Introduction and main results
We define Z + = Z [ 0 , + ∞ ) = { n ∈ Z : 0 ≤ n < + ∞ } \mathbb{Z}_{+}=\mathbb{Z}[0,+\infty)=\{n\in\mathbb{Z}:0\le n<+\infty\} Z + = Z [ 0 , + ∞ ) = { n ∈ Z : 0 ≤ n < + ∞ } and Z + = Z ( 0 , + ∞ ) = { n ∈ Z : 0 < n < + ∞ } \mathbb{Z}^{+}=\mathbb{Z}(0,+\infty)=\{n\in\mathbb{Z}:0<n<+\infty\} Z + = Z ( 0 , + ∞ ) = { n ∈ Z : 0 < n < + ∞ } . On C m \mathbb{C}^{m} C m , we define
∂ z i = ∂ ∂ z i , ∂ z j z i 2 = ∂ 2 ∂ z j ∂ z i , … , ∂ z i l i = ∂ l i ∂ z i l i and ∂ I = ∂ ∣ I ∣ ∂ z 1 i 1 ⋯ ∂ z m i m . \partial_{z_i}=\frac{\partial}{\partial z_i},\quad\partial_{z_jz_i}^{2}=\frac{\partial^2}{\partial z_j\partial z_i},\ldots,\partial_{z_i}^{l_i}=\frac{\partial^{l_i}}{\partial z_i^{l_i}}\quad\text{and}\quad\partial^I=\frac{\partial^{|I|}}{\partial z_1^{i_1}\cdots\partial z_m^{i_m}}. ∂ z i = ∂ z i ∂ , ∂ z j z i 2 = ∂ z j ∂ z i ∂ 2 , … , ∂ z i l i = ∂ z i l i ∂ l i and ∂ I = ∂ z 1 i 1 ⋯ ∂ z m i m ∂ ∣ I ∣ .
where l i ∈ Z + l_i\in\mathbb{Z}^{+} l i ∈ Z + (i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m ) and I = ( i 1 , … , i m ) ∈ Z + m I=(i_1,\ldots,i_m)\in\mathbb{Z}_{+}^{m} I = ( i 1 , … , i m ) ∈ Z + m is a multi-index such that
∣ I ∣ = ∑ j = 1 m i j . |I|=\sum_{j=1}^{m}i_j. ∣ I ∣ = j = 1 ∑ m i j .
Let G ≠ ∅ G\ne\varnothing G = ∅ be an open subset of C m \mathbb{C}^{m} C m . Let f f f be a holomorphic function on G G G . For a point a ∈ C m a\in\mathbb{C}^{m} a ∈ C m , we write f ( z ) = ∑ i = 0 ∞ P i ( z − a ) f(z)=\sum_{i=0}^{\infty}P_i(z-a) f ( z ) = ∑ i = 0 ∞ P i ( z − a ) , where the term P i ( z − a ) P_i(z-a) P i ( z − a ) is either identically zero or a homogeneous polynomial of degree i i i . Denote the zero-multiplicity of f f f at a a a by μ f 0 ( a ) = min { i : P i ( z − a ) ≢ 0 } \mu_f^0(a)=\min\{i:P_i(z-a)\not\equiv0\} μ f 0 ( a ) = min { i : P i ( z − a ) ≡ 0 } in the sense of [9 Definition 2.1] .
Let f f f be a meromorphic function on G G G . Then there exist holomorphic functions g g g and h h h such that h f = g hf=g h f = g on G G G and dim z I f ≤ m − 2 \dim_z I_f\le m-2 dim z I f ≤ m − 2 , where I f = { z ∈ C m : g ( z ) = h ( z ) = 0 } I_f=\{z\in\mathbb{C}^{m}:g(z)=h(z)=0\} I f = { z ∈ C m : g ( z ) = h ( z ) = 0 } . Therefore the c c c -multiplicity of f f f is just μ f c = μ g − c h 0 \mu_f^c=\mu_{g-ch}^0 μ f c = μ g − c h 0 if c ∈ C c\in\mathbb{C} c ∈ C and μ f ∞ = μ h 0 \mu_f^\infty=\mu_h^0 μ f ∞ = μ h 0 if c = ∞ c=\infty c = ∞ . The function μ f c : C m → Z \mu_f^c:\mathbb{C}^m\to\mathbb{Z} μ f c : C m → Z is nonnegative and is called the c c c -divisor of f f f . If f ≢ 0 f\not\equiv0 f ≡ 0 on each component of G G G , then ν = μ f = μ f 0 − μ f ∞ \nu=\mu_f=\mu_f^0-\mu_f^\infty ν = μ f = μ f 0 − μ f ∞ is called the divisor of f f f . For k ∈ N k\in\mathbb{N} k ∈ N , define the truncated multiplicity functions on C m \mathbb{C}^m C m by μ f , k a ( z ) = min { μ f a ( z ) , k } \mu_{f,k}^a(z)=\min\{\mu_f^a(z),k\} μ f , k a ( z ) = min { μ f a ( z ) , k } .
Let f f f , g g g and a a a be meromorphic functions on C m \mathbb{C}^{m} C m . Then one can find three pairs of entire functions f 1 f_1 f 1 and f 2 f_2 f 2 , g 1 g_1 g 1 and g 2 g_2 g 2 , and a 1 a_1 a 1 and a 2 a_2 a 2 , in which each pair is coprime at each point in C m \mathbb{C}^{m} C m such that f = f 2 / f 1 f=f_2/f_1 f = f 2 / f 1 , g = g 2 / g 1 g=g_2/g_1 g = g 2 / g 1 and a = a 2 / a 1 a=a_2/a_1 a = a 2 / a 1 . We say that f f f and g g g share a a a by counting multiplicities (CM) if
μ a 1 f 2 − a 2 f 1 0 = μ a 1 g 2 − a 2 g 1 0 ( a ≠ ∞ ) and μ f 1 0 = μ g 1 0 ( a = ∞ ) . \mu_{a_1f_2-a_2f_1}^0=\mu_{a_1g_2-a_2g_1}^0\quad(a\ne\infty)\quad\text{and}\quad\mu_{f_1}^0=\mu_{g_1}^0\quad(a=\infty). μ a 1 f 2 − a 2 f 1 0 = μ a 1 g 2 − a 2 g 1 0 ( a = ∞ ) and μ f 1 0 = μ g 1 0 ( a = ∞ ) .
Again we say that f f f and g g g share a a a by ignoring multiplicities (IM) if
μ a 1 f 2 − a 2 f 1 , 1 0 = μ a 1 g 2 − a 2 g 1 , 1 0 ( a ≠ ∞ ) and μ f 1 , 1 0 = μ g 1 , 1 0 ( a = ∞ ) . \mu_{a_1f_2-a_2f_1,1}^0=\mu_{a_1g_2-a_2g_1,1}^0\quad(a\ne\infty)\quad\text{and}\quad\mu_{f_1,1}^0=\mu_{g_1,1}^0\quad(a=\infty). μ a 1 f 2 − a 2 f 1 , 1 0 = μ a 1 g 2 − a 2 g 1 , 1 0 ( a = ∞ ) and μ f 1 , 1 0 = μ g 1 , 1 0 ( a = ∞ ) .
Typeset by AMS-LaTeX.
Understanding the relationship between meromorphic functions by their shared zeros and poles is a key area of research in complex analysis. This allows for the characterization of functions based on their value distribution, particularly in relation to their derivatives, a field known as Nevanlinna Theory. This helps in problems such as determining if a function is uniquely determined by its zeros and poles.
Nevanlinna theory gives powerful tools to compare the growth and value distribution of meromorphic functions. A foundational result is that if two meromorphic functions share five distinct values IM, they must be identical.
Two meromorphic functions f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) in C \mathbb{C} C sharing the same zeros and poles with their multiplicities implies they differ only by a constant multiplier. More generally, if their zeros and poles are the CM shared, they are closely related through a limited set of transformations, such as being related by a Möbius transformation or an exponential function. Research in this area uses Nevanlinna theory to understand how sharing zeros and poles with derivatives can determine the relationship between functions. In the paper, we are interested in the theory behind meromorphic functions sharing zeros and poles, and also the scenario where some derivatives of these functions share zeros. This falls under the broader framework of uniqueness theory, particularly using Nevanlinna theory and value distribution.
Research on “Meromorphic functions sharing zeros and poles and also some of their derivatives sharing zeros” focuses on the uniqueness of meromorphic functions by establishing conditions under which two non-constant meromorphic functions f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) in C \mathbb{C} C , must be related, when they share specific zeros and poles, and when their derivatives share zeros. Researchers study conditions under which f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) are related if their derivatives f ( k ) ( z ) f^{(k)}(z) f ( k ) ( z ) and g ( k ) ( z ) g^{(k)}(z) g ( k ) ( z ) share certain values CM or IM. By weakening the sharing conditions for the functions and their derivatives, mathematicians can derive more precise and comprehensive conclusions about the relationship between these functions. This area has applications in complex dynamics:
(a) Studying how iterations of meromorphic functions behave,
(b) Functions sharing poles/zeros with their iterates can exhibit special types of Julia sets or Fatou sets (stable/chaotic regions)
and in differential equations:
(a) If solutions to a differential equation share poles/zeros or those of their derivatives, that can help classify all solutions or establish uniqueness,
(b) Used in Painlevé equations, where meromorphic solutions have special value-sharing properties.
For non-constant entire functions f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) in C \mathbb{C} C with finite order, C. C. Yang [38 ] and G. G. Gundersen [12 ] studied the case where f ( j ) ( z ) f^{(j)}(z) f ( j ) ( z ) and g ( j ) ( z ) g^{(j)}(z) g ( j ) ( z ) share 0 CM for j = 0 , 1 j=0,1 j = 0 , 1 and examined what implications this strong form of sharing (CM sharing of 0 for both the functions and their first derivatives) has on the relationship between f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) .
For non-constant meromorphic functions f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) in C \mathbb{C} C , we know that f ( j ) ( z ) f^{(j)}(z) f ( j ) ( z ) and g ( j ) ( z ) g^{(j)}(z) g ( j ) ( z ) share 0 and ∞ \infty ∞ CM for each non-negative integer j j j whenever f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) satisfy one of the following four conditions:
(i) f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c \in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } ;
(ii) f ( z ) = e a z + b f(z)=e^{az+b} f ( z ) = e a z + b and g ( z ) = e c z + d g(z)=e^{cz+d} g ( z ) = e cz + d , where a , c ∈ C ∖ { 0 } a,c \in\mathbb{C}\setminus\{0\} a , c ∈ C ∖ { 0 } and b , d ∈ C b,d \in\mathbb{C} b , d ∈ C ;
(iii) f ( z ) = a ( 1 − b e c z ) f(z)=a(1-be^{cz}) f ( z ) = a ( 1 − b e cz ) and g ( z ) = d ( e − c z − b ) g(z)=d(e^{-cz}-b) g ( z ) = d ( e − cz − b ) , where a , b , c , d ∈ C ∖ { 0 } a,b,c,d \in\mathbb{C}\setminus\{0\} a , b , c , d ∈ C ∖ { 0 } ;
(iv) f ( z ) = a / ( 1 − b e β ( z ) ) f(z)=a/(1-be^{\beta(z)}) f ( z ) = a / ( 1 − b e β ( z ) ) and g ( z ) = a / ( e − β ( z ) − b ) g(z)=a/(e^{-\beta(z)}-b) g ( z ) = a / ( e − β ( z ) − b ) , where a , b ∈ C ∖ { 0 } a,b \in\mathbb{C}\setminus\{0\} a , b ∈ C ∖ { 0 } and β \beta β is a non-constant entire function.
In 1984, A. Hinkkanen [2 p. 492] proposed the following problem:
Question A. (Hinkkanen’s Problem) Does there exist an integer n ≥ 2 n \ge2 n ≥ 2 such that f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) satisfy one of the conditions (i)-(iv) when f ( j ) ( z ) f^{(j)}(z) f ( j ) ( z ) and g ( j ) ( z ) g^{(j)}(z) g ( j ) ( z ) share the values 0 0 0 and ∞ \infty ∞ CM for j = 0 , 1 , … , n j=0,1,\ldots,n j = 0 , 1 , … , n ?
Hinkkanen’s Problem refers to a notable result in the uniqueness theory of meromorphic functions. Hinkkanen originally raised this question in the context of how many derivative zero sets are needed to uniquely determine a meromorphic function.
In 1989, Köhler [16 ] proved that n = 6 n=6 n = 6 solves the problem. In 1990, Tohge [34 ] considered the case n = 2 , 3 n=2,3 n = 2 , 3 under restrictions on the growth of f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) . According to the answer to Hinkkanen’s Problem, Frank et al. [8 ] demonstrated that a meromorphic function may be uniquely identified if we know the positions of its zeros, poles, and the zeros of its first four derivatives. In fact, Frank et al. [8 ] proved that: The sharp answer to Hinkkanen’s problem is n = 4 n=4 n = 4 . In Example 1 [8 ] , Frank et al. [8 ] demonstrated the sharpness of Hinkkanen’s Problem for n = 4 n=4 n = 4 .
In 1989, Köhler [16 ] examined Hinkkanen’s Problem with the additional condition that the order of f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) are finite. He proved the following result.
Theorem A. [16 ] [Theorem 2] Let f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) be two non-constant meromorphic functions in C \mathbb{C} C of finite order such that f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) share 0 0 0 and ∞ \infty ∞ CM. Suppose f ( 1 ) ( z ) f^{(1)}(z) f ( 1 ) ( z ) and g ( 1 ) ( z ) g^{(1)}(z) g ( 1 ) ( z ) share 0 0 0 CM and for k > 1 k>1 k > 1 , f ( k ) ( z ) f^{(k)}(z) f ( k ) ( z ) and g ( k ) ( z ) g^{(k)}(z) g ( k ) ( z ) share 0 0 0 CM. Then f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) satisfy one of the above case (i)-(iv) and β ( z ) \beta(z) β ( z ) in (iv) is a non-constant polynomial.
By the example f ( z ) = exp ( exp ( z ) ) f(z)=\exp(\exp(z)) f ( z ) = exp ( exp ( z )) and g ( z ) = exp ( exp ( − z ) ) g(z)=\exp(\exp(-z)) g ( z ) = exp ( exp ( − z )) , it is clear that Theorem A does not hold for functions of infinite order.
Theorem A leads to the following natural question:
Question 1.1. Does Theorem A work in several complex variables?
We respond to Question 1.1 in the paper with the following outcome.
Theorem 1.1. Let f f f and g g g be two non-constant finite order meromorphic function in C m \mathbb{C}^{m} C m such that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM. Suppose ∂ z i ( f ) \partial_{z_i}(f) ∂ z i ( f ) and ∂ z i ( g ) \partial_{z_i}(g) ∂ z i ( g ) share 0 0 0 CM and for one n > 1 n>1 n > 1 , ∂ z i n ( f ) \partial_{z_i}^{n}(f) ∂ z i n ( f ) and ∂ z i n ( g ) \partial_{z_i}^{n}(g) ∂ z i n ( g ) share 0 0 0 CM, where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Then f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) satisfy one of the following four conditions:
f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } ;
f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a f(z)=e^{a_1z_1+a_2z_2+\cdots+a_mz_m+a} f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a and g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b g(z)=e^{b_1z_1+b_2z_2+\cdots+b_mz_m+b} g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b , where a , a 1 , … , a m , b 1 , … , b m , b ∈ C a,a_1,\ldots,a_m,b_1,\ldots,b_m,b\in\mathbb{C} a , a 1 , … , a m , b 1 , … , b m , b ∈ C such that ( a 1 , … , a m ) ≠ ( 0 , … , 0 ) (a_1,\ldots,a_m)\ne(0,\ldots,0) ( a 1 , … , a m ) = ( 0 , … , 0 ) and ( b 1 , … , b m ) ≠ ( 0 , … , 0 ) (b_1,\ldots,b_m)\ne(0,\ldots,0) ( b 1 , … , b m ) = ( 0 , … , 0 ) .
f ( z ) = a ( 1 − b e ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) ) f(z)=a\left(1-be^{(a_1z_1+a_2z_2+\cdots+a_mz_m)}\right) f ( z ) = a ( 1 − b e ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) ) and g ( z ) = c ( e − ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) − b ) g(z)=c\left(e^{-(a_1z_1+a_2z_2+\cdots+a_mz_m)}-b\right) g ( z ) = c ( e − ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) − b ) , where a , b , c ∈ C ∖ { 0 } a,b,c\in\mathbb{C}\setminus\{0\} a , b , c ∈ C ∖ { 0 } and a 1 , a 2 , … , a m ∈ C a_1,a_2,\ldots,a_m\in\mathbb{C} a 1 , a 2 , … , a m ∈ C such that ( a 1 , a 2 , … , a m ) ≠ ( 0 , 0 , … , 0 ) (a_1,a_2,\ldots,a_m)\ne(0,0,\ldots,0) ( a 1 , a 2 , … , a m ) = ( 0 , 0 , … , 0 ) ;
f ( z ) = d 1 − e − P ( z ) f(z)=\frac{d}{1-e^{-P(z)}} f ( z ) = 1 − e − P ( z ) d and g ( z ) = d e P ( z ) − c g(z)=\frac{d}{e^{P(z)}-c} g ( z ) = e P ( z ) − c d , where c , d ∈ C ∖ { 0 } c,d\in\mathbb{C}\setminus\{0\} c , d ∈ C ∖ { 0 } and P P P is a non-constant polynomial in C m \mathbb{C}^{m} C m .
Let
f ( z ) = ( e 2 ( z 1 + ⋯ + z m ) − 1 ) exp ( − i e z 1 + ⋯ + z m ) f(z)=\left(e^{2(z_1+\cdots+z_m)}-1\right)\exp\left(-ie^{z_1+\cdots+z_m}\right) f ( z ) = ( e 2 ( z 1 + ⋯ + z m ) − 1 ) exp ( − i e z 1 + ⋯ + z m )
and
g ( z ) = ( 1 − e − 2 ( z 1 + ⋯ + z m ) ) exp ( i e − ( z 1 + ⋯ + z m ) ) . g(z)=\left(1-e^{-2(z_1+\cdots+z_m)}\right)\exp\left(ie^{-(z_1+\cdots+z_m)}\right). g ( z ) = ( 1 − e − 2 ( z 1 + ⋯ + z m ) ) exp ( i e − ( z 1 + ⋯ + z m ) ) .
Clearly f f f and g g g share 0 0 0 and ∞ \infty ∞ CM, ∂ z i ( f ) \partial_{z_i}(f) ∂ z i ( f ) and ∂ z i ( g ) \partial_{z_i}(g) ∂ z i ( g ) share 0 0 0 CM and ∂ z i 2 ( f ) \partial_{z_i}^{2}(f) ∂ z i 2 ( f ) and ∂ z i 2 ( g ) \partial_{z_i}^{2}(g) ∂ z i 2 ( g ) share 0 0 0 , where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m , but f ( z ) f(z) f ( z ) and g ( z ) g(z) g ( z ) do not satisfy any case of (i)-(iv) of Theorem 1.1. Consequently for meromorphic functions of infinite order Theorem 1.1 is invalid.
Before going to the proof of our main result, we will briefly recall some of the standard notation of Nevanlinna theory in C m \mathbb{C}^{m} C m (see [14 , 33 ] ). Nevanlinna theory, originally formulated for meromorphic functions of a single complex variable, has been extended to the setting of several complex variables to study the value distribution of meromorphic maps from C m \mathbb{C}^{m} C m into complex projective spaces or complex manifolds. It provides profound insights into: (a) The growth and value distribution of meromorphic mappings; (b) Uniqueness and rigidity phenomena for solutions of linear and nonlinear PDEs and (c) Criteria for compactness and convergence in families of meromorphic maps. Its applications span complex analysis, partial differential equations, complex geometry, and mathematical physics, offering both powerful techniques and elegant results. These references [1 ] , [4 ] , [6 ] –[7 ] , [15 ] , [17 ] –[22 ] , [24 ] , [25 ] , [26 ] , [27 ] , [28 ] , [29 ] , [30 ] , [31 ] , [32 ] , [35 ] , [36 ] and [37 ] provide a foundation for understanding the current state of research in Nevanlinna value distribution theory in several complex variables.
On C m \mathbb{C}^{m} C m , the exterior derivative d d d splits d = ∂ + ∂ ˉ d=\partial+\bar{\partial} d = ∂ + ∂ ˉ and twists to d c = ι 4 π ( ∂ ˉ − ∂ ) d^{c}=\frac{\iota}{4\pi}(\bar{\partial}-\partial) d c = 4 π ι ( ∂ ˉ − ∂ ) . Clearly
d d c = ι 2 π ∂ ∂ ˉ . dd^{c}=\frac{\iota}{2\pi}\partial\bar{\partial}. d d c = 2 π ι ∂ ∂ ˉ .
A non-negative function τ : C m → R [ 0 , b ) \tau:\mathbb{C}^{m}\to\mathbb{R}[0,b) τ : C m → R [ 0 , b ) (0 < b ≤ ∞ 0<b\leq\infty 0 < b ≤ ∞ ) of class C ∞ C^{\infty} C ∞ is said to be an exhaustion of C m \mathbb{C}^{m} C m if τ − 1 ( K ) \tau^{-1}(K) τ − 1 ( K ) is compact whenever K K K is. An exhaustion τ m \tau_{m} τ m of C m \mathbb{C}^{m} C m is defined by τ m ( z ) = ∥ z ∥ 2 \tau_{m}(z)=\lVert z\rVert^{2} τ m ( z ) = ∥ z ∥ 2 . The standard Kaehler metric on C m \mathbb{C}^{m} C m is given by υ m = d d c τ m > 0 \upsilon_{m}=dd^{c}\tau_{m}>0 υ m = d d c τ m > 0 . On C m ∖ { 0 } \mathbb{C}^{m}\setminus\{0\} C m ∖ { 0 } , we define
ω m = d d c log τ m ≥ 0 and σ m = d c log τ m ∧ ω m m − 1 . \omega_{m}=dd^{c}\log\tau_{m}\geq0\quad\text{and}\quad\sigma_{m}=d^{c}\log\tau_{m}\wedge\omega_{m}^{m-1}. ω m = d d c log τ m ≥ 0 and σ m = d c log τ m ∧ ω m m − 1 .
For any S ⊆ C m S\subseteq\mathbb{C}^{m} S ⊆ C m , let S [ r ] S[r] S [ r ] and S ⟨ r ⟩ S\langle r\rangle S ⟨ r ⟩ be the intersection of S S S with respectively the closed ball and the sphere of radius r > 0 r>0 r > 0 centered at 0 ∈ C m 0\in\mathbb{C}^{m} 0 ∈ C m . Also for t > 0 t>0 t > 0 , the counting function n ν n_{\nu} n ν is defined by
n ν ( t ) = t − 2 ( m − 1 ) ∫ A [ t ] ν υ m m − 1 , n_{\nu}(t)=t^{-2(m-1)}\int_{A[t]}\nu\upsilon_{m}^{m-1}, n ν ( t ) = t − 2 ( m − 1 ) ∫ A [ t ] ν υ m m − 1 ,
where A = supp ν = supp μ f = { z ∈ G : ν ( z ) ≠ 0 } A=\operatorname{supp}\nu=\operatorname{supp}\mu_{f}=\{z\in G:\nu(z)\ne0\} A = supp ν = supp μ f = { z ∈ G : ν ( z ) = 0 } . The valence function of ν \nu ν is defined by
N ν ( r ) = N ν ( r , r 0 ) = ∫ r 0 r n ν ( t ) d t t ( r ≥ r 0 ) . N_{\nu}(r)=N_{\nu}(r,r_{0})=\int_{r_{0}}^{r}n_{\nu}(t)\frac{dt}{t}\quad(r\geq r_{0}). N ν ( r ) = N ν ( r , r 0 ) = ∫ r 0 r n ν ( t ) t d t ( r ≥ r 0 ) .
For a ∈ P 1 = C ∪ { ∞ } a\in\mathbb{P}^{1}=\mathbb{C}\cup\{\infty\} a ∈ P 1 = C ∪ { ∞ } , we write N μ f a ( r ) = N ( r , a ; f ) N_{\mu_{f}^{a}}(r)=N(r,a;f) N μ f a ( r ) = N ( r , a ; f ) if a ∈ C a\in\mathbb{C} a ∈ C and N μ f a ( r ) = N ( r , f ) N_{\mu_{f}^{a}}(r)=N(r,f) N μ f a ( r ) = N ( r , f ) if a = ∞ a=\infty a = ∞ . For k ∈ N k\in\mathbb{N} k ∈ N , define the truncated multiplicity functions on C m \mathbb{C}^{m} C m by
μ ˉ f = k a ( z ) = { 1 , if μ f a ( z ) = k 0 , if μ f a ( z ) ≠ k and μ ˉ f ( k a ( z ) = { 1 , if μ f a ( z ) ≥ k 0 , if μ f a ( z ) < k . \bar{\mu}_{f=k}^{a}(z)=
\begin{cases}
1, & \text{if }\mu_{f}^{a}(z)=k\\
0, & \text{if }\mu_{f}^{a}(z)\ne k
\end{cases}
\quad\text{and}\quad
\bar{\mu}_{f(k}^{a}(z)=
\begin{cases}
1, & \text{if }\mu_{f}^{a}(z)\geq k\\
0, & \text{if }\mu_{f}^{a}(z)<k.
\end{cases} μ ˉ f = k a ( z ) = { 1 , 0 , if μ f a ( z ) = k if μ f a ( z ) = k and μ ˉ f ( k a ( z ) = { 1 , 0 , if μ f a ( z ) ≥ k if μ f a ( z ) < k .
and the truncated valence functions
N ν ( t ) = { N ‾ ( t , a ; f ) , if ν = μ f , 1 a N ‾ ( k ( t , a ; f ) , if ν = μ ˉ f ( k a . N_{\nu}(t)=
\begin{cases}
\overline{N}(t,a;f), & \text{if }\nu=\mu_{f,1}^{a}\\
\overline{N}_{(k}(t,a;f), & \text{if }\nu=\bar{\mu}_{f(k}^{a}.
\end{cases} N ν ( t ) = { N ( t , a ; f ) , N ( k ( t , a ; f ) , if ν = μ f , 1 a if ν = μ ˉ f ( k a .
Definition 1.1. An algebraic subset X X X of C m \mathbb{C}^{m} C m is defined as a subset
X = { z ∈ C m : P j ( z ) = 0 , 1 ≤ j ≤ l } X=\{z\in\mathbb{C}^{m}:P_{j}(z)=0,\ 1\leq j\leq l\} X = { z ∈ C m : P j ( z ) = 0 , 1 ≤ j ≤ l } with finitely many polynomials P 1 ( z ) , … , P l ( z ) P_{1}(z),\ldots,P_{l}(z) P 1 ( z ) , … , P l ( z ) . A divisor ν \nu ν on C m \mathbb{C}^{m} C m is said to be algebraic if ν \nu ν is the zero divisor of a polynomial. In this case the counting function n ν n_{\nu} n ν is bounded (see [10 , 33 ] ).
With the help of the positive logarithm function, we define the proximity function of f f f by
m ( r , f ) = C m ⟨ r ; log + ∣ f ∣ ⟩ = ∫ C m ⟨ r ⟩ log + ∣ f ∣ σ m . m(r,f)=\mathbb{C}^{m}\langle r;\log^{+}|f|\rangle=\int_{\mathbb{C}^{m}\langle r\rangle}\log^{+}|f|\,\sigma_{m}. m ( r , f ) = C m ⟨ r ; log + ∣ f ∣ ⟩ = ∫ C m ⟨ r ⟩ log + ∣ f ∣ σ m .
The characteristic function of f f f is defined by T ( r , f ) = m ( r , f ) + N ( r , f ) T(r,f)=m(r,f)+N(r,f) T ( r , f ) = m ( r , f ) + N ( r , f ) . We define m ( r , a ; f ) = m ( r , f ) m(r,a;f)=m(r,f) m ( r , a ; f ) = m ( r , f ) if a = ∞ a=\infty a = ∞ and m ( r , a ; f ) = m ( r , 1 / ( f − a ) ) m(r,a;f)=m(r,1/(f-a)) m ( r , a ; f ) = m ( r , 1/ ( f − a )) if a a a is a finite complex number. If a ∈ C a\in\mathbb{C} a ∈ C , then the first main theorem states that m ( r , a ; f ) + N ( r , a ; f ) = T ( r , f ) + O ( 1 ) m(r,a;f)+N(r,a;f)=T(r,f)+O(1) m ( r , a ; f ) + N ( r , a ; f ) = T ( r , f ) + O ( 1 ) , where O ( 1 ) O(1) O ( 1 ) denotes a bounded function when r r r is sufficiently large. We define the order of f f f by
ρ ( f ) : = lim sup r → ∞ log T ( r , f ) log r . \rho(f) := \limsup_{r\to\infty} \frac{\log T(r,f)}{\log r}. ρ ( f ) := r → ∞ lim sup log r log T ( r , f ) .
Definition 1.2. Let U ⊂ C m U \subset\mathbb{C}^{m} U ⊂ C m be an open subset. A closed subset A ⊂ U A \subset U A ⊂ U is said to be analytic, if for each a ∈ A a \in A a ∈ A , there are a finite number of holomorphic functions f 1 ( z ) , f 2 ( z ) , … , f l ( z ) f_1(z), f_2(z), \ldots, f_l(z) f 1 ( z ) , f 2 ( z ) , … , f l ( z ) defined in a neighbourhood N ( a ) N(a) N ( a ) of a a a such that (see [23 pp. 42] )
A ∩ N ( a ) = { z ∈ N ( a ) : f 1 ( z ) = f 2 ( z ) = ⋯ = f l ( z ) = 0 } . A \cap N(a) = \{z \in N(a): f_1(z) = f_2(z) = \cdots= f_l(z) = 0\}. A ∩ N ( a ) = { z ∈ N ( a ) : f 1 ( z ) = f 2 ( z ) = ⋯ = f l ( z ) = 0 } . If f j ( z ) f_j(z) f j ( z ) , 1 ≤ j ≤ l 1 \leq j \leq l 1 ≤ j ≤ l can be taken so that their differentials at a ∈ A a \in A a ∈ A , d f 1 ( a ) , d f 2 ( a ) , … , d f l ( a ) df_1(a), df_2(a), \ldots, df_l(a) d f 1 ( a ) , d f 2 ( a ) , … , d f l ( a ) are linearly independent, then a a a is called a regular point of A A A . A point of A A A which is not regular called a singular point.
Auxiliary lemmas
In the proof of Theorem 1.1, we use the following key lemmas.
Lemma 2.1. [14 Lemma 1.37] Let f : C m → P 1 f:\mathbb{C}^{m}\to\mathbb{P}^{1} f : C m → P 1 be a non-constant meromorphic function and let I = ( i 1 , … , i m ) ∈ Z + m I=(i_1,\ldots,i_m)\in\mathbb{Z}_{+}^{m} I = ( i 1 , … , i m ) ∈ Z + m be a multi-index. Then for any ε > 0 \varepsilon>0 ε > 0 , we have
m ( r , ∂ I ( f ) f ) ≤ ∣ I ∣ log + T ( r , f ) + ∣ I ∣ ( 1 + ε ) log + log T ( r , f ) + O ( 1 ) , m\left(r,\frac{\partial^{I}(f)}{f}\right)\leq|I|\log^{+}T(r,f)+|I|(1+\varepsilon)\log^{+}\log T(r,f)+O(1), m ( r , f ∂ I ( f ) ) ≤ ∣ I ∣ log + T ( r , f ) + ∣ I ∣ ( 1 + ε ) log + log T ( r , f ) + O ( 1 ) , for all large r r r outside a set E E E with ∫ E d log r < ∞ \int_E d\log r<\infty ∫ E d log r < ∞ .
Lemma 2.2. [13 Proposition 3.2] Let P P P be a non-constant entire function in C m \mathbb{C}^{m} C m . Then
ρ ( e P ) = { deg ( P ) if P is a polynomial , + ∞ otherwise . \rho(e^{P}) =
\begin{cases}
\deg(P) & \text{if } P \text{ is a polynomial},\\
+\infty& \text{otherwise}.
\end{cases} ρ ( e P ) = { deg ( P ) + ∞ if P is a polynomial , otherwise .
Lemma 2.3. Let F F F be a non-constant meromorphic function in C m \mathbb{C}^{m} C m and f k = ∂ z k ( F ) F f_k=\frac{\partial_{z_k}(F)}{F} f k = F ∂ z k ( F ) , where k = 1 , 2 , … , m k=1,2,\ldots,m k = 1 , 2 , … , m . Then
∂ z k l ( F ) F = f k l + l ( l − 1 ) 2 f k l − 2 ∂ z k 2 ( f k ) + A ~ l f k l − 3 ∂ z k 3 ( f k ) + B ~ l f k l − 4 ( ∂ z k ( f k ) ) 2 + P l − 3 ( f k ) , \frac{\partial_{z_k}^{l}(F)}{F}=f_k^{l}+\frac{l(l-1)}{2}f_k^{l-2}\partial_{z_k}^{2}(f_k)+\widetilde{A}_l f_k^{l-3}\partial_{z_k}^{3}(f_k)+\widetilde{B}_l f_k^{l-4}\bigl(\partial_{z_k}(f_k)\bigr)^2+P_{l-3}(f_k), F ∂ z k l ( F ) = f k l + 2 l ( l − 1 ) f k l − 2 ∂ z k 2 ( f k ) + A l f k l − 3 ∂ z k 3 ( f k ) + B l f k l − 4 ( ∂ z k ( f k ) ) 2 + P l − 3 ( f k ) , where
A ~ l = l ( l − 1 ) ( l − 2 ) 6 , B ~ l = l ( l − 1 ) ( l − 2 ) ( l − 3 ) 8 , \widetilde{A}_l=\frac{l(l-1)(l-2)}{6},\qquad\widetilde{B}_l=\frac{l(l-1)(l-2)(l-3)}{8}, A l = 6 l ( l − 1 ) ( l − 2 ) , B l = 8 l ( l − 1 ) ( l − 2 ) ( l − 3 ) , and P l − 3 ( f k ) P_{l-3}(f_k) P l − 3 ( f k ) is a differential polynomial with constant coefficients, which vanishes identically for l ≤ 3 l\leq3 l ≤ 3 and has degree l − 3 l-3 l − 3 when l > 3 l>3 l > 3 .
Proof. We have
f k = ∂ z k ( F ) F . (1) f_k=\frac{\partial_{z_k}(F)}{F}. \tag*{(1)} f k = F ∂ z k ( F ) . (1)
where k = 1 , 2 , … , m k=1,2,\ldots,m k = 1 , 2 , … , m . Clearly from (1) , we have
∂ z k 2 ( F ) = f k ∂ z k ( F ) + F ∂ z k ( f k ) = f k 2 F + F ∂ z k ( f k ) = ( f k 2 + ∂ z k ( f k ) ) F , \partial_{z_k}^{2}(F)=f_k\partial_{z_k}(F)+F\partial_{z_k}(f_k)=f_k^{2}F+F\partial_{z_k}(f_k)=\bigl(f_k^{2}+\partial_{z_k}(f_k)\bigr)F, ∂ z k 2 ( F ) = f k ∂ z k ( F ) + F ∂ z k ( f k ) = f k 2 F + F ∂ z k ( f k ) = ( f k 2 + ∂ z k ( f k ) ) F ,
i.e.,
∂ z k 2 ( F ) F = f k 2 + ∂ z k ( f k ) . \frac{\partial_{z_k}^{2}(F)}{F}=f_k^{2}+\partial_{z_k}(f_k). F ∂ z k 2 ( F ) = f k 2 + ∂ z k ( f k ) .
We inductively deduce (see [11 Lemma 3.5] ) that
∂ z k l ( F ) F = f k l + l ( l − 1 ) 2 f k l − 2 ∂ z k 2 ( f k ) + A ~ l f k l − 3 ∂ z k 3 ( f k ) + B ~ l f k l − 4 ( ∂ z k ( f k ) ) 2 + P l − 3 ( f k ) , \frac{\partial_{z_k}^{l}(F)}{F}=f_k^{l}+\frac{l(l-1)}{2}f_k^{l-2}\partial_{z_k}^{2}(f_k)+\widetilde{A}_l f_k^{l-3}\partial_{z_k}^{3}(f_k)+\widetilde{B}_l f_k^{l-4}\bigl(\partial_{z_k}(f_k)\bigr)^2+P_{l-3}(f_k), F ∂ z k l ( F ) = f k l + 2 l ( l − 1 ) f k l − 2 ∂ z k 2 ( f k ) + A l f k l − 3 ∂ z k 3 ( f k ) + B l f k l − 4 ( ∂ z k ( f k ) ) 2 + P l − 3 ( f k ) ,
where
A ~ l = l ( l − 1 ) ( l − 2 ) 6 , B ~ l = l ( l − 1 ) ( l − 2 ) ( l − 3 ) 8 , \widetilde{A}_l=\frac{l(l-1)(l-2)}{6},\qquad\widetilde{B}_l=\frac{l(l-1)(l-2)(l-3)}{8}, A l = 6 l ( l − 1 ) ( l − 2 ) , B l = 8 l ( l − 1 ) ( l − 2 ) ( l − 3 ) ,
and P l − 3 ( f k ) P_{l-3}(f_k) P l − 3 ( f k ) is a differential polynomial with constant coefficients, which vanishes identically for l ≤ 3 l \le3 l ≤ 3 and has degree l − 3 l-3 l − 3 when l > 3 l>3 l > 3 .
Lemma 2.4. Let f f f and g g g be two non-constant finite order meromorphic function in C m \mathbb{C}^{m} C m such that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM. Suppose ∂ z i ( f ) \partial_{z_i}(f) ∂ z i ( f ) and ∂ z i ( g ) \partial_{z_i}(g) ∂ z i ( g ) share 0 0 0 CM and for one n > 1 n>1 n > 1 , ∂ z i n ( f ) \partial_{z_i}^{n}(f) ∂ z i n ( f ) and ∂ z i n ( g ) \partial_{z_i}^{n}(g) ∂ z i n ( g ) share 0 0 0 CM, where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . If either ∂ z i n ( f ) ≡ 0 \partial_{z_i}^{n}(f)\equiv0 ∂ z i n ( f ) ≡ 0 or ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is a rational function for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m , then we have either f ≡ c g f\equiv cg f ≡ c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } or
f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a and g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b , f(z)=e^{a_1z_1+a_2z_2+\cdots+a_mz_m+a}\quad\text{and}\quad g(z)=e^{b_1z_1+b_2z_2+\cdots+b_mz_m+b}, f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a and g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b , where a , a 1 , … , a m , b 1 , … , b m , b ∈ C a,a_1,\ldots,a_m,b_1,\ldots,b_m,b\in\mathbb{C} a , a 1 , … , a m , b 1 , … , b m , b ∈ C such that ( a 1 , … , a m ) ≠ ( 0 , … , 0 ) (a_1,\ldots,a_m)\ne(0,\ldots,0) ( a 1 , … , a m ) = ( 0 , … , 0 ) and ( b 1 , … , b m ) ≠ ( 0 , … , 0 ) (b_1,\ldots,b_m)\ne(0,\ldots,0) ( b 1 , … , b m ) = ( 0 , … , 0 ) .
Proof. By the given conditions, we have f f f and g g g share 0 0 0 and ∞ \infty ∞ CM, ∂ z i ( f ) \partial_{z_i}(f) ∂ z i ( f ) and ∂ z i ( g ) \partial_{z_i}(g) ∂ z i ( g ) share 0 0 0 CM and ∂ z i n ( f ) \partial_{z_i}^{n}(f) ∂ z i n ( f ) and ∂ z i n ( g ) \partial_{z_i}^{n}(g) ∂ z i n ( g ) share 0 0 0 CM, where n > 1 n>1 n > 1 and i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m .
Now we consider following two cases.
Case 1. Suppose ∂ z i n ( f ( z ) ) ≡ 0 \partial_{z_i}^{n}(f(z))\equiv0 ∂ z i n ( f ( z )) ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Since ∂ z i n ( f ) \partial_{z_i}^{n}(f) ∂ z i n ( f ) and ∂ z i n ( g ) \partial_{z_i}^{n}(g) ∂ z i n ( g ) share 0 0 0 CM, it follows that ∂ z i n ( g ( z ) ) ≡ 0 \partial_{z_i}^{n}(g(z))\equiv0 ∂ z i n ( g ( z )) ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Therefore we have
∂ n f ( z ) ∂ z i n ≡ 0 and ∂ n g ( z ) ∂ z i n ≡ 0 , (2) \frac{\partial^{n}f(z)}{\partial z_i^{n}}\equiv0\quad\text{and}\quad\frac{\partial^{n}g(z)}{\partial z_i^{n}}\equiv0, \tag*{(2)} ∂ z i n ∂ n f ( z ) ≡ 0 and ∂ z i n ∂ n g ( z ) ≡ 0 , (2)
for i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m .
We now want to prove that f ( z ) f(z) f ( z ) is a polynomial in C m \mathbb{C}^{m} C m . Here we prove that f ( z ) f(z) f ( z ) is a polynomial in C m \mathbb{C}^{m} C m by induction on the number m m m of variables. If m = 1 m=1 m = 1 , from (2) , we get that f ( z ) f(z) f ( z ) is a polynomial in C \mathbb{C} C . Let us suppose m = 2 m=2 m = 2 . Since
∂ n f ( z 1 , z 2 ) ∂ z 2 n ≡ 0 , \frac{\partial^{n}f(z_1,z_2)}{\partial z_2^{n}}\equiv0, ∂ z 2 n ∂ n f ( z 1 , z 2 ) ≡ 0 ,
on integration, we have
f ( z 1 , z 2 ) = ϕ 1 ( z 1 ) z 2 n − 1 + ϕ 2 ( z 1 ) z 2 n − 2 + ⋯ + ϕ n ( z 1 ) , (3) f(z_1,z_2)=\phi_1(z_1)z_2^{n-1}+\phi_2(z_1)z_2^{n-2}+\cdots+\phi_n(z_1), \tag*{(3)} f ( z 1 , z 2 ) = ϕ 1 ( z 1 ) z 2 n − 1 + ϕ 2 ( z 1 ) z 2 n − 2 + ⋯ + ϕ n ( z 1 ) , (3)
where ϕ i ( z 1 ) \phi_i(z_1) ϕ i ( z 1 ) ’s are entire functions in C \mathbb{C} C in the variable z 1 z_1 z 1 . Note that ∂ n f ( z 1 , z 2 ) ∂ z 1 n ≡ 0 \frac{\partial^{n}f(z_1,z_2)}{\partial z_1^{n}}\equiv0 ∂ z 1 n ∂ n f ( z 1 , z 2 ) ≡ 0 and so (3) gives
ϕ 1 ( n ) ( z 1 ) z 2 n − 1 + ϕ 2 ( n ) ( z 1 ) z 2 n − 2 + ⋯ + ϕ n ( n ) ( z 1 ) ≡ 0 , \phi_1^{(n)}(z_1)z_2^{n-1}+\phi_2^{(n)}(z_1)z_2^{n-2}+\cdots+\phi_n^{(n)}(z_1)\equiv0, ϕ 1 ( n ) ( z 1 ) z 2 n − 1 + ϕ 2 ( n ) ( z 1 ) z 2 n − 2 + ⋯ + ϕ n ( n ) ( z 1 ) ≡ 0 ,
which shows that ϕ i ( n ) ( z 1 ) ≡ 0 \phi_i^{(n)}(z_1)\equiv0 ϕ i ( n ) ( z 1 ) ≡ 0 for i = 1 , 2 , … , n i=1,2,\ldots,n i = 1 , 2 , … , n . Therefore on integration, we get
ϕ i ( z 1 ) = c n − 1 , i z 1 n − 1 + c n − 2 , i z 1 n − 2 + c 0 , i , \phi_i(z_1)=c_{n-1,i}z_1^{n-1}+c_{n-2,i}z_1^{n-2}+c_{0,i}, ϕ i ( z 1 ) = c n − 1 , i z 1 n − 1 + c n − 2 , i z 1 n − 2 + c 0 , i ,
where c i , j c_{i,j} c i , j (0 ≤ i ≤ n − 1 0\le i\le n-1 0 ≤ i ≤ n − 1 , j = 1 , 2 , … , n j=1,2,\ldots,n j = 1 , 2 , … , n ) are constants in C \mathbb{C} C . Then from (3) , we get
f ( z 1 , z 2 ) = ( c n − 1 , 1 z 1 n − 1 + c n − 2 , 1 z 1 n − 2 + c 0 , 1 ) z 2 n − 1 + ⋯ + ( c n − 1 , n z 1 n − 1 + c n − 2 , n z 1 n − 2 + c 0 , n ) , f(z_1,z_2)=(c_{n-1,1}z_1^{n-1}+c_{n-2,1}z_1^{n-2}+c_{0,1})z_2^{n-1}+\cdots+(c_{n-1,n}z_1^{n-1}+c_{n-2,n}z_1^{n-2}+c_{0,n}), f ( z 1 , z 2 ) = ( c n − 1 , 1 z 1 n − 1 + c n − 2 , 1 z 1 n − 2 + c 0 , 1 ) z 2 n − 1 + ⋯ + ( c n − 1 , n z 1 n − 1 + c n − 2 , n z 1 n − 2 + c 0 , n ) ,
which shows that f ( z 1 , z 2 ) f(z_1,z_2) f ( z 1 , z 2 ) is a polynomial in C 2 \mathbb{C}^{2} C 2 . Now we fix m ≥ 2 m\ge2 m ≥ 2 and assume that f ( z ) f(z) f ( z ) is a polynomial for variables of number at most m − 1 m-1 m − 1 . Since ∂ n f ( z ) ∂ z m n ≡ 0 \frac{\partial^{n}f(z)}{\partial z_m^{n}}\equiv0 ∂ z m n ∂ n f ( z ) ≡ 0 , we have
f ( z 1 , z 2 , … , z m ) = A 1 ( z 1 , z 2 , … , z m − 1 ) z m n − 1 + A 2 ( z 1 , z 2 , … , z m − 1 ) z m n − 2 + ⋯ + A n ( z 1 , z 2 , … , z m − 1 ) . (4) \begin{aligned}
f(z_1,z_2,\ldots,z_m)=A_1(z_1,z_2,\ldots,z_{m-1})z_m^{n-1}+A_2(z_1,z_2,\ldots,z_{m-1})z_m^{n-2}+\cdots\\
&+A_n(z_1,z_2,\ldots,z_{m-1}). \tag*{(4)}
\end{aligned} f ( z 1 , z 2 , … , z m ) = A 1 ( z 1 , z 2 , … , z m − 1 ) z m n − 1 + A 2 ( z 1 , z 2 , … , z m − 1 ) z m n − 2 + ⋯ + A n ( z 1 , z 2 , … , z m − 1 ) . (4)
Now using (2) to (4) , we get
∂ n A 1 ( z 1 , … , z m − 1 ) ∂ z i n z m n − 1 + ∂ 2 A 2 ( z 1 , … , z m − 1 ) ∂ z i 2 z m n − 2 + ⋯ + ∂ n A n ( z 1 , … , z m − 1 ) ∂ z i n ≡ 0 (5) \frac{\partial^{n}A_1(z_1,\ldots,z_{m-1})}{\partial z_i^{n}}z_m^{n-1}+\frac{\partial^{2}A_2(z_1,\ldots,z_{m-1})}{\partial z_i^{2}}z_m^{n-2}+\cdots+\frac{\partial^{n}A_n(z_1,\ldots,z_{m-1})}{\partial z_i^{n}}\equiv0 \tag*{(5)} ∂ z i n ∂ n A 1 ( z 1 , … , z m − 1 ) z m n − 1 + ∂ z i 2 ∂ 2 A 2 ( z 1 , … , z m − 1 ) z m n − 2 + ⋯ + ∂ z i n ∂ n A n ( z 1 , … , z m − 1 ) ≡ 0 (5)
for i = 1 , 2 , … , m − 1 i=1,2,\ldots,m-1 i = 1 , 2 , … , m − 1 . Therefore (5) yields
∂ n A j ( z 1 , z 2 , … , z m − 1 ) ∂ z i n ≡ 0 (6) \frac{\partial^n A_j(z_1,z_2,\ldots,z_{m-1})}{\partial z_i^n}\equiv0 \tag*{(6)} ∂ z i n ∂ n A j ( z 1 , z 2 , … , z m − 1 ) ≡ 0 (6)
for all j = 1 , 2 , … , n j=1,2,\ldots,n j = 1 , 2 , … , n , where i = 1 , 2 , … , m − 1 i=1,2,\ldots,m-1 i = 1 , 2 , … , m − 1 . Then by the induction assumptions, we find from (6) that A i ( z 1 , z 2 , … , z m − 1 ) A_i(z_1,z_2,\ldots,z_{m-1}) A i ( z 1 , z 2 , … , z m − 1 ) (∀ i = 1 , 2 , … , n \forall i=1,2,\ldots,n ∀ i = 1 , 2 , … , n ) are polynomials in the variables z 1 , z 2 , … , z m − 1 z_1,z_2,\ldots,z_{m-1} z 1 , z 2 , … , z m − 1 . Therefore from (4) , we see that f ( z 1 , z 2 , … , z m ) f(z_1,z_2,\ldots,z_m) f ( z 1 , z 2 , … , z m ) is a polynomial in z 1 , z 2 , … , z m z_1,z_2,\ldots,z_m z 1 , z 2 , … , z m . Similarly g ( z 1 , z 2 , … , z m ) g(z_1,z_2,\ldots,z_m) g ( z 1 , z 2 , … , z m ) is a polynomial in z 1 , z 2 , … , z m z_1,z_2,\ldots,z_m z 1 , z 2 , … , z m . Since f f f and g g g share 0 CM, it follows that f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } .
Case 2. Suppose ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is a rational function for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Since f f f and g g g share 0 and ∞ \infty ∞ CM, there exists a polynomial P P P in C m \mathbb{C}^m C m such that f = g e P f=ge^P f = g e P . Clearly, we have
∂ z i ( f ) f = ∂ z i ( g ) g + ∂ z i ( P ) , \frac{\partial_{z_i}(f)}{f}=\frac{\partial_{z_i}(g)}{g}+\partial_{z_i}(P), f ∂ z i ( f ) = g ∂ z i ( g ) + ∂ z i ( P ) ,
which shows that ∂ z i ( g ) / g \partial_{z_i}(g)/g ∂ z i ( g ) / g is also a rational function for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . By the given conditions, we have f f f and g g g share 0 and ∞ \infty ∞ CM and ∂ z i ( f ) \partial_{z_i}(f) ∂ z i ( f ) and ∂ z i ( g ) \partial_{z_i}(g) ∂ z i ( g ) share 0 CM for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Consequently ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f and ∂ z i ( g ) / g \partial_{z_i}(g)/g ∂ z i ( g ) / g share 0 and ∞ \infty ∞ CM. As a result, we have
∂ z i ( f ) f = A i ∂ z i ( g ) g , (7) \frac{\partial_{z_i}(f)}{f}=A_i\frac{\partial_{z_i}(g)}{g}, \tag*{(7)} f ∂ z i ( f ) = A i g ∂ z i ( g ) , (7)
where A i ∈ C ∖ { 0 } A_i\in\mathbb{C}\setminus\{0\} A i ∈ C ∖ { 0 } and i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . On integration, we get
f ( z ) = g A i ( z ) e g i ( z ) , (8) f(z)=g^{A_i}(z)e^{g_i(z)}, \tag*{(8)} f ( z ) = g A i ( z ) e g i ( z ) , (8)
where g i ( z ) = g ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) g_i(z)=g(z_1,z_2,\ldots,z_{i-1},z_{i+1},\ldots,z_m) g i ( z ) = g ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) is a polynomial for i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now we consider following two sub-cases.
Sub-case 2.1. Suppose 0 and ∞ \infty ∞ are not both Picard exceptional values of f f f . Since f f f and g g g share 0 and ∞ \infty ∞ CM, it follows that 0 and ∞ \infty ∞ are not both Picard exceptional values of g g g . Now from (8) , we deduce that A i = 1 A_i=1 A i = 1 , where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m and so from (6) and (8) , we get respectively
∂ z j ( f ) f = ∂ z j ( g ) g , (9) \frac{\partial_{z_j}(f)}{f}=\frac{\partial_{z_j}(g)}{g}, \tag*{(9)} f ∂ z j ( f ) = g ∂ z j ( g ) , (9)
where j = 1 , 2 , … , m j=1,2,\ldots,m j = 1 , 2 , … , m and
f ( z ) = g ( z ) e g i ( z ) , (10) f(z)=g(z)e^{g_i(z)}, \tag*{(10)} f ( z ) = g ( z ) e g i ( z ) , (10)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Taking partial derivative with respect to z j z_j z j on the both sides of (10) and using (9) , we get ∂ z j ( g i ( z ) ) ≡ 0 \partial_{z_j}(g_i(z))\equiv0 ∂ z j ( g i ( z )) ≡ 0 , where i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m such that i ≠ j i\ne j i = j . Since g i ( z ) g_i(z) g i ( z ) is independent of z i z_i z i , we conclude that g i ( z ) g_i(z) g i ( z ) is a constant. Now in view of (10) , we can assume that e g i ( z ) = c e^{g_i(z)}=c e g i ( z ) = c for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Therefore f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } .
Case 2.2. Suppose 0 and ∞ \infty ∞ are both Picard exceptional values of f f f . Since f f f and g g g share 0 and ∞ \infty ∞ CM, it follows that 0 and ∞ \infty ∞ are both Picard exceptional values of g g g . Consequently ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f and ∂ z i ( g ) / g \partial_{z_i}(g)/g ∂ z i ( g ) / g are polynomials in C m \mathbb{C}^m C m for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Let g = exp ( Q ) g=\exp(Q) g = exp ( Q ) , where Q Q Q is a polynomial in C m \mathbb{C}^m C m . Clearly
∂ z i ( g ) g = ∂ z i ( Q ) . (11) \frac{\partial_{z_i}(g)}{g}=\partial_{z_i}(Q). \tag*{(11)} g ∂ z i ( g ) = ∂ z i ( Q ) . (11)
Now from (7) , we get
∂ z i ( f ) f = A i ∂ z i ( Q ) (12) \frac{\partial_{z_i}(f)}{f}=A_i\partial_{z_i}(Q) \tag*{(12)} f ∂ z i ( f ) = A i ∂ z i ( Q ) (12)
and so on integration, we have
f ( z ) = e A i Q ( z ) + H i ( z ) , (13) f(z)=e^{A_iQ(z)+H_i(z)}, \tag*{(13)} f ( z ) = e A i Q ( z ) + H i ( z ) , (13)
where H i ( z ) = H ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) H_i(z)=H(z_1,z_2,\ldots,z_{i-1},z_{i+1},\ldots,z_m) H i ( z ) = H ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) is a polynomial, i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Note that (13) gives
∂ z j ( f ) f = A i ∂ z j ( Q ) + ∂ z j ( H i ) , (14) \frac{\partial_{z_j}(f)}{f}=A_i\partial_{z_j}(Q)+\partial_{z_j}(H_i), \tag*{(14)} f ∂ z j ( f ) = A i ∂ z j ( Q ) + ∂ z j ( H i ) , (14)
where i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j .
By Lemma 2.3 , we can say that ∂ z i n ( f ) / f \partial_{z_i}^{n}(f)/f ∂ z i n ( f ) / f and ∂ z i n ( g ) / g \partial_{z_i}^{n}(g)/g ∂ z i n ( g ) / g are rational functions for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Also by the given conditions, we have f f f and g g g share 0 0 0 and ∞ \infty ∞ CM and ∂ z i n ( f ) \partial_{z_i}^{n}(f) ∂ z i n ( f ) and ∂ z i n ( g ) \partial_{z_i}^{n}(g) ∂ z i n ( g ) share 0 0 0 CM for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Consequently ∂ z i n ( f ) / f \partial_{z_i}^{n}(f)/f ∂ z i n ( f ) / f and ∂ z i n ( g ) / g \partial_{z_i}^{n}(g)/g ∂ z i n ( g ) / g share 0 0 0 and ∞ \infty ∞ CM. As a result, we have
∂ z i n ( f ) f = B i ∂ z i n ( g ) g , (15) \frac{\partial_{z_i}^{n}(f)}{f}=B_i\frac{\partial_{z_i}^{n}(g)}{g}, \tag*{(15)} f ∂ z i n ( f ) = B i g ∂ z i n ( g ) , (15)
where B i ∈ C ∖ { 0 } B_i\in\mathbb{C}\setminus\{0\} B i ∈ C ∖ { 0 } , where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now in view of (11) and (12) and using Lemma 2.3 to (15) , we have
( A i ∂ z i ( Q ) ) n + n ( n − 1 ) 2 ( A i ∂ z i ( Q ) ) n − 2 ∂ z i ( A i ∂ z i ( Q ) ) + A ~ n ( A i ∂ z i ( Q ) ) n − 3 ∂ z i 2 ( A i ∂ z i ( Q ) ) + B ~ n ( A i ∂ z i ( Q ) ) n − 4 ( ∂ z i ( A i ∂ z i ( Q ) ) ) 2 + P n − 3 ( A i ∂ z i ( Q ) ) = B i ( ( ∂ z i ( Q ) ) n + n ( n − 1 ) 2 ( ∂ z i ( Q ) ) n − 2 ∂ z i ( ∂ z i ( Q ) ) + A ~ n ( ∂ z i ( Q ) ) n − 3 ∂ z i 2 ( ∂ z i ( Q ) ) + B ~ n ( ∂ z i ( Q ) ) n − 4 ( ∂ z i ( ∂ z i ( Q ) ) ) 2 + P n − 3 ( ∂ z i ( Q ) ) ) (16) \begin{aligned}
&(A_i\partial_{z_i}(Q))^n+\frac{n(n-1)}{2}(A_i\partial_{z_i}(Q))^{n-2}\partial_{z_i}(A_i\partial_{z_i}(Q))+\widetilde{A}_n(A_i\partial_{z_i}(Q))^{n-3}\partial_{z_i}^2(A_i\partial_{z_i}(Q))\\
&\quad+\widetilde{B}_n(A_i\partial_{z_i}(Q))^{n-4}(\partial_{z_i}(A_i\partial_{z_i}(Q)))^2+P_{n-3}(A_i\partial_{z_i}(Q))\\
&=B_i\left((\partial_{z_i}(Q))^n+\frac{n(n-1)}{2}(\partial_{z_i}(Q))^{n-2}\partial_{z_i}(\partial_{z_i}(Q))+\widetilde{A}_n(\partial_{z_i}(Q))^{n-3}\partial_{z_i}^2(\partial_{z_i}(Q))\right.\\
&\qquad\left.+\widetilde{B}_n(\partial_{z_i}(Q))^{n-4}(\partial_{z_i}(\partial_{z_i}(Q)))^2+P_{n-3}(\partial_{z_i}(Q))\right) \tag*{(16)}
\end{aligned} ( A i ∂ z i ( Q ) ) n + 2 n ( n − 1 ) ( A i ∂ z i ( Q ) ) n − 2 ∂ z i ( A i ∂ z i ( Q )) + A n ( A i ∂ z i ( Q ) ) n − 3 ∂ z i 2 ( A i ∂ z i ( Q )) + B n ( A i ∂ z i ( Q ) ) n − 4 ( ∂ z i ( A i ∂ z i ( Q )) ) 2 + P n − 3 ( A i ∂ z i ( Q )) = B i ( ( ∂ z i ( Q ) ) n + 2 n ( n − 1 ) ( ∂ z i ( Q ) ) n − 2 ∂ z i ( ∂ z i ( Q )) + A n ( ∂ z i ( Q ) ) n − 3 ∂ z i 2 ( ∂ z i ( Q )) + B n ( ∂ z i ( Q ) ) n − 4 ( ∂ z i ( ∂ z i ( Q )) ) 2 + P n − 3 ( ∂ z i ( Q )) ) (16)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m ,
A ~ n = n ( n − 1 ) ( n − 2 ) 6 , B ~ n = n ( n − 1 ) ( n − 2 ) ( n − 3 ) 8 , \widetilde{A}_n=\frac{n(n-1)(n-2)}{6},\qquad\widetilde{B}_n=\frac{n(n-1)(n-2)(n-3)}{8}, A n = 6 n ( n − 1 ) ( n − 2 ) , B n = 8 n ( n − 1 ) ( n − 2 ) ( n − 3 ) ,
and P n − 3 ( ∂ z i ( P ) ) P_{n-3}(\partial_{z_i}(P)) P n − 3 ( ∂ z i ( P )) is a differential polynomial with constant coefficients, which vanishes identically for n ≤ 3 n\le3 n ≤ 3 and has degree n − 3 n-3 n − 3 when n > 3 n>3 n > 3 .
We consider following two sub-cases.
Sub-case 2.2.1. Suppose deg ( ∂ z i ( Q ) ) ≥ 1 \deg(\partial_{z_i}(Q))\ge1 deg ( ∂ z i ( Q )) ≥ 1 for all i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Now from (16) , we deduce that A i n = B i A_i^n=B_i A i n = B i and A i n − 1 = B i A_i^{n-1}=B_i A i n − 1 = B i for all i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Clearly A i = 1 A_i=1 A i = 1 for all i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Now proceeding in the same way as done in the proof of Sub-case 2.1, we easily get f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } .
Sub-case 2.2.2. Suppose deg ( ∂ z i ( Q ) ) < 1 \deg(\partial_{z_i}(Q))<1 deg ( ∂ z i ( Q )) < 1 for at least one i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . We know that ∂ z j ( f ) / f \partial_{z_j}(f)/f ∂ z j ( f ) / f and ∂ j ( g ) / g \partial_j(g)/g ∂ j ( g ) / g share 0 0 0 CM, where i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j . Therefore from (11) and (14) , we see that ∂ z j ( Q ) \partial_{z_j}(Q) ∂ z j ( Q ) and A i ∂ z j ( Q ) + ∂ z j ( H i ) A_i\partial_{z_j}(Q)+\partial_{z_j}(H_i) A i ∂ z j ( Q ) + ∂ z j ( H i ) share 0 0 0 CM, where i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j . Consequently ∂ z j ( Q ) \partial_{z_j}(Q) ∂ z j ( Q ) and ∂ z j ( H i ) \partial_{z_j}(H_i) ∂ z j ( H i ) share 0 0 0 CM, where i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j . Therefore, we can write
C j ∂ z j ( Q ) ≡ A i ∂ z j ( Q ) + ∂ z j ( H i ) (17) C_j\partial_{z_j}(Q)\equiv A_i\partial_{z_j}(Q)+\partial_{z_j}(H_i) \tag*{(17)} C j ∂ z j ( Q ) ≡ A i ∂ z j ( Q ) + ∂ z j ( H i ) (17)
and
D j ∂ z j ( Q ) ≡ ∂ z j ( H i ) , (18) D_j\partial_{z_j}(Q)\equiv\partial_{z_j}(H_i), \tag*{(18)} D j ∂ z j ( Q ) ≡ ∂ z j ( H i ) , (18)
where C j , D ∈ C ∖ { 0 } C_j,D\in\mathbb{C}\setminus\{0\} C j , D ∈ C ∖ { 0 } and i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j .
First suppose C j = A i C_j=A_i C j = A i for all i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j . Then from (17) , we get ∂ z j ( H i ) ≡ 0 \partial_{z_j}(H_i)\equiv0 ∂ z j ( H i ) ≡ 0 , where i , j ∈ { 1 , 2 , … , m } i,j\in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i\ne j i = j . Clearly H i ( z ) H_i(z) H i ( z ) is a constant, say $D_i, where i ∈ { 1 , 2 , … , m } i \in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Also from (18) , we get ∂ z j ( Q ) ≡ 0 \partial_{z_j}(Q) \equiv0 ∂ z j ( Q ) ≡ 0 , where i ∈ { 1 , 2 , … , m } i \in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . This shows that Q ( z ) Q(z) Q ( z ) is a polynomial in C m \mathbb{C}^{m} C m of degree one. Therefore from (13) , we have
f = e A i Q + D i , (19) f=e^{A_iQ+D_i}, \tag*{(19)} f = e A i Q + D i , (19)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now from (19) , we get
e ( A i − A j ) Q + D i − D j = 1 , (20) e^{(A_i-A_j)Q+D_i-D_j}=1, \tag*{(20)} e ( A i − A j ) Q + D i − D j = 1 , (20)
where i , j ∈ { 1 , 2 , … , m } i,j \in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i \ne j i = j . Therefore from (20) , one can easily deduce that A 1 = A 2 = ⋯ = A m A_1=A_2=\cdots=A_m A 1 = A 2 = ⋯ = A m and D 1 = D 2 = ⋯ = D m D_1=D_2=\cdots=D_m D 1 = D 2 = ⋯ = D m . Consequently, we can take f = e A Q + B f=e^{AQ+B} f = e A Q + B , where A ∈ C ∖ { 0 } A \in\mathbb{C}\setminus\{0\} A ∈ C ∖ { 0 } and B ∈ C B \in\mathbb{C} B ∈ C . Finally, we have
f ( z ) = e A ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) + a and g ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + b , f(z)=e^{A(a_1z_1+a_2z_2+\cdots+a_mz_m)+a}\quad\text{and}\quad g(z)=e^{a_1z_1+a_2z_2+\cdots+a_mz_m+b}, f ( z ) = e A ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) + a and g ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + b ,
where a , A , a 1 , a 2 , … , a m , b ∈ C a,A,a_1,a_2,\ldots,a_m,b \in\mathbb{C} a , A , a 1 , a 2 , … , a m , b ∈ C such that A ≠ 0 A \ne0 A = 0 and ( a 1 , a 2 , … , a m ) ≠ ( 0 , 0 , … , 0 ) (a_1,a_2,\ldots,a_m)\ne(0,0,\ldots,0) ( a 1 , a 2 , … , a m ) = ( 0 , 0 , … , 0 ) .
Next suppose C j ≠ A i C_j \ne A_i C j = A i for at least one set i , j ∈ { 1 , 2 , … , m } i,j \in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i \ne j i = j . Without loss of generality, we may assume that C j ≠ A i C_j \ne A_i C j = A i for all i , j ∈ { 1 , 2 , … , m } i,j \in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i \ne j i = j . Then from (17) , we get
( C j − A i ) ∂ z j ( Q ) = ∂ z j ( H i ) , (21) (C_j-A_i)\partial_{z_j}(Q)=\partial_{z_j}(H_i), \tag*{(21)} ( C j − A i ) ∂ z j ( Q ) = ∂ z j ( H i ) , (21)
where i , j ∈ { 1 , 2 , … , m } i,j \in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i \ne j i = j . Note that ∂ z j ( H i ) \partial_{z_j}(H_i) ∂ z j ( H i ) is independent of z i z_i z i , where i , j ∈ { 1 , 2 , … , m } i,j \in\{1,2,\ldots,m\} i , j ∈ { 1 , 2 , … , m } such that i ≠ j i \ne j i = j . Therefore from (21) , we can say that ∂ z j ( Q ) \partial_{z_j}(Q) ∂ z j ( Q ) is independent of z j z_j z j , where j ∈ { 1 , 2 , … , m } j \in\{1,2,\ldots,m\} j ∈ { 1 , 2 , … , m } . Since Q Q Q is a polynomial in C m \mathbb{C}^{m} C m , one can easily deduce that Q Q Q is a polynomial of degree one. Therefore from (21) , it follows that H i ( z ) = H ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) H_i(z)=H(z_1,z_2,\ldots,z_{i-1},z_{i+1},\ldots,z_m) H i ( z ) = H ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) is also a polynomial of degree one, where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Clearly from (17) , we conclude that deg ( A i Q + H i ) = 1 \deg(A_iQ+H_i)=1 deg ( A i Q + H i ) = 1 , where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Finally, we can take
f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a and g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b , f(z)=e^{a_1z_1+a_2z_2+\cdots+a_mz_m+a}\quad\text{and}\quad g(z)=e^{b_1z_1+b_2z_2+\cdots+b_mz_m+b}, f ( z ) = e a 1 z 1 + a 2 z 2 + ⋯ + a m z m + a and g ( z ) = e b 1 z 1 + b 2 z 2 + ⋯ + b m z m + b ,
where a , a 1 , … , a m , b 1 , … , b m , b ∈ C a,a_1,\ldots,a_m,b_1,\ldots,b_m,b \in\mathbb{C} a , a 1 , … , a m , b 1 , … , b m , b ∈ C such that ( a 1 , … , a m ) ≠ ( 0 , … , 0 ) (a_1,\ldots,a_m)\ne(0,\ldots,0) ( a 1 , … , a m ) = ( 0 , … , 0 ) and ( b 1 , … , b m ) ≠ ( 0 , … , 0 ) (b_1,\ldots,b_m)\ne(0,\ldots,0) ( b 1 , … , b m ) = ( 0 , … , 0 ) .
Hence the proof is complete. □
Lemma 2.5. Let f f f and g g g be two non-constant finite order meromorphic function in C m \mathbb{C}^{m} C m such that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM. If Φ 0 , i ≡ 0 \Phi_{0,i}\equiv0 Φ 0 , i ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m , where Φ 0 , i \Phi_{0,i} Φ 0 , i is defined by
Φ 0 , i = ∂ z i ( f ) f − ∂ z i ( g ) g , (22) \Phi_{0,i}=\frac{\partial_{z_i}(f)}{f}-\frac{\partial_{z_i}(g)}{g}, \tag*{(22)} Φ 0 , i = f ∂ z i ( f ) − g ∂ z i ( g ) , (22) then f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } .
Proof. By the given conditions, we know that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM and so f = g e P f=ge^{P} f = g e P , where P P P is an entire function in C m \mathbb{C}^{m} C m . Since ρ ( f ) < + ∞ \rho(f)<+\infty ρ ( f ) < + ∞ and ρ ( g ) < + ∞ \rho(g)<+\infty ρ ( g ) < + ∞ , using Lemma 2.2, we can say that P P P is a polynomial in C m \mathbb{C}^{m} C m . Now using f = g e P f=ge^{P} f = g e P to (22) , we have Φ 0 , i = ∂ z i ( P ) \Phi_{0,i}=\partial_{z_i}(P) Φ 0 , i = ∂ z i ( P ) . Since Φ 0 , i ≡ 0 \Phi_{0,i}\equiv0 Φ 0 , i ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m , we get ∂ z i ( P ) ≡ 0 \partial_{z_i}(P)\equiv0 ∂ z i ( P ) ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m and so P P P is a constant. Therefore f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } . □
Lemma 2.6. Let f f f and g g g be two non-constant finite order meromorphic function in C m \mathbb{C}^{m} C m such that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM. If Φ 1 , i j ≡ 0 \Phi_{1,ij}\equiv0 Φ 1 , ij ≡ 0 for all i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m , where Φ 1 , i j \Phi_{1,ij} Φ 1 , ij is defined by
Φ 1 , i j = ∂ z j 2 z i ( f ) ∂ z i ( f ) − ∂ z j 2 z i ( g ) ∂ z i ( g ) , (23) \Phi_{1,ij}=\frac{\partial_{z_j}^{2}z_i(f)}{\partial_{z_i}(f)}-\frac{\partial_{z_j}^{2}z_i(g)}{\partial_{z_i}(g)}, \tag*{(23)} Φ 1 , ij = ∂ z i ( f ) ∂ z j 2 z i ( f ) − ∂ z i ( g ) ∂ z j 2 z i ( g ) , (23) then either f = c g f=cg f = c g , where c ∈ C ∖ { 0 } c\in\mathbb{C}\setminus\{0\} c ∈ C ∖ { 0 } or
f = d 1 − c e − P and g = d e P − c , f=\frac{d}{1-ce^{-P}}\quad\text{and}\quad g=\frac{d}{e^{P}-c}, f = 1 − c e − P d and g = e P − c d , энд_mid=1
where c , d ∈ C \ { 0 } c,d \in\mathbb{C}\backslash\{0\} c , d ∈ C \ { 0 } and P P P is a non-constant polynomial in C m \mathbb{C}^{m} C m .
Proof. By the given conditions, we know that f f f and g g g share 0 0 0 and ∞ \infty ∞ CM and so f = g e P f=ge^{P} f = g e P , where P P P is a polynomial in C m \mathbb{C}^{m} C m . If P P P is a constant, then we have f = c g f=cg f = c g , where c ∈ C \ { 0 } c\in\mathbb{C}\backslash\{0\} c ∈ C \ { 0 } . Hence we assume that P P P is a non-constant polynomial. Note that Φ 1 , i j ≡ 0 \Phi_{1,ij}\equiv0 Φ 1 , ij ≡ 0 for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Then from (23) , we get
∂ z j z i 2 ( f ) ∂ z i ( g ) − ∂ z j z i 2 ( g ) ∂ z i ( f ) ≡ 0. \partial_{z_jz_i}^{2}(f)\partial_{z_i}(g)-\partial_{z_jz_i}^{2}(g)\partial_{z_i}(f)\equiv0. ∂ z j z i 2 ( f ) ∂ z i ( g ) − ∂ z j z i 2 ( g ) ∂ z i ( f ) ≡ 0.
Let F i = ∂ z i ( f ) F_i=\partial_{z_i}(f) F i = ∂ z i ( f ) and G i = ∂ z i ( g ) G_i=\partial_{z_i}(g) G i = ∂ z i ( g ) . If possible, suppose F F F and G G G are linearly independent. Then by Corollary 1.40 [14 ] , there exists j ∈ Z [ 1 , m ] j\in\mathbb{Z}[1,m] j ∈ Z [ 1 , m ] such that
W ( F i , G i ) = ∣ F i G i ∂ z j ( F i ) ∂ z j ( G i ) ∣ ≢ 0 W(F_i,G_i)=\begin{vmatrix} F_i & G_i \\ \partial_{z_j}(F_i) & \partial_{z_j}(G_i) \end{vmatrix}\not\equiv0 W ( F i , G i ) = F i ∂ z j ( F i ) G i ∂ z j ( G i ) ≡ 0
and so ∂ z j z i 2 ( f ) ∂ z i ( g ) − ∂ z j z i 2 ( g ) ∂ z i ( f ) ≢ 0 \partial_{z_jz_i}^{2}(f)\partial_{z_i}(g)-\partial_{z_jz_i}^{2}(g)\partial_{z_i}(f)\not\equiv0 ∂ z j z i 2 ( f ) ∂ z i ( g ) − ∂ z j z i 2 ( g ) ∂ z i ( f ) ≡ 0 , which is absurd. Hence F i F_i F i and G i G_i G i are linearly dependent and so
∂ z i ( f ) = c i ∂ z i ( g ) , (24) \partial_{z_i}(f)=c_i\partial_{z_i}(g), \tag*{(24)} ∂ z i ( f ) = c i ∂ z i ( g ) , (24)
where c i ∈ C \ { 0 } c_i\in\mathbb{C}\backslash\{0\} c i ∈ C \ { 0 } and i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Since f = g e P f=ge^{P} f = g e P , we have
∂ z i ( f ) = ( ∂ z i ( g ) + g ∂ z i ( P ) ) e P , (25) \partial_{z_i}(f)=\left(\partial_{z_i}(g)+g\partial_{z_i}(P)\right)e^{P}, \tag*{(25)} ∂ z i ( f ) = ( ∂ z i ( g ) + g ∂ z i ( P ) ) e P , (25)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now using (24) and (25) , we have
∂ z i ( g ) g = ∂ z i ( P ) e P c i − e P and ∂ z i ( f ) f = c i ∂ z i ( P ) e − P c i e − P − 1 , (26) \frac{\partial_{z_i}(g)}{g}=\frac{\partial_{z_i}(P)e^{P}}{c_i-e^{P}}\quad\text{and}\quad\frac{\partial_{z_i}(f)}{f}=\frac{c_i\partial_{z_i}(P)e^{-P}}{c_ie^{-P}-1}, \tag*{(26)} g ∂ z i ( g ) = c i − e P ∂ z i ( P ) e P and f ∂ z i ( f ) = c i e − P − 1 c i ∂ z i ( P ) e − P , (26)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . On integration, we get
g ( z ) = e R 1 i ( z ) c i − e P ( z ) and f ( z ) = e R 2 i ( z ) c i e − P − 1 , (27) g(z)=\frac{e^{R_{1i}(z)}}{c_i-e^{P(z)}}\quad\text{and}\quad f(z)=\frac{e^{R_{2i}(z)}}{c_ie^{-P}-1}, \tag*{(27)} g ( z ) = c i − e P ( z ) e R 1 i ( z ) and f ( z ) = c i e − P − 1 e R 2 i ( z ) , (27)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m and R j i ( z ) = R j ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) R_{ji}(z)=R_j(z_1,z_2,\ldots,z_{i-1},z_{i+1},\ldots,z_m) R j i ( z ) = R j ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) is a polynomial. It is easy to get from (27) that c 1 = c 2 = … = c m = c c_1=c_2=\ldots=c_m=c c 1 = c 2 = … = c m = c and so
f ( z ) = e R 2 i ( z ) c e − P − 1 and g ( z ) = e R 1 i ( z ) c − e P ( z ) , (28) f(z)=\frac{e^{R_{2i}(z)}}{ce^{-P}-1}\quad\text{and}\quad g(z)=\frac{e^{R_{1i}(z)}}{c-e^{P(z)}}, \tag*{(28)} f ( z ) = c e − P − 1 e R 2 i ( z ) and g ( z ) = c − e P ( z ) e R 1 i ( z ) , (28)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now from (28) , we deduce that R 2 i − R 2 j R_{2i}-R_{2j} R 2 i − R 2 j and R 1 i − R 1 j R_{1i}-R_{1j} R 1 i − R 1 j are constants for i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m and i ≠ j i\ne j i = j . Clearly R 11 − R 12 , R 11 − R 13 , … , R 11 − R 1 m R_{11}-R_{12}, R_{11}-R_{13},\ldots,R_{11}-R_{1m} R 11 − R 12 , R 11 − R 13 , … , R 11 − R 1 m are constants. So ∂ z 2 ( R 11 ) ≡ 0 \partial_{z_2}(R_{11})\equiv0 ∂ z 2 ( R 11 ) ≡ 0 , ∂ z 3 ( R 11 ) ≡ 0 \partial_{z_3}(R_{11})\equiv0 ∂ z 3 ( R 11 ) ≡ 0 , … \ldots … , ∂ z m ( R 11 ) ≡ 0 \partial_{z_m}(R_{11})\equiv0 ∂ z m ( R 11 ) ≡ 0 . Consequently R 11 R_{11} R 11 is a constant. Similarly R 12 , R 13 , … , R 1 m R_{12},R_{13},\ldots,R_{1m} R 12 , R 13 , … , R 1 m are also constants. In the like manner, we can prove that R 21 , R 12 , … , R 2 m R_{21},R_{12},\ldots,R_{2m} R 21 , R 12 , … , R 2 m are constants. Finally from (28) , we may assume that
f ( z ) = a c e − P − 1 and g ( z ) = b c − e P ( z ) , f(z)=\frac{a}{ce^{-P}-1}\quad\text{and}\quad g(z)=\frac{b}{c-e^{P(z)}}, f ( z ) = c e − P − 1 a and g ( z ) = c − e P ( z ) b ,
where a , b ∈ C \ { 0 } a,b\in\mathbb{C}\backslash\{0\} a , b ∈ C \ { 0 } . Hence the proof is complete. □
Proof of Theorem 1.1
Proof. By the given conditions, we have
f = g e P , ∂ z i ( f ) = ∂ z i ( g ) e Q i and ∂ z i n ( f ) = ∂ z i n ( g ) e R i , (29) f=ge^{P},\quad\partial_{z_i}(f)=\partial_{z_i}(g)e^{Q_i}\quad\text{and}\quad\partial_{z_i}^{n}(f)=\partial_{z_i}^{n}(g)e^{R_i}, \tag*{(29)} f = g e P , ∂ z i ( f ) = ∂ z i ( g ) e Q i and ∂ z i n ( f ) = ∂ z i n ( g ) e R i , (29)
where P P P , Q i Q_i Q i (i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m ) and R i R_i R i (i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m ) are polynomials in C m \mathbb{C}^{m} C m . Here we consider the auxiliary functions Φ 0 , i \Phi_{0,i} Φ 0 , i and Φ 1 , i j \Phi_{1,ij} Φ 1 , ij defined as in (22) and (23) respectively. If Φ 0 , i ≡ 0 \Phi_{0,i}\equiv0 Φ 0 , i ≡ 0 for i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m , then by Lemma 2.5, we have f = c g f=cg f = c g , where c ∈ C \ { 0 } c\in\mathbb{C}\backslash\{0\} c ∈ C \ { 0 } . Henceforth we assume that Φ 0 , i ≢ 0 \Phi_{0,i}\not\equiv0 Φ 0 , i ≡ 0 for at least one i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . If Φ 1 , i j ≡ 0 \Phi_{1,ij}\equiv0 Φ 1 , ij ≡ 0 for i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m , then by Lemma 2.6 , we have either f = c g f=cg f = c g , where c ∈ C \ { 0 } c\in\mathbb{C}\backslash\{0\} c ∈ C \ { 0 } or
f = d 1 − c e − P and g = d e P − c , f=\frac{d}{1-ce^{-P}}\quad\text{and}\quad g=\frac{d}{e^P-c}, f = 1 − c e − P d and g = e P − c d ,
where c , d ∈ C \ { 0 } c,d\in\mathbb{C}\backslash\{0\} c , d ∈ C \ { 0 } and P P P is a non-constant polynomial in C m \mathbb{C}^m C m . Henceforth, we assume that Φ 1 , i j ≢ 0 \Phi_{1,ij}\not\equiv0 Φ 1 , ij ≡ 0 for at least one ( i , j ) ∈ { i = 1 , 2 , … , m } × { 1 , 2 , … , m } (i,j)\in\{i=1,2,\ldots,m\}\times\{1,2,\ldots,m\} ( i , j ) ∈ { i = 1 , 2 , … , m } × { 1 , 2 , … , m } .
Define I = I f ∪ I g I=I_f\cup I_g I = I f ∪ I g and S = ⋃ a ∈ { 0 , ∞ } ( supp μ f a ) s ∪ ( supp μ g a ) s S=\bigcup_{a\in\{0,\infty\}}(\operatorname{supp}\mu_f^a)_s\cup(\operatorname{supp}\mu_g^a)_s S = ⋃ a ∈ { 0 , ∞ } ( supp μ f a ) s ∪ ( supp μ g a ) s , where A s A_s A s denotes the set of singular points of the analytic set A A A . Note that dim ( I ∪ S ) ≤ m − 2 \dim(I\cup S)\leq m-2 dim ( I ∪ S ) ≤ m − 2 . Take z 0 ∈ C m − ( I ∪ S ) z_0\in\mathbb{C}^m-(I\cup S) z 0 ∈ C m − ( I ∪ S ) .
Suppose p = μ f 0 ( z 0 ) > 0 p=\mu_f^0(z_0)>0 p = μ f 0 ( z 0 ) > 0 . By the given condition, we have p = μ g 0 ( z 0 ) > 0 p=\mu_g^0(z_0)>0 p = μ g 0 ( z 0 ) > 0 . Note that f − 1 ( { 0 } ) s f^{-1}(\{0\})_s f − 1 ({ 0 } ) s is an analytic subset of C m \mathbb{C}^m C m of dimension ≤ m − 2 \leq m-2 ≤ m − 2 , where f − 1 ( { 0 } ) s f^{-1}(\{0\})_s f − 1 ({ 0 } ) s is the set of singular points of f − 1 ( { 0 } ) f^{-1}(\{0\}) f − 1 ({ 0 }) . Clearly z 0 ∈ C m − f − 1 ( { 0 } ) s z_0\in\mathbb{C}^m-f^{-1}(\{0\})_s z 0 ∈ C m − f − 1 ({ 0 } ) s . Then there is a holomorphic coordinate system ( U ; φ 1 , … , φ m ) (U;\varphi_1,\ldots,\varphi_m) ( U ; φ 1 , … , φ m ) of z 0 z_0 z 0 in C m \mathbb{C}^m C m such that
U ∩ f − 1 ( { 0 } ) = { z ∈ U ∣ φ 1 ( z ) = 0 } and φ j ( z 0 ) = 0 U\cap f^{-1}(\{0\})=\{z\in U\mid\varphi_1(z)=0\}\quad\text{and}\quad\varphi_j(z_0)=0 U ∩ f − 1 ({ 0 }) = { z ∈ U ∣ φ 1 ( z ) = 0 } and φ j ( z 0 ) = 0
for j = 1 , 2 , … , m j=1,2,\ldots,m j = 1 , 2 , … , m (see Lemma 2.3 [3 ] and the proof of Lemma 2.3 [22 ] ). Therefore biholomorphic coordinate transformation z j = z j ( φ 1 , … , φ m ) z_j=z_j(\varphi_1,\ldots,\varphi_m) z j = z j ( φ 1 , … , φ m ) , j = 1 , … , m j=1,\ldots,m j = 1 , … , m near 0 0 0 exists such that z 0 = z ( 0 ) = ( z 1 ( 0 ) , … , z m ( 0 ) ) z_0=z(0)=(z_1(0),\ldots,z_m(0)) z 0 = z ( 0 ) = ( z 1 ( 0 ) , … , z m ( 0 )) . So we can write
f = a p φ 1 p + a p + 1 φ 1 p + 1 + a p + 2 φ 1 p + 2 + φ 1 p + 3 f ^ ( φ 1 , … , φ m ) (30) f=a_p\varphi_1^p+a_{p+1}\varphi_1^{p+1}+a_{p+2}\varphi_1^{p+2}+\varphi_1^{p+3}\hat f(\varphi_1,\ldots,\varphi_m) \tag*{(30)} f = a p φ 1 p + a p + 1 φ 1 p + 1 + a p + 2 φ 1 p + 2 + φ 1 p + 3 f ^ ( φ 1 , … , φ m ) (30)
and
g = b p φ 1 p + b p + 1 φ 1 p + 1 + b p + 2 φ 1 p + 2 + φ 1 p + 3 g ^ ( φ 1 , … , φ m ) , (31) g=b_p\varphi_1^p+b_{p+1}\varphi_1^{p+1}+b_{p+2}\varphi_1^{p+2}+\varphi_1^{p+3}\hat g(\varphi_1,\ldots,\varphi_m), \tag*{(31)} g = b p φ 1 p + b p + 1 φ 1 p + 1 + b p + 2 φ 1 p + 2 + φ 1 p + 3 g ^ ( φ 1 , … , φ m ) , (31)
where a p ( ≠ 0 ) , a p + 1 , a p + 2 , b p ( ≠ 0 ) , b p + 1 , b p + 2 ∈ C a_p(\ne0),a_{p+1},a_{p+2},b_p(\ne0),b_{p+1},b_{p+2}\in\mathbb{C} a p ( = 0 ) , a p + 1 , a p + 2 , b p ( = 0 ) , b p + 1 , b p + 2 ∈ C and f ^ \hat f f ^ and g ^ \hat g g ^ are holomorphic function near 0 0 0 . Clearly (3.2) and (3.3) give
∂ z i ( f ) = p a p φ 1 p − 1 ∂ z i ( φ 1 ) + ( p + 1 ) a p + 1 φ 1 p ∂ z i ( φ 1 ) + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z i ( φ 1 ) + … , (32) \partial_{z_i}(f)=pa_p\varphi_1^{p-1}\partial_{z_i}(\varphi_1)+(p+1)a_{p+1}\varphi_1^p\partial_{z_i}(\varphi_1)+(p+2)a_{p+2}\varphi_1^{p+1}\partial_{z_i}(\varphi_1)+\ldots, \tag*{(32)} ∂ z i ( f ) = p a p φ 1 p − 1 ∂ z i ( φ 1 ) + ( p + 1 ) a p + 1 φ 1 p ∂ z i ( φ 1 ) + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z i ( φ 1 ) + … , (32)
∂ z j z i 2 ( f ) = p ( p − 1 ) a p φ 1 p − 2 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + p a p φ 1 p − 1 ∂ z j z i 2 ( φ 1 ) + ( p + 1 ) p a p + 1 φ 1 p − 1 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 1 ) a p + 1 φ 1 p ∂ z j z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) a p + 2 φ 1 p ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z j z i 2 ( φ 1 ) + … , (33) \begin{aligned}
\partial^2_{z_jz_i}(f)={}&p(p-1)a_p\varphi_1^{p-2}\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+pa_p\varphi_1^{p-1}\partial^2_{z_jz_i}(\varphi_1)\\
&+(p+1)pa_{p+1}\varphi_1^{p-1}\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+(p+1)a_{p+1}\varphi_1^p\partial^2_{z_jz_i}(\varphi_1)\\
&+(p+2)(p+1)a_{p+2}\varphi_1^p\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+(p+2)a_{p+2}\varphi_1^{p+1}\partial^2_{z_jz_i}(\varphi_1)+\ldots, \tag*{(33)}
\end{aligned} ∂ z j z i 2 ( f ) = p ( p − 1 ) a p φ 1 p − 2 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + p a p φ 1 p − 1 ∂ z j z i 2 ( φ 1 ) + ( p + 1 ) p a p + 1 φ 1 p − 1 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 1 ) a p + 1 φ 1 p ∂ z j z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) a p + 2 φ 1 p ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z j z i 2 ( φ 1 ) + … , (33)
∂ z i 2 ( f ) = p ( p − 1 ) a p φ 1 p − 2 ( ∂ z i ( φ 1 ) ) 2 + p a p φ 1 p − 1 ∂ z i 2 ( φ 1 ) + ( p + 1 ) p a p + 1 φ 1 p − 1 ( ∂ z i ( φ 1 ) ) 2 + ( p + 1 ) a p + 1 φ 1 p ∂ z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) a p + 2 φ 1 p ( ∂ z i ( φ 1 ) ) 2 + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z i 2 ( φ 1 ) + … , \begin{aligned}
\partial^2_{z_i}(f)={}&p(p-1)a_p\varphi_1^{p-2}(\partial_{z_i}(\varphi_1))^2+pa_p\varphi_1^{p-1}\partial^2_{z_i}(\varphi_1)+(p+1)pa_{p+1}\varphi_1^{p-1}(\partial_{z_i}(\varphi_1))^2\\
&+(p+1)a_{p+1}\varphi_1^p\partial^2_{z_i}(\varphi_1)+(p+2)(p+1)a_{p+2}\varphi_1^p(\partial_{z_i}(\varphi_1))^2\\
&+(p+2)a_{p+2}\varphi_1^{p+1}\partial^2_{z_i}(\varphi_1)+\ldots,
\end{aligned} ∂ z i 2 ( f ) = p ( p − 1 ) a p φ 1 p − 2 ( ∂ z i ( φ 1 ) ) 2 + p a p φ 1 p − 1 ∂ z i 2 ( φ 1 ) + ( p + 1 ) p a p + 1 φ 1 p − 1 ( ∂ z i ( φ 1 ) ) 2 + ( p + 1 ) a p + 1 φ 1 p ∂ z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) a p + 2 φ 1 p ( ∂ z i ( φ 1 ) ) 2 + ( p + 2 ) a p + 2 φ 1 p + 1 ∂ z i 2 ( φ 1 ) + … ,
∂ z i n ( f ) = p ( p − 1 ) ⋯ ( p − n + 1 ) a p φ 1 p − n ( ∂ z i ( φ 1 ) ) n + … + ( p + 1 ) p ⋯ ( p − n + 2 ) a p + 1 φ 1 p − n + 1 ( ∂ z i ( φ 1 ) ) n + ( p + 2 ) ( p + 1 ) ⋯ ( p − n + 3 ) a p + 2 φ 1 p − n + 2 ( ∂ z i ( φ 1 ) ) n + … , (34) \begin{aligned}
\partial^n_{z_i}(f)={}&p(p-1)\cdots(p-n+1)a_p\varphi_1^{p-n}(\partial_{z_i}(\varphi_1))^n+\ldots\\
&+(p+1)p\cdots(p-n+2)a_{p+1}\varphi_1^{p-n+1}(\partial_{z_i}(\varphi_1))^n\\
&+(p+2)(p+1)\cdots(p-n+3)a_{p+2}\varphi_1^{p-n+2}(\partial_{z_i}(\varphi_1))^n+\ldots, \tag*{(34)}
\end{aligned} ∂ z i n ( f ) = p ( p − 1 ) ⋯ ( p − n + 1 ) a p φ 1 p − n ( ∂ z i ( φ 1 ) ) n + … + ( p + 1 ) p ⋯ ( p − n + 2 ) a p + 1 φ 1 p − n + 1 ( ∂ z i ( φ 1 ) ) n + ( p + 2 ) ( p + 1 ) ⋯ ( p − n + 3 ) a p + 2 φ 1 p − n + 2 ( ∂ z i ( φ 1 ) ) n + … , (34)
∂ z i ( g ) = p b p φ 1 p − 1 ∂ z i ( φ 1 ) + ( p + 1 ) b p + 1 φ 1 p ∂ z i ( φ 1 ) + ( p + 2 ) b p + 2 φ 1 p + 1 ∂ z i ( φ 1 ) + … , (35) \partial_{z_i}(g)=pb_p\varphi_1^{p-1}\partial_{z_i}(\varphi_1)+(p+1)b_{p+1}\varphi_1^p\partial_{z_i}(\varphi_1)+(p+2)b_{p+2}\varphi_1^{p+1}\partial_{z_i}(\varphi_1)+\ldots, \tag*{(35)} ∂ z i ( g ) = p b p φ 1 p − 1 ∂ z i ( φ 1 ) + ( p + 1 ) b p + 1 φ 1 p ∂ z i ( φ 1 ) + ( p + 2 ) b p + 2 φ 1 p + 1 ∂ z i ( φ 1 ) + … , (35)
∂ z j z i 2 ( g ) = p ( p − 1 ) b p φ 1 p − 2 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + p b p φ 1 p − 1 ∂ z j z i 2 ( φ 1 ) + ( p + 1 ) p b p + 1 φ 1 p − 1 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 1 ) b p + 1 φ 1 p ∂ z j z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) b p + 2 φ 1 p ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 2 ) b p + 2 φ 1 p + 1 ∂ z j z i 2 ( φ 1 ) + … , (36) \begin{aligned}
\partial^2_{z_jz_i}(g)={}&p(p-1)b_p\varphi_1^{p-2}\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+pb_p\varphi_1^{p-1}\partial^2_{z_jz_i}(\varphi_1)\\
&+(p+1)pb_{p+1}\varphi_1^{p-1}\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+(p+1)b_{p+1}\varphi_1^p\partial^2_{z_jz_i}(\varphi_1)\\
&+(p+2)(p+1)b_{p+2}\varphi_1^p\partial_{z_j}(\varphi_1)\partial_{z_i}(\varphi_1)+(p+2)b_{p+2}\varphi_1^{p+1}\partial^2_{z_jz_i}(\varphi_1)+\ldots, \tag*{(36)}
\end{aligned} ∂ z j z i 2 ( g ) = p ( p − 1 ) b p φ 1 p − 2 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + p b p φ 1 p − 1 ∂ z j z i 2 ( φ 1 ) + ( p + 1 ) p b p + 1 φ 1 p − 1 ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 1 ) b p + 1 φ 1 p ∂ z j z i 2 ( φ 1 ) + ( p + 2 ) ( p + 1 ) b p + 2 φ 1 p ∂ z j ( φ 1 ) ∂ z i ( φ 1 ) + ( p + 2 ) b p + 2 φ 1 p + 1 ∂ z j z i 2 ( φ 1 ) + … , (36)
and
∂ z i n ( g ) = p ( p − 1 ) ⋯ ( p − n + 1 ) b p φ 1 p − n ( ∂ z i ( φ 1 ) ) n + ⋯ (37) \partial_{z_i}^{n}(g)=p(p-1)\cdots(p-n+1)b_p\varphi_1^{p-n}(\partial_{z_i}(\varphi_1))^n+\cdots \tag*{(37)} ∂ z i n ( g ) = p ( p − 1 ) ⋯ ( p − n + 1 ) b p φ 1 p − n ( ∂ z i ( φ 1 ) ) n + ⋯ (37)
∂ z i n ( g ) = + ( p + 1 ) p ⋯ ( p − n + 2 ) b p + 1 φ 1 p − n + 1 ( ∂ z i ( φ 1 ) ) n \phantom{\partial_{z_i}^{n}(g)={}}+(p+1)p\cdots(p-n+2)b_{p+1}\varphi_1^{p-n+1}(\partial_{z_i}(\varphi_1))^n ∂ z i n ( g ) = + ( p + 1 ) p ⋯ ( p − n + 2 ) b p + 1 φ 1 p − n + 1 ( ∂ z i ( φ 1 ) ) n
∂ z i n ( g ) = + ( p + 2 ) ( p + 1 ) ⋯ ( p − n + 3 ) b p + 2 φ 1 p − n + 2 ( ∂ z i ( φ 1 ) ) n + ⋯ . \phantom{\partial_{z_i}^{n}(g)={}}+(p+2)(p+1)\cdots(p-n+3)b_{p+2}\varphi_1^{p-n+2}(\partial_{z_i}(\varphi_1))^n+\cdots. ∂ z i n ( g ) = + ( p + 2 ) ( p + 1 ) ⋯ ( p − n + 3 ) b p + 2 φ 1 p − n + 2 ( ∂ z i ( φ 1 ) ) n + ⋯ .
Now we consider following two cases.
Case 1. Suppose ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is a rational function for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Then by Lemma 2.4 , we get the desired conclusions.
Case 2. Suppose ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is transcendental for atleast one i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Without loss of generality, we may assume that ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is transcendental for all i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . By Lemma 2.1 , we have
m ( r , ∂ z i ( f ) f ) = O ( log r ) . m\left(r,\frac{\partial_{z_i}(f)}{f}\right)=O(\log r). m ( r , f ∂ z i ( f ) ) = O ( log r ) .
Also it is easy to deduce that
N ( r , ∂ z i ( f ) f ) ≤ N ‾ ( r , f ) + N ‾ ( r , 0 ; f ) = N ‾ ( r , f ) + ∑ k = 1 n − 1 N ‾ = k ( r , 0 ; f ) + N ‾ ( n ( r , 0 ; f ) . N\left(r,\frac{\partial_{z_i}(f)}{f}\right)\leq\overline{N}(r,f)+\overline{N}(r,0;f)=\overline{N}(r,f)+\sum_{k=1}^{n-1}\overline{N}_{=k}(r,0;f)+\overline{N}_{(n}(r,0;f). N ( r , f ∂ z i ( f ) ) ≤ N ( r , f ) + N ( r , 0 ; f ) = N ( r , f ) + k = 1 ∑ n − 1 N = k ( r , 0 ; f ) + N ( n ( r , 0 ; f ) .
Consequently
T ( r , ∂ z i ( f ) f ) ≤ N ‾ ( r , f ) + ∑ k = 1 n − 1 N ‾ = i ( r , 0 ; f ) + N ‾ ( n ( r , 0 ; f ) + O ( log r ) . (38) T\left(r,\frac{\partial_{z_i}(f)}{f}\right)\leq\overline{N}(r,f)+\sum_{k=1}^{n-1}\overline{N}_{=i}(r,0;f)+\overline{N}_{(n}(r,0;f)+O(\log r). \tag*{(38)} T ( r , f ∂ z i ( f ) ) ≤ N ( r , f ) + k = 1 ∑ n − 1 N = i ( r , 0 ; f ) + N ( n ( r , 0 ; f ) + O ( log r ) . (38)
Let
S k = { z ∈ C m : μ ‾ f = k 0 ( z ) > 0 } , S ( n = { z ∈ C m : μ ‾ f ( n 0 ( z ) > 0 } and S = { z ∈ C m : μ ‾ f ∞ ( z ) > 0 } , S_k=\{z\in\mathbb{C}^m:\overline{\mu}^{0}_{f=k}(z)>0\},\ S_{(n}=\{z\in\mathbb{C}^m:\overline{\mu}^{0}_{f(n}(z)>0\}\text{ and }S=\{z\in\mathbb{C}^m:\overline{\mu}^{\infty}_{f}(z)>0\}, S k = { z ∈ C m : μ f = k 0 ( z ) > 0 } , S ( n = { z ∈ C m : μ f ( n 0 ( z ) > 0 } and S = { z ∈ C m : μ f ∞ ( z ) > 0 } ,
where k = 1 , 2 , … , n − 1 k=1,2,\ldots,n-1 k = 1 , 2 , … , n − 1 . Since ∂ z i ( f ) / f \partial_{z_i}(f)/f ∂ z i ( f ) / f is transcendental, from (38) it follows that the set ⋃ k = 1 n − 1 S k ∪ S ( n ∪ S \bigcup_{k=1}^{n-1}S_k\cup S_{(n}\cup S ⋃ k = 1 n − 1 S k ∪ S ( n ∪ S is not algebraic.
We now consider following two sub-cases.
Sub-case 2.1. Suppose the set S k S_k S k is not algebraic for at least one value of k ∈ { 1 , 2 , … , n − 1 } k\in\{1,2,\ldots,n-1\} k ∈ { 1 , 2 , … , n − 1 } . Let
Ψ k , i j = ( k + 1 ) Φ 0 , j − k Φ 1 , i j = ( k + 1 ) ( ∂ z j ( f ) f − ∂ z j ( g ) g ) − k ( ∂ z j z i 2 ( f ) ∂ z i ( f ) − ∂ z j z i 2 ( g ) ∂ z i ( g ) ) , (39) \Psi_{k,ij}=(k+1)\Phi_{0,j}-k\Phi_{1,ij}=(k+1)\left(\frac{\partial_{z_j}(f)}{f}-\frac{\partial_{z_j}(g)}{g}\right)-k\left(\frac{\partial^2_{z_jz_i}(f)}{\partial_{z_i}(f)}-\frac{\partial^2_{z_jz_i}(g)}{\partial_{z_i}(g)}\right), \tag*{(39)} Ψ k , ij = ( k + 1 ) Φ 0 , j − k Φ 1 , ij = ( k + 1 ) ( f ∂ z j ( f ) − g ∂ z j ( g ) ) − k ( ∂ z i ( f ) ∂ z j z i 2 ( f ) − ∂ z i ( g ) ∂ z j z i 2 ( g ) ) , (39)
k = 1 , 2 , … , n − 1 k=1,2,\ldots,n-1 k = 1 , 2 , … , n − 1 and i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m . On the other hand from (29) and (39) , we deduce that
Ψ k , i j = ( k + 1 ) ∂ z j ( P ) − k ∂ z j ( Q i ) (40) \Psi_{k,ij}=(k+1)\partial_{z_j}(P)-k\partial_{z_j}(Q_i) \tag*{(40)} Ψ k , ij = ( k + 1 ) ∂ z j ( P ) − k ∂ z j ( Q i ) (40)
k = 1 , 2 , … , n − 1 k=1,2,\ldots,n-1 k = 1 , 2 , … , n − 1 and i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m .
Now we divide following two sub-cases.
Sub-case 2.1.1. Suppose Ψ k , i j ≢ 0 \Psi_{k,ij}\not\equiv0 Ψ k , ij ≡ 0 for at least one ( k , i , j ) ∈ { 1 , 2 , … , n − 1 } × { 1 , 2 , … , m } × { 1 , 2 , … , m } (k,i,j)\in\{1,2,\ldots,n-1\}\times\{1,2,\ldots,m\}\times\{1,2,\ldots,m\} ( k , i , j ) ∈ { 1 , 2 , … , n − 1 } × { 1 , 2 , … , m } × { 1 , 2 , … , m } . Let z 0 ∈ S k z_0\in S_k z 0 ∈ S k such that k = μ f 0 ( z 0 ) > 0 k=\mu_f^0(z_0)>0 k = μ f 0 ( z 0 ) > 0 . By the given condition, we have k = μ g 0 ( z 0 ) > 0 k=\mu_g^0(z_0)>0 k = μ g 0 ( z 0 ) > 0 . Let us take p = k p=k p = k . Now we apply (30) –(33) , (35) and (36) to (39) and then by a elementary calculation, we can conclude that Ψ k , i j ( z 0 ) = 0 \Psi_{k,ij}(z_0)=0 Ψ k , ij ( z 0 ) = 0 . Consequently the set S k S_k S k is algebraic, which is absurd.
Sub-case 2.1.2. Suppose Ψ k , i j ≡ 0 \Psi_{k,ij}\equiv0 Ψ k , ij ≡ 0 for all ( k , i , j ) ∈ { 1 , 2 , … , n − 1 } × { 1 , 2 , … , m } × { 1 , 2 , … , m } (k,i,j)\in\{1,2,\ldots,n-1\}\times\{1,2,\ldots,m\}\times\{1,2,\ldots,m\} ( k , i , j ) ∈ { 1 , 2 , … , n − 1 } × { 1 , 2 , … , m } × { 1 , 2 , … , m } . Then from (40) , we get
( k + 1 ) ∂ z j ( P ) − k ∂ z j ( Q i ) ≡ 0 , (41) (k+1)\partial_{z_j}(P)-k\partial_{z_j}(Q_i)\equiv0, \tag*{(41)} ( k + 1 ) ∂ z j ( P ) − k ∂ z j ( Q i ) ≡ 0 , (41)
where k = 1 , 2 , … , n − 1 k = 1, 2, \ldots, n - 1 k = 1 , 2 , … , n − 1 and i , j = 1 , 2 , … , m i, j = 1, 2, \ldots, m i , j = 1 , 2 , … , m . On integration, we get
Q i ( z ) = k + 1 k P ( z ) + R ~ j ( z ) , (42) Q_i(z) = \frac{k+1}{k}P(z) + \widetilde{R}_j(z), \tag*{(42)} Q i ( z ) = k k + 1 P ( z ) + R j ( z ) , (42)
where R ~ j ( z ) = R ~ ( z 1 , z 2 , … , z j − 1 , z j + 1 , … , z m ) \widetilde{R}_j(z) = \widetilde{R}(z_1,z_2,\ldots,z_{j-1},z_{j+1},\ldots,z_m) R j ( z ) = R ( z 1 , z 2 , … , z j − 1 , z j + 1 , … , z m ) is a polynomial. Clearly from (42) , we get
Q 1 ( z ) = k + 1 k P ( z ) + R ~ 1 ( z ) , Q 1 ( z ) = k + 1 k P ( z ) + R ~ 2 ( z ) , … , Q 1 ( z ) = k + 1 k P ( z ) + R ~ m ( z ) Q_1(z) = \frac{k+1}{k}P(z) + \widetilde{R}_1(z), Q_1(z) = \frac{k+1}{k}P(z) + \widetilde{R}_2(z), \ldots, Q_1(z) = \frac{k+1}{k}P(z) + \widetilde{R}_m(z) Q 1 ( z ) = k k + 1 P ( z ) + R 1 ( z ) , Q 1 ( z ) = k k + 1 P ( z ) + R 2 ( z ) , … , Q 1 ( z ) = k k + 1 P ( z ) + R m ( z )
which demands that R ~ 1 ( z ) = R ~ 2 ( z ) = … = R ~ m ( z ) = A ∈ C \widetilde{R}_1(z) = \widetilde{R}_2(z) = \ldots= \widetilde{R}_m(z) = A \in\mathbb{C} R 1 ( z ) = R 2 ( z ) = … = R m ( z ) = A ∈ C . Therefore from (42) , we get
Q i ( z ) = k + 1 k P ( z ) + A . (43) Q_i(z) = \frac{k+1}{k}P(z) + A. \tag*{(43)} Q i ( z ) = k k + 1 P ( z ) + A . (43)
Now from (29) , we have
∂ z i ( f ) f = ∂ z i ( g ) g + ∂ z i ( P ) (44) \frac{\partial_{z_i}(f)}{f} = \frac{\partial_{z_i}(g)}{g} + \partial_{z_i}(P) \tag*{(44)} f ∂ z i ( f ) = g ∂ z i ( g ) + ∂ z i ( P ) (44)
where i = 1 , 2 , … , m i = 1, 2, \ldots, m i = 1 , 2 , … , m . Again from (29) , we get
∂ z i ( f ) f = ∂ z i ( g ) g e Q i − P , (45) \frac{\partial_{z_i}(f)}{f} = \frac{\partial_{z_i}(g)}{g}e^{Q_i-P}, \tag*{(45)} f ∂ z i ( f ) = g ∂ z i ( g ) e Q i − P , (45)
where i = 1 , 2 , … , m i = 1, 2, \ldots, m i = 1 , 2 , … , m . Therefore from (44) and (45) , we obtain
∂ z i ( g ) g = − ∂ z i ( P ) 1 − e Q i − P and ∂ z i ( f ) f = − ∂ z i ( P ) e − Q i + P − 1 , (46) \frac{\partial_{z_i}(g)}{g} = \frac{-\partial_{z_i}(P)}{1-e^{Q_i-P}} \quad\text{and} \quad\frac{\partial_{z_i}(f)}{f} = \frac{-\partial_{z_i}(P)}{e^{-Q_i+P}-1}, \tag*{(46)} g ∂ z i ( g ) = 1 − e Q i − P − ∂ z i ( P ) and f ∂ z i ( f ) = e − Q i + P − 1 − ∂ z i ( P ) , (46)
where i = 1 , 2 , … , m i = 1, 2, \ldots, m i = 1 , 2 , … , m .
Note that the set S k S_k S k is not algebraic for at least one value of k ∈ { 1 , 2 , … , n − 1 } k \in\{1, 2, \ldots, n - 1\} k ∈ { 1 , 2 , … , n − 1 } . Since ∂ z i ( g ) / g \partial_{z_i}(g)/g ∂ z i ( g ) / g has only simple poles, it follows from (46) that deg ( Q i − P ) = 1 \deg(Q_i-P)=1 deg ( Q i − P ) = 1 . Consequently from (41) , we can conclude that deg ( P ) = deg ( Q i ) = 1 \deg(P)=\deg(Q_i)=1 deg ( P ) = deg ( Q i ) = 1 . Again using (43) to (46) , we get
∂ z i ( g ) g = k 1 k ∂ z i ( P ) e − 1 k P − A 1 − e − 1 k P − A and ∂ z i ( f ) f = k 1 k ∂ z i ( P ) e 1 k P + A e 1 k P + A − 1 , (47) \frac{\partial_{z_i}(g)}{g} = k\frac{\frac{1}{k}\partial_{z_i}(P)e^{-\frac{1}{k}P-A}}{1-e^{-\frac{1}{k}P-A}} \quad\text{and} \quad\frac{\partial_{z_i}(f)}{f} = k\frac{\frac{1}{k}\partial_{z_i}(P)e^{\frac{1}{k}P+A}}{e^{\frac{1}{k}P+A}-1}, \tag*{(47)} g ∂ z i ( g ) = k 1 − e − k 1 P − A k 1 ∂ z i ( P ) e − k 1 P − A and f ∂ z i ( f ) = k e k 1 P + A − 1 k 1 ∂ z i ( P ) e k 1 P + A , (47)
where i , j = 1 , 2 , … , m i,j = 1, 2, \ldots, m i , j = 1 , 2 , … , m . On integration, we get from (47) that
g ( z ) = ( 1 − e − 1 k P ( z ) − A ) k e R ^ 1 i ( z ) and f ( z ) = ( e 1 k P ( z ) + A − 1 ) k e R ^ 2 i ( z ) , (48) g(z)=\left(1-e^{-\frac{1}{k}P(z)-A}\right)^k e^{\widehat{R}_{1i}(z)} \quad\text{and} \quad f(z)=\left(e^{\frac{1}{k}P(z)+A}-1\right)^k e^{\widehat{R}_{2i}(z)}, \tag*{(48)} g ( z ) = ( 1 − e − k 1 P ( z ) − A ) k e R 1 i ( z ) and f ( z ) = ( e k 1 P ( z ) + A − 1 ) k e R 2 i ( z ) , (48)
where i = 1 , 2 , … , m i = 1, 2, \ldots, m i = 1 , 2 , … , m and R ^ j i ( z ) = R ^ j ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) \widehat{R}_{ji}(z) = \widehat{R}_j(z_1,z_2,\ldots,z_{i-1},z_{i+1},\ldots,z_m) R j i ( z ) = R j ( z 1 , z 2 , … , z i − 1 , z i + 1 , … , z m ) is a polynomial. Now proceeding in the same way as done in the proof of Lemma 2.6, we can prove that R ^ 1 i ( z ) \widehat{R}_{1i}(z) R 1 i ( z ) and R ^ 2 i ( z ) \widehat{R}_{2i}(z) R 2 i ( z ) are constants. Consequently from (48) , we have
g ( z ) = A i ( 1 − e − 1 k P ( z ) − A ) k and f ( z ) = B i ( e 1 k P ( z ) + A − 1 ) k , (49) g(z)=A_i\left(1-e^{-\frac{1}{k}P(z)-A}\right)^k \quad\text{and} \quad f(z)=B_i\left(e^{\frac{1}{k}P(z)+A}-1\right)^k, \tag*{(49)} g ( z ) = A i ( 1 − e − k 1 P ( z ) − A ) k and f ( z ) = B i ( e k 1 P ( z ) + A − 1 ) k , (49)
where A i , B i ∈ C \ { 0 } A_i,B_i \in\mathbb{C}\backslash\{0\} A i , B i ∈ C \ { 0 } and i = 1 , 2 , … , m i = 1, 2, \ldots, m i = 1 , 2 , … , m . Again from (49) , one can easily prove that A 1 = A 2 = … = A m A_1=A_2=\ldots=A_m A 1 = A 2 = … = A m and B 1 = B 2 = … = B m B_1=B_2=\ldots=B_m B 1 = B 2 = … = B m . Let us take
P ( z ) = a 1 z 1 + a 2 z 2 + … + a m z m + a 0 , P(z)=a_1z_1+a_2z_2+\ldots+a_mz_m+a_0, P ( z ) = a 1 z 1 + a 2 z 2 + … + a m z m + a 0 ,
where a 0 , a 1 , a 2 , … , a m ∈ C a_0,a_1,a_2,\ldots,a_m \in\mathbb{C} a 0 , a 1 , a 2 , … , a m ∈ C such that ( a 1 , a 2 , … , a m ) ≠ ( 0 , 0 , … , 0 ) (a_1,a_2,\ldots,a_m)\ne(0,0,\ldots,0) ( a 1 , a 2 , … , a m ) = ( 0 , 0 , … , 0 ) . Now in view of (49) , we may assume that
f ( z ) = a ( 1 − b e 1 k ( a 1 z 1 + a 2 z 2 + … + a m z m ) ) k and g ( z ) = c ( e − 1 k ( a 1 z 1 + a 2 z 2 + … + a m z m ) − b ) k , (50) f(z)=a\left(1-be^{\frac{1}{k}(a_1z_1+a_2z_2+\ldots+a_mz_m)}\right)^k \quad\text{and} \quad g(z)=c\left(e^{-\frac{1}{k}(a_1z_1+a_2z_2+\ldots+a_mz_m)}-b\right)^k, \tag*{(50)} f ( z ) = a ( 1 − b e k 1 ( a 1 z 1 + a 2 z 2 + … + a m z m ) ) k and g ( z ) = c ( e − k 1 ( a 1 z 1 + a 2 z 2 + … + a m z m ) − b ) k , (50)
where a , b , c ∈ C \ { 0 } a,b,c \in\mathbb{C}\backslash\{0\} a , b , c ∈ C \ { 0 } . Applying the binomial theorem, we get
f ( z ) = a { 1 − k b e 1 k ( a 1 z 1 + … + a m z m ) + ( k 2 ) b 2 e 2 k ( a 1 z 1 + … + a m z m ) − … + ( − b ) k e a 1 z 1 + … + a m z m } (51) f(z)=a\left\{1-kbe^{\frac{1}{k}(a_1z_1+\ldots+a_mz_m)}+\binom{k}{2}b^2e^{\frac{2}{k}(a_1z_1+\ldots+a_mz_m)}-\ldots+(-b)^ke^{a_1z_1+\ldots+a_mz_m}\right\} \tag*{(51)} f ( z ) = a { 1 − k b e k 1 ( a 1 z 1 + … + a m z m ) + ( 2 k ) b 2 e k 2 ( a 1 z 1 + … + a m z m ) − … + ( − b ) k e a 1 z 1 + … + a m z m } (51)
and
g ( z ) = c { e − ( a 1 z 1 + ⋯ + a m z m ) − k b e − k − 1 k ( a 1 z 1 + ⋯ + a m z m ) + ( k 2 ) b 2 e − k − 2 k ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + k ( − b ) k − 1 e − 1 k ( a 1 z 1 + ⋯ + a m z m ) + ( − b ) k } . (52) g(z)=c\left\{e^{-(a_1z_1+\cdots+a_mz_m)}-kbe^{-\frac{k-1}{k}(a_1z_1+\cdots+a_mz_m)}+\binom{k}{2}b^2e^{-\frac{k-2}{k}(a_1z_1+\cdots+a_mz_m)}
-\cdots+k(-b)^{k-1}e^{-\frac{1}{k}(a_1z_1+\cdots+a_mz_m)}+(-b)^k\right\}. \tag*{(52)} g ( z ) = c { e − ( a 1 z 1 + ⋯ + a m z m ) − k b e − k k − 1 ( a 1 z 1 + ⋯ + a m z m ) + ( 2 k ) b 2 e − k k − 2 ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + k ( − b ) k − 1 e − k 1 ( a 1 z 1 + ⋯ + a m z m ) + ( − b ) k } . (52)
Now differentiating partially (51) and (52) n n n -times with respect to z i z_i z i , we get respectively
∂ n ( f ( z ) ) ∂ z i n = a ( a i k ) n ( − k b e 1 k ( a 1 z 1 + ⋯ + a m z m ) + ( k 2 ) b 2 2 n e 2 k ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + ( − b ) k k n e a 1 z 1 + ⋯ + a m z m ) (53) \frac{\partial^n(f(z))}{\partial z_i^n}=a\left(\frac{a_i}{k}\right)^n\left(-kbe^{\frac{1}{k}(a_1z_1+\cdots+a_mz_m)}+\binom{k}{2}b^2 2^n e^{\frac{2}{k}(a_1z_1+\cdots+a_mz_m)}-\cdots+(-b)^k k^n e^{a_1z_1+\cdots+a_mz_m}\right) \tag*{(53)} ∂ z i n ∂ n ( f ( z )) = a ( k a i ) n ( − k b e k 1 ( a 1 z 1 + ⋯ + a m z m ) + ( 2 k ) b 2 2 n e k 2 ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + ( − b ) k k n e a 1 z 1 + ⋯ + a m z m ) (53)
and
∂ n ( g ( z ) ) ∂ z i n = c ( − a i k ) n ( k n e − ( a 1 z 1 + ⋯ + a m z m ) − k b ( k − 1 ) n e − k − 1 k ( a 1 z 1 + ⋯ + a m z m ) + ( k 2 ) b 2 ( k − 2 ) n e − k − 2 k ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + k ( − b ) k − 1 e − 1 k ( a 1 z 1 + ⋯ + a m z m ) ) . (54) \frac{\partial^n(g(z))}{\partial z_i^n}=c\left(-\frac{a_i}{k}\right)^n\left(k^n e^{-(a_1z_1+\cdots+a_mz_m)}-kb(k-1)^n e^{-\frac{k-1}{k}(a_1z_1+\cdots+a_mz_m)}+\binom{k}{2}b^2(k-2)^n e^{-\frac{k-2}{k}(a_1z_1+\cdots+a_mz_m)}-\cdots+k(-b)^{k-1}e^{-\frac{1}{k}(a_1z_1+\cdots+a_mz_m)}\right). \tag*{(54)} ∂ z i n ∂ n ( g ( z )) = c ( − k a i ) n ( k n e − ( a 1 z 1 + ⋯ + a m z m ) − k b ( k − 1 ) n e − k k − 1 ( a 1 z 1 + ⋯ + a m z m ) + ( 2 k ) b 2 ( k − 2 ) n e − k k − 2 ( a 1 z 1 + ⋯ + a m z m ) − ⋯ + k ( − b ) k − 1 e − k 1 ( a 1 z 1 + ⋯ + a m z m ) ) . (54)
Since ∂ n ∂ z i n ( f ( z ) ) \frac{\partial^n}{\partial z_i^n}(f(z)) ∂ z i n ∂ n ( f ( z )) and ∂ n ∂ z i n ( g ( z ) ) \frac{\partial^n}{\partial z_i^n}(g(z)) ∂ z i n ∂ n ( g ( z )) share 0 CM, from (53) and (54) , one can easily conclude that k = 1 k=1 k = 1 . Therefore from (50) , we have
f ( z ) = a ( 1 − b e ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) ) f(z)=a\left(1-be^{(a_1z_1+a_2z_2+\cdots+a_mz_m)}\right) f ( z ) = a ( 1 − b e ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) )
and
g ( z ) = c ( e − ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) − b ) , g(z)=c\left(e^{-(a_1z_1+a_2z_2+\cdots+a_mz_m)}-b\right), g ( z ) = c ( e − ( a 1 z 1 + a 2 z 2 + ⋯ + a m z m ) − b ) ,
where a , b , c ∈ C \ { 0 } a,b,c\in\mathbb{C}\backslash\{0\} a , b , c ∈ C \ { 0 } and a 1 , a 2 , … , a m ∈ C a_1,a_2,\ldots,a_m\in\mathbb{C} a 1 , a 2 , … , a m ∈ C such that ( a 1 , a 2 , … , a m ) ≠ ( 0 , 0 , … , 0 ) (a_1,a_2,\ldots,a_m)\ne(0,0,\ldots,0) ( a 1 , a 2 , … , a m ) = ( 0 , 0 , … , 0 ) .
Sub-case 2.2. Suppose the set S k S_k S k is algebraic for all k ∈ { 1 , 2 , … , n − 1 } k\in\{1,2,\ldots,n-1\} k ∈ { 1 , 2 , … , n − 1 } . In this case, the set S n ∪ S S_n\cup S S n ∪ S is not algebraic. We define
Φ n , i j = ∂ z j z i n n + 1 ( f ) ∂ z i n ( f ) − ∂ z j z i n n + 1 ( g ) ∂ z i n ( g ) , (55) \Phi_{n,ij}=\frac{\partial^{n+1}_{z_jz_i^n}(f)}{\partial^n_{z_i}(f)}-\frac{\partial^{n+1}_{z_jz_i^n}(g)}{\partial^n_{z_i}(g)}, \tag*{(55)} Φ n , ij = ∂ z i n ( f ) ∂ z j z i n n + 1 ( f ) − ∂ z i n ( g ) ∂ z j z i n n + 1 ( g ) , (55)
where i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m and j = 1 , 2 , … , m j=1,2,\ldots,m j = 1 , 2 , … , m .
Suppose p = μ f 0 ( z 0 ) ≥ n p=\mu_f^0(z_0)\ge n p = μ f 0 ( z 0 ) ≥ n . By the given condition, we have p = μ g 0 ( z 0 ) ≥ n p=\mu_g^0(z_0)\ge n p = μ g 0 ( z 0 ) ≥ n . Set
A 0 : = A 0 ( z 0 ) = a p + 1 a p − b p + 1 b p (56) A_0:=A_0(z_0)=\frac{a_{p+1}}{a_p}-\frac{b_{p+1}}{b_p} \tag*{(56)} A 0 := A 0 ( z 0 ) = a p a p + 1 − b p b p + 1 (56)
B 0 : = B 0 ( z 0 ) = a p + 2 a p − b p + 2 b p (57) B_0:=B_0(z_0)=\frac{a_{p+2}}{a_p}-\frac{b_{p+2}}{b_p} \tag*{(57)} B 0 := B 0 ( z 0 ) = a p a p + 2 − b p b p + 2 (57)
and
C 0 : = C 0 ( z 0 ) = A 0 ( z 0 ) ( a p + 1 a p + b p + 1 b p ) . (58) C_0:=C_0(z_0)=A_0(z_0)\left(\frac{a_{p+1}}{a_p}+\frac{b_{p+1}}{b_p}\right). \tag*{(58)} C 0 := C 0 ( z 0 ) = A 0 ( z 0 ) ( a p a p + 1 + b p b p + 1 ) . (58)
Now using (30) -(37) to (22) , (23) and (55) , we get respectively
Φ 0 , j ( z ) = A 0 + 2 B 0 φ 1 − C 0 φ 1 + O ( φ 1 2 ) , (59) \Phi_{0,j}(z)=A_0+2B_0\varphi_1-C_0\varphi_1+O(\varphi_1^2), \tag*{(59)} Φ 0 , j ( z ) = A 0 + 2 B 0 φ 1 − C 0 φ 1 + O ( φ 1 2 ) , (59)
Φ 1 , i j ( z ) = A 0 p + 1 p + 2 B 0 p + 2 p φ 1 − C 0 ( p + 1 ) 2 p 2 φ 1 + O ( φ 1 2 ) (60) \Phi_{1,ij}(z)=A_0\frac{p+1}{p}+2B_0\frac{p+2}{p}\varphi_1-C_0\frac{(p+1)^2}{p^2}\varphi_1+O(\varphi_1^2) \tag*{(60)} Φ 1 , ij ( z ) = A 0 p p + 1 + 2 B 0 p p + 2 φ 1 − C 0 p 2 ( p + 1 ) 2 φ 1 + O ( φ 1 2 ) (60)
and
Φ n , i j ( z ) = A 0 ( p + 1 n ) ( p n ) + 2 B 0 ( p + 2 n ) ( p n ) φ 1 − C 0 ( p + 1 n ) 2 ( p n ) 2 φ 1 + O ( φ 1 2 ) . (61) \Phi_{n,ij}(z)=A_0\frac{\binom{p+1}{n}}{\binom{p}{n}}+2B_0\frac{\binom{p+2}{n}}{\binom{p}{n}}\varphi_1-C_0\frac{\binom{p+1}{n}^2}{\binom{p}{n}^2}\varphi_1+O(\varphi_1^2). \tag*{(61)} Φ n , ij ( z ) = A 0 ( n p ) ( n p + 1 ) + 2 B 0 ( n p ) ( n p + 2 ) φ 1 − C 0 ( n p ) 2 ( n p + 1 ) 2 φ 1 + O ( φ 1 2 ) . (61)
Suppose p = μ f ∞ ( z 0 ) > 0 p=\mu_f^\infty(z_0)>0 p = μ f ∞ ( z 0 ) > 0 . By the given condition, we have p = μ g ∞ ( z 0 ) > 0 p=\mu_g^\infty(z_0)>0 p = μ g ∞ ( z 0 ) > 0 . Then there is a holomorphic coordinate system ( U ; u 1 , … , u m ) (U;u_1,\ldots,u_m) ( U ; u 1 , … , u m ) of z 0 z_0 z 0 in C m − ( I ∪ S ) \mathbb{C}^m-(I\cup S) C m − ( I ∪ S ) such that
U ∩ supp μ f ∞ = { z ∈ U ∣ u 1 ( z ) = 0 } and u j ( z 0 ) = 0 U\cap\operatorname{supp}\mu_f^\infty=\{z\in U\mid u_1(z)=0\}\quad\text{and}\quad u_j(z_0)=0 U ∩ supp μ f ∞ = { z ∈ U ∣ u 1 ( z ) = 0 } and u j ( z 0 ) = 0
for j = 1 , 2 , … , m j=1,2,\ldots,m j = 1 , 2 , … , m . Therefore biholomorphic coordinate transformation z j = z j ( u 1 , … , u m ) z_j=z_j(u_1,\ldots,u_m) z j = z j ( u 1 , … , u m ) , j = 1 , … , m j=1,\ldots,m j = 1 , … , m near 0 0 0 exists such that z 0 = z ( 0 ) = ( z 1 ( 0 ) , … , z m ( 0 ) ) z_0=z(0)=(z_1(0),\ldots,z_m(0)) z 0 = z ( 0 ) = ( z 1 ( 0 ) , … , z m ( 0 )) . Therefore, we can write
f = c p φ 1 − p + c p + 1 φ 1 − p + 1 + c p + 2 φ 1 − p + 2 + φ 1 − p + 3 f ~ ( u 1 , … , u m ) (62) f=c_p\varphi_1^{-p}+c_{p+1}\varphi_1^{-p+1}+c_{p+2}\varphi_1^{-p+2}+\varphi_1^{-p+3}\widetilde{f}(u_1,\ldots,u_m) \tag*{(62)} f = c p φ 1 − p + c p + 1 φ 1 − p + 1 + c p + 2 φ 1 − p + 2 + φ 1 − p + 3 f ( u 1 , … , u m ) (62)
and
g = d p φ 1 − p + d − p + 1 φ 1 p + 1 + d − p + 2 φ 1 − p + 2 + φ 1 − p + 3 g ~ ( u 1 , … , u m ) , (63) g=d_p\varphi_1^{-p}+d_{-p+1}\varphi_1^{p+1}+d_{-p+2}\varphi_1^{-p+2}+\varphi_1^{-p+3}\widetilde{g}(u_1,\ldots,u_m), \tag*{(63)} g = d p φ 1 − p + d − p + 1 φ 1 p + 1 + d − p + 2 φ 1 − p + 2 + φ 1 − p + 3 g ( u 1 , … , u m ) , (63)
where c p ( ≠ 0 ) , c p + 1 , c p + 2 , d p ( ≠ 0 ) , d p + 1 , d p + 2 ∈ C c_p(\ne0),c_{p+1},c_{p+2},d_p(\ne0),d_{p+1},d_{p+2}\in\mathbb{C} c p ( = 0 ) , c p + 1 , c p + 2 , d p ( = 0 ) , d p + 1 , d p + 2 ∈ C and f ~ \widetilde{f} f and g ~ \widetilde{g} g are holomorphic function near 0 0 0 .
Set
A ∞ : = A ∞ ( z 0 ) = c p + 1 c p − d p + 1 d p (64) A_\infty:=A_\infty(z_0)=\frac{c_{p+1}}{c_p}-\frac{d_{p+1}}{d_p} \tag*{(64)} A ∞ := A ∞ ( z 0 ) = c p c p + 1 − d p d p + 1 (64)
B ∞ : = B ∞ ( z 0 ) = c p + 2 c p − d p + 2 d p (65) B_\infty:=B_\infty(z_0)=\frac{c_{p+2}}{c_p}-\frac{d_{p+2}}{d_p} \tag*{(65)} B ∞ := B ∞ ( z 0 ) = c p c p + 2 − d p d p + 2 (65)
and
C ∞ : = C ∞ ( z 0 ) = A ∞ ( z 0 ) ( c p + 1 c p + d p + 1 d p ) . (66) C_\infty:=C_\infty(z_0)=A_\infty(z_0)\left(\frac{c_{p+1}}{c_p}+\frac{d_{p+1}}{d_p}\right). \tag*{(66)} C ∞ := C ∞ ( z 0 ) = A ∞ ( z 0 ) ( c p c p + 1 + d p d p + 1 ) . (66)
In this case also, we have
Φ 0 , j ( z ) = A ∞ + 2 B ∞ u 1 − C ∞ u 1 2 + O ( u 1 2 ) , (67) \Phi_{0,j}(z)=A_\infty+2B_\infty u_1-C_\infty u_1^2+O(u_1^2), \tag*{(67)} Φ 0 , j ( z ) = A ∞ + 2 B ∞ u 1 − C ∞ u 1 2 + O ( u 1 2 ) , (67)
Φ 1 , i j ( z ) = A ∞ p + 1 p + 2 B ∞ p + 2 p u 1 − C ∞ ( p + 1 ) 2 p 2 u 1 + O ( u 1 2 ) (68) \Phi_{1,ij}(z)=A_\infty\frac{p+1}{p}+2B_\infty\frac{p+2}{p}u_1-C_\infty\frac{(p+1)^2}{p^2}u_1+O(u_1^2) \tag*{(68)} Φ 1 , ij ( z ) = A ∞ p p + 1 + 2 B ∞ p p + 2 u 1 − C ∞ p 2 ( p + 1 ) 2 u 1 + O ( u 1 2 ) (68)
and
Φ n , i j ( z ) = A ∞ ( p + 1 n ) ( p n ) + 2 B ∞ ( p + 2 n ) ( p n ) u 1 − C ∞ ( p + 1 n ) 2 ( p n ) 2 u 1 + O ( u 1 2 ) . (69) \Phi_{n,ij}(z)=A_\infty\frac{\binom{p+1}{n}}{\binom{p}{n}}+2B_\infty\frac{\binom{p+2}{n}}{\binom{p}{n}}u_1-C_\infty\frac{\binom{p+1}{n}^2}{\binom{p}{n}^2}u_1+O(u_1^2). \tag*{(69)} Φ n , ij ( z ) = A ∞ ( n p ) ( n p + 1 ) + 2 B ∞ ( n p ) ( n p + 2 ) u 1 − C ∞ ( n p ) 2 ( n p + 1 ) 2 u 1 + O ( u 1 2 ) . (69)
Let S ^ 0 = { z ∈ C m : μ f , n 0 ( z ) > 0 and A 0 ( z ) = 0 } \widehat{S}_0=\{z\in\mathbb{C}^m:\mu_{f,n}^0(z)>0\text{ and }A_0(z)=0\} S 0 = { z ∈ C m : μ f , n 0 ( z ) > 0 and A 0 ( z ) = 0 } and S ^ ∞ = { z ∈ C m : μ f , 1 ∞ ( z ) ≥ 1 and A ∞ ( z ) = 0 } \widehat{S}_\infty=\{z\in\mathbb{C}^m:\mu_{f,1}^\infty(z)\ge1\text{ and }A_\infty(z)=0\} S ∞ = { z ∈ C m : μ f , 1 ∞ ( z ) ≥ 1 and A ∞ ( z ) = 0 } .
Now we consider following two sub-cases.
Sub-case 2.2.1. Suppose the set S ^ 0 ∪ S ^ ∞ \widehat{S}_0\cup\widehat{S}_\infty S 0 ∪ S ∞ is not algebraic. By assumption, we have Φ 0 , i ≢ 0 \Phi_{0,i}\not\equiv0 Φ 0 , i ≡ 0 for at least one value of i ∈ { 1 , 2 , … , m } i\in\{1,2,\ldots,m\} i ∈ { 1 , 2 , … , m } . Again using (29) to (22) , we get Φ 0 , i = ∂ z i ( P ) \Phi_{0,i}=\partial_{z_i}(P) Φ 0 , i = ∂ z i ( P ) and so T ( r , Φ 0 , i ) = O ( log r ) T(r,\Phi_{0,i})=O(\log r) T ( r , Φ 0 , i ) = O ( log r ) for i = 1 , 2 , … , m i=1,2,\ldots,m i = 1 , 2 , … , m . Now using (56) to (59) and (64) to (67) , we get
N ‾ ( n ( r , 0 ; f ) + N ‾ ( r , f ) ≤ N ( r , 0 ; Φ 0 , i ) ≤ T ( r , Φ 0 , i ) + O ( 1 ) = O ( log r ) , \overline{N}(n(r,0;f)+\overline{N}(r,f)\leq N(r,0;\Phi_{0,i})\leq T(r,\Phi_{0,i})+O(1)=O(\log r), N ( n ( r , 0 ; f ) + N ( r , f ) ≤ N ( r , 0 ; Φ 0 , i ) ≤ T ( r , Φ 0 , i ) + O ( 1 ) = O ( log r ) ,
which shows that the set S ( n ∪ S S_{(n}\cup S S ( n ∪ S is algebraic. Therefore, we get a contradiction.
Sub-case 2.2.2. Suppose the set S ^ 0 ∪ S ^ ∞ \widehat{S}_0\cup\widehat{S}_\infty S 0 ∪ S ∞ is algebraic. In this case, the set S ~ 0 ∪ S ~ ∞ \widetilde{S}_0\cup\widetilde{S}_\infty S 0 ∪ S ∞ is not algebraic, where
S ~ 0 = { z ∈ C m : μ f , n 0 ( z ) > 0 and A 0 ( z ) ≠ 0 } and S ~ ∞ = { z ∈ C m : μ f , 1 ∞ ( z ) > 0 and A ∞ ( z ) ≠ 0 } . \widetilde{S}_0=\{z\in\mathbb{C}^m:\mu_{f,n}^0(z)>0\text{ and }A_0(z)\ne0\}\quad\text{and}\quad\widetilde{S}_\infty=\{z\in\mathbb{C}^m:\mu_{f,1}^\infty(z)>0\text{ and }A_\infty(z)\ne0\}. S 0 = { z ∈ C m : μ f , n 0 ( z ) > 0 and A 0 ( z ) = 0 } and S ∞ = { z ∈ C m : μ f , 1 ∞ ( z ) > 0 and A ∞ ( z ) = 0 } .
Also in this case, Φ 0 , i ≢ 0 \Phi_{0,i}\not\equiv0 Φ 0 , i ≡ 0 and Φ s , i j ≢ 0 \Phi_{s,ij}\not\equiv0 Φ s , ij ≡ 0 for all i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m and s = 1 , n s=1,n s = 1 , n . Now using (29) to (22) , (23) and (55) , we get respectively Φ 0 , i = ∂ z i ( P ) \Phi_{0,i}=\partial_{z_i}(P) Φ 0 , i = ∂ z i ( P ) , Φ 1 , i j = ∂ z j ( Q i ) \Phi_{1,ij}=\partial_{z_j}(Q_i) Φ 1 , ij = ∂ z j ( Q i ) and Φ n , i j ) = ∂ z j ( R i ) \Phi_{n,ij})=\partial_{z_j}(R_i) Φ n , ij ) = ∂ z j ( R i ) . Therefore T ( r , Φ 0 , i ) = O ( log r ) T(r,\Phi_{0,i})=O(\log r) T ( r , Φ 0 , i ) = O ( log r ) and T ( r , Φ s , i j ) = O ( log r ) T(r,\Phi_{s,ij})=O(\log r) T ( r , Φ s , ij ) = O ( log r ) , where s = 1 , n s=1,n s = 1 , n . Let
H i j = n − 1 Φ 0 , j − n Φ 1 , i j + 1 Φ n , i j , (70) H_{ij}=\frac{n-1}{\Phi_{0,j}}-\frac{n}{\Phi_{1,ij}}+\frac{1}{\Phi_{n,ij}}, \tag*{(70)} H ij = Φ 0 , j n − 1 − Φ 1 , ij n + Φ n , ij 1 , (70)
where i , j = 1 , 2 , … , m i,j=1,2,\ldots,m i , j = 1 , 2 , … , m . Since T ( r , Φ 0 , i ) = O ( log r ) T(r,\Phi_{0,i})=O(\log r) T ( r , Φ 0 , i ) = O ( log r ) and T ( r , Φ s , i j ) = O ( log r ) T(r,\Phi_{s,ij})=O(\log r) T ( r , Φ s , ij ) = O ( log r ) , where s = 1 , n s=1,n s = 1 , n , it follows that T ( r , H i j ) = O ( log r ) T(r,H_{ij})=O(\log r) T ( r , H ij ) = O ( log r ) .
Now if z 0 ∈ S ~ 0 z_0\in\widetilde{S}_0 z 0 ∈ S 0 , then using (59) –(61) to (70) , we get H i j ( z 0 ) = 0 H_{ij}(z_0)=0 H ij ( z 0 ) = 0 . Again if z 0 ∈ S ~ ∞ z_0\in\widetilde{S}_\infty z 0 ∈ S ∞ , then using (67) –(69) to (70) , we get H i j ( z 0 ) = 0 H_{ij}(z_0)=0 H ij ( z 0 ) = 0 . Since T ( r , H i j ) = O ( log r ) T(r,H_{ij})=O(\log r) T ( r , H ij ) = O ( log r ) , we can conclude that the set S ~ 0 ∪ S ~ ∞ \widetilde{S}_0\cup\widetilde{S}_\infty S 0 ∪ S ∞ is algebraic, which is absurd. Hence H i j ≡ 0 H_{ij}\equiv0 H ij ≡ 0 and so from (29) , we get
n − 1 Φ 0 , j − n Φ 1 , i j + 1 Φ n , i j ≡ 0. (71) \frac{n-1}{\Phi_{0,j}}-\frac{n}{\Phi_{1,ij}}+\frac{1}{\Phi_{n,ij}}\equiv0. \tag*{(71)} Φ 0 , j n − 1 − Φ 1 , ij n + Φ n , ij 1 ≡ 0. (71)
Now if z 0 ∈ S ~ 0 z_0\in\widetilde{S}_0 z 0 ∈ S 0 , then using (59) –(61) to (71) , we get B 0 ( z 0 ) = 0 B_0(z_0)=0 B 0 ( z 0 ) = 0 . Again if z 0 ∈ S ~ ∞ z_0\in\widetilde{S}_\infty z 0 ∈ S ∞ , then using (67) –(69) to (71) , we get B ∞ ( z 0 ) = 0 B_\infty(z_0)=0 B ∞ ( z 0 ) = 0 . Therefore if z 0 ∈ S ~ 0 z_0\in\widetilde{S}_0 z 0 ∈ S 0 , then from (59) –(61) , we get
1 Φ 0 , j ( z ) = 1 A 0 + C 0 A 0 2 φ 1 + O ( φ 1 2 ) (72) \frac{1}{\Phi_{0,j}(z)}=\frac{1}{A_0}+\frac{C_0}{A_0^2}\varphi_1+O(\varphi_1^2) \tag*{(72)} Φ 0 , j ( z ) 1 = A 0 1 + A 0 2 C 0 φ 1 + O ( φ 1 2 ) (72)
and
1 ϕ s , i j ( z ) = ( p s ) ( p s ( p + 1 ) A 0 ) + C 0 A 0 2 φ 1 + O ( φ 1 2 ) , (73) \frac{1}{\phi_{s,ij}(z)}=\frac{(p_s)}{(p_s^{(p+1)}A_0)}+\frac{C_0}{A_0^2}\varphi_1+O(\varphi_1^2), \tag*{(73)} ϕ s , ij ( z ) 1 = ( p s ( p + 1 ) A 0 ) ( p s ) + A 0 2 C 0 φ 1 + O ( φ 1 2 ) , (73)
where s = 1 , n s=1,n s = 1 , n and if z 0 ∈ S ~ ∞ z_0\in\widetilde{S}_\infty z 0 ∈ S ∞ , then from (67) –(69) , we get
1 Φ 0 , j ( z ) = 1 A ∞ + C ∞ A ∞ 2 u 1 + O ( u 1 2 ) . (74) \frac{1}{\Phi_{0,j}(z)}=\frac{1}{A_\infty}+\frac{C_\infty}{A_\infty^2}u_1+O(u_1^2). \tag*{(74)} Φ 0 , j ( z ) 1 = A ∞ 1 + A ∞ 2 C ∞ u 1 + O ( u 1 2 ) . (74)
and
1 Φ s , i j ( z ) = ( p s ) ( p s ( p + 1 ) A ∞ ) + C ∞ A ∞ 2 u 1 + O ( u 1 2 ) , (75) \frac{1}{\Phi_{s,ij}(z)}=\frac{(p_s)}{(p_s^{(p+1)}A_\infty)}+\frac{C_\infty}{A_\infty^2}u_1+O(u_1^2), \tag*{(75)} Φ s , ij ( z ) 1 = ( p s ( p + 1 ) A ∞ ) ( p s ) + A ∞ 2 C ∞ u 1 + O ( u 1 2 ) , (75)
where s = 1 , n s=1,n s = 1 , n . Let
G s , i j : = ∂ z i ( 1 Φ 0 , j − 1 Φ s , i j ) , (76) G_{s,ij}:=\partial_{z_i}\left(\frac{1}{\Phi_{0,j}}-\frac{1}{\Phi_{s,ij}}\right), \tag*{(76)} G s , ij := ∂ z i ( Φ 0 , j 1 − Φ s , ij 1 ) , (76)
where s = 1 , n s=1,n s = 1 , n .
Since T ( r , Φ 0 , i ) = O ( log r ) T(r,\Phi_{0,i})=O(\log r) T ( r , Φ 0 , i ) = O ( log r ) and T ( r , Φ s , i j ) = O ( log r ) T(r,\Phi_{s,ij})=O(\log r) T ( r , Φ s , ij ) = O ( log r ) , where s = 1 , n s=1,n s = 1 , n , it follows that T ( r , G s , i j ) = O ( log r ) T(r,G_{s,ij})=O(\log r) T ( r , G s , ij ) = O ( log r ) . Now using (72) –(75) to (76) , we deduce that ∂ z i ( G s , i j ( z 0 ) ) = 0 \partial_{z_i}(G_{s,ij}(z_0))=0 ∂ z i ( G s , ij ( z 0 )) = 0 for s = 1 , n s=1,n s = 1 , n . Consequently, the set S ~ 0 ∪ S ~ ∞ \widetilde{S}_0\cup\widetilde{S}_\infty S 0 ∪ S ∞ is algebraic, which is absurd. Hence G s , i j ≡ 0 G_{s,ij}\equiv0 G s , ij ≡ 0 for s = 1 , n s=1,n s = 1 , n and so
1 Φ 0 , j − 1 Φ s , i j = s ( p + 1 ) A = s M (77) \frac{1}{\Phi_{0,j}}-\frac{1}{\Phi_{s,ij}}=\frac{s}{(p+1)A}=sM \tag*{(77)} Φ 0 , j 1 − Φ s , ij 1 = ( p + 1 ) A s = s M (77)
is a constant, where M ∈ C ∖ { 0 } M\in\mathbb{C}\setminus\{0\} M ∈ C ∖ { 0 } . We know that ∞ \infty ∞ is a Picard exceptional value of Φ 0 , j \Phi_{0,j} Φ 0 , j , Φ 1 , i j \Phi_{1,ij} Φ 1 , ij and Φ n , i j \Phi_{n,ij} Φ n , ij . Therefore from (76) , it is easy to verify that 1 / M 1/M 1/ M and 1 / n M 1/nM 1/ n M are Picard exceptional values of Φ 0 , j \Phi_{0,j} Φ 0 , j . Consequently Φ 0 , j \Phi_{0,j} Φ 0 , j is a constant. Now from (77) , one can conclude that Φ 1 , i j \Phi_{1,ij} Φ 1 , ij and Φ n , i j \Phi_{n,ij} Φ n , ij are both constants.
On the other hand from (72) and (74) , we can deduce that C 0 = 0 C_0=0 C 0 = 0 and C ∞ = 0 C_\infty=0 C ∞ = 0 and so by (58) and (66) , we have a p + 1 / a p = − b p + 1 / b p a_{p+1}/a_p=-b_{p+1}/b_p a p + 1 / a p = − b p + 1 / b p and c p + 1 / c p = − d p + 1 / d p c_{p+1}/c_p=-d_{p+1}/d_p c p + 1 / c p = − d p + 1 / d p . Thus
A 0 = 2 a p + 1 / a p and A ∞ = 2 c p + 1 / c p . (78) A_0=2a_{p+1}/a_p\quad\text{and}\quad A_\infty=2c_{p+1}/c_p. \tag*{(78)} A 0 = 2 a p + 1 / a p and A ∞ = 2 c p + 1 / c p . (78)
Let z 0 ∈ S ~ 0 z_0 \in\widetilde{S}_0 z 0 ∈ S 0 . The coefficients of φ 1 2 \varphi_1^2 φ 1 2 in the expansions given by (59) –(61) of Φ 0 , i \Phi_{0,i} Φ 0 , i , Φ 1 , i j \Phi_{1,ij} Φ 1 , ij and Φ n , i j \Phi_{n,ij} Φ n , ij about z 0 z_0 z 0 respectively, yield a homogeneous linear equation system. It is easy to see that ( a p + 1 / a p ) 3 = 0 (a_{p+1}/a_p)^3=0 ( a p + 1 / a p ) 3 = 0 . Then from (78) , we have A 0 ( z 0 ) = 0 A_0(z_0)=0 A 0 ( z 0 ) = 0 , contradicting the choice of z 0 z_0 z 0 (see proof of Theorem 4 [16 ] ).
Let z 0 ∈ S ~ ∞ z_0 \in\widetilde{S}_\infty z 0 ∈ S ∞ . The coefficients of u 1 2 u_1^2 u 1 2 in the expansions given by (67) –(69) of Φ 0 , i \Phi_{0,i} Φ 0 , i , Φ 1 , i j \Phi_{1,ij} Φ 1 , ij and Φ n , i j \Phi_{n,ij} Φ n , ij about z 0 z_0 z 0 respectively, yield a homogeneous linear equation system. It is easy to see that ( a p + 1 / a p ) 3 = 0 (a_{p+1}/a_p)^3=0 ( a p + 1 / a p ) 3 = 0 . Then from (78) , we have A ∞ ( z 0 ) = 0 A_\infty(z_0)=0 A ∞ ( z 0 ) = 0 , contradicting the choice of z 0 z_0 z 0 . This completes the proof. □
Compliance of Ethical Standards:
Conflict of Interest. The authors declare that there is no conflict of interest regarding the publication of this paper.
Data availability statement. Data sharing not applicable to this article as no data sets were generated or analysed during the current study.
Funding. This work was supported by the National Natural Science Foundation of China (12161074) and the Foundation of Education Department of Jiangxi (GJJ190876, GJJ202303, GJJ201813, GJJ201343) of China, and the Talent Introduction Research Foundation of Suqian University (106-CK00042/028), Suqian Sci & Tech Program Grant No. M202206 and sponsored by Qing Lan Project of Jiangsu province and Suqian Talent Xiongying Plan of Suqian.
References