. Value distribution theory[M]. Berlin: Springer-Verlag, 1993.