Greatest common divisors of analytic functions and Nevanlinna theory on algebraic tori

Aaron Levin; Julie Tzu-Yueh Wang

doi:10.1515/crelle-2019-0033

Open Access Published by De Gruyter November 9, 2019

Greatest common divisors of analytic functions and Nevanlinna theory on algebraic tori

Aaron Levin and Julie Tzu-Yueh Wang

From the journal Journal für die reine und angewandte Mathematik (Crelles Journal)

https://doi.org/10.1515/crelle-2019-0033

Abstract

We study upper bounds for the counting function of common zeros of two meromorphic functions in various contexts. The proofs and results are inspired by recent work involving greatest common divisors in Diophantine approximation, to which we introduce additional techniques to take advantage of the stronger inequalities available in Nevanlinna theory. In particular, we prove a general version of a conjectural “asymptotic gcd” inequality of Pasten and the second author, and consider moving targets versions of our results.

1 Introduction

We prove upper bounds for the counting function of common zeros of two meromorphic functions in various contexts. A starting point for such results (in a geometric formulation) comes from the study of holomorphic curves in semi-abelian varieties by Noguchi, Winkelmann, and Yamanoi, who proved the following:

Theorem 1.1 (Noguchi, Winkelmann, Yamanoi [14, Theorem5.1] (see also [13, Sectiom 6.5])).

Let f : C → A be a holomorphic map to a semi-abelian variety A with Zariski dense image. Let Y be a closed subscheme of A with codim ⁡ Y ≥ 2 and let ε > 0 .

Then

N f ⁢ ( Y , r ) ≤ exc ε ⁢ T f ⁢ ( r ) .
There exists a compactification A ¯ of A , independent of ε , such that for the Zariski closure Y ¯ of Y in A ¯ ,

T Y ¯ , f ⁢ ( r ) ≤ exc ε ⁢ T f ⁢ ( r ) .

Here N f ⁢ ( Y , r ) is a counting function associated to f and Y, T Y ¯ , f ⁢ ( r ) is a Nevanlinna characteristic (or height) function associated to f and Y ¯ , and T f ⁢ ( r ) is any characteristic function associated to an appropriate ample line bundle (see [14] for more discussion and Section 2 for the relevant definitions from Nevanlinna theory). The notation ≤ exc means that the estimate holds for all r outside a set of finite Lebesgue measure, possibly depending on ε.

More generally, Noguchi, Winkelmann, and Yamanoi proved a result for k-jet lifts of holomorphic maps to semi-abelian varieties. The case when A is an abelian variety was proved by Yamanoi [26].

A first goal of our work is to obtain a new proof of Theorem 1.1 when A = ( ℂ * ) n is the complex algebraic torus. In this case we obtain the following version of Theorem 1.1, where in part b one may use any nonsingular projective toric compactification.

Theorem 1.2.

Let Y be a closed subscheme of ( C * ) n of codimension at least 2. Let g = ( g 1 , … , g n ) be a holomorphic map from C to ( C * ) n with Zariski dense image (equivalently, g 1 , … , g n are entire functions without zeros, and g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ C for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } ). Let ε > 0 .

Then

N 𝐠 ⁢ ( Y , r ) ≤ exc ε ⁢ T 𝐠 ⁢ ( r ) .
Let X be a nonsingular projective toric compactification of ( ℂ * ) n . Let Y ¯ be the Zariski closure of Y in X , and suppose that Y ¯ is in general position with the boundary of ( ℂ * ) n in X . Then

T Y ¯ , 𝐠 ⁢ ( r ) ≤ exc ε ⁢ T 𝐠 ⁢ ( r ) .

Here, we say that Y ¯ ⊂ X is in general position with the boundary X ∖ ( ℂ * ) n if Y ¯ does not contain any point of intersection of n distinct irreducible components of X ∖ ( ℂ * ) n .

Alternatively, the counting function N 𝐠 ⁢ ( Y , r ) can be expressed as a counting function of common zeros of functions obtained by composing 𝐠 with polynomials generating the defining ideal of Y. In this context, we prove the following reformulation of Theorem 1.2 (which is again a special case of Noguchi–Winkelmann–Yamanoi’s Theorem 1.1):

Theorem 1.3.

Let F , G ∈ C ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials, and let g 1 , … , g n be entire functions without zeros. Assume that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ C for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } . Let ε > 0 .

Then

N gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .
If F and G do not both vanish at the origin ( 0 , … , 0 ) , then

T gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .

Here, as will be discussed in more detail later, N gcd ⁢ ( f , g , r ) and T gcd ⁢ ( f , g , r ) are analogues of the greatest common divisor of two integers. The function N gcd ⁢ ( f , g , r ) is simply the counting function of common zeros of f and g.

A particularly simple consequence of Theorem 1.3 is the following result.

Corollary 1.4.

Let f and g be multiplicatively independent non-constant entire functions without zeros. Then for any ε > 0 ,

N gcd ⁢ ( f - 1 , g - 1 , r ) ≤ exc ε ⁢ max ⁡ { T f ⁢ ( r ) , T g ⁢ ( r ) } .

An elementary proof of Corollary 1.4 was previously obtained in [16] by adapting the number-theoretical arguments of [5] to Nevanlinna theory.

It is natural to try to extend Theorem 1.3 (and Corollary 1.4) in an appropriate way to entire functions possibly admitting zeros, or more generally, to meromorphic functions. In this direction, by adapting the ideas and methods of Silverman [21] to a pair of meromorphic functions, the following estimate was established by Pasten and the second author in [16, Proposition 7.2] under the assumption of the Nevanlinna theoretic analogue of Vojta’s conjecture [24, Conjecture 15.2] for a blow-up of ℙ 1 × ℙ 1 at a single point: If f and g are algebraically independent complex meromorphic functions, then for all ε > 0 ,

N gcd ( f - 1 , g - 1 , r ) ≤ exc ε max { T f ( r ) , T g ( r ) } + 1 1 + ε 4 ( N f ( 0 , r ) + N g ( 0 , r )

+ N f ( ∞ , r ) + N g ( ∞ , r ) ) .

Again under the assumption of [24, Conjecture 15.2] for a blow-up of ℙ 1 × ℙ 1 at finitely many points, the following asymptotic gcd estimate is formulated in [16, Proposition 7.4]: If f and g are multiplicatively independent meromorphic functions, then for any ε > 0 , there exists n 0 such that for all n ≥ n 0 ,

(1.1) N gcd ⁢ ( f n - 1 , g n - 1 , r ) ≤ exc ε ⁢ max ⁡ { T f n ⁢ ( r ) , T g n ⁢ ( r ) } .

A second goal of this article is to prove (unconditional) asymptotic gcd estimates in a more general context:

Theorem 1.5.

Let F , G ∈ C ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials such that not both of them vanish at ( 0 , … , 0 ) . Let g 1 , … , g n be meromorphic functions such that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ C for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } . Then for any ε > 0 , there exists k 0 such that for all k ≥ k 0 ,

one has

N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } ,
one has

T gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) }

if g 1 , … , g n are entire functions.

In particular, we prove the conjectured inequality (1.1):

Corollary 1.6.

Let f and g be multiplicatively independent meromorphic functions. Then for any ε > 0 , there exists k 0 such that for all k ≥ k 0 ,

N gcd ⁢ ( f k - 1 , g k - 1 , r ) ≤ exc ε ⁢ max ⁡ { T f k ⁢ ( r ) , T g k ⁢ ( r ) } .

When f and g are algebraically independent meromorphic functions, Corollary 1.6 was recently obtained by Guo and the second author in [10] with ε replaced by 1 2 + ε . We refer to [16] for further discussion of gcd problems for both complex and non-archimedean meromorphic functions.

Even in the special case when g 1 , … , g n are complex polynomials, Theorem 1.5 gives new results.

Remark 1.7.

Let F , G ∈ ℂ ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials such that not both of them vanish at ( 0 , … , 0 ) . Let g 1 , … , g n ∈ ℂ ⁢ [ z ] be complex polynomials such that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ ℂ for any index set ( i 1 , … , i n ) ∈ ℤ n ∖ { ( 0 , … , 0 ) } . It is elementary that in this case,

(1.2) N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r )

= ( deg ⁡ gcd ⁡ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) ) ) ⁢ log ⁡ r + O ⁢ ( 1 )

and

max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } = k ⁢ max 1 ≤ i ≤ n ⁡ { deg ⁡ g i } ⁢ log ⁡ r + O ⁢ ( 1 ) ,

where the gcd on the right-hand side of (1.2) is the greatest common divisor in the polynomial ring ℂ ⁢ [ z ] . Then Theorem 1.5 implies that for any ε > 0 , there exists k 0 such that for all k ≥ k 0 ,

deg ⁡ gcd ⁡ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) ) < ε ⁢ k .

More generally, Theorem 1.5 gives a similar statement for rational functions g 1 , … , g n ∈ ℂ ⁢ ( z ) . When n > 2 , the only previous result in this direction appears to be a result of Ostafe [15, Theorem 1.3], which considers special polynomials such as F = x 1 ⁢ ⋯ ⁢ x r - 1 , G = x r + 1 ⁢ ⋯ ⁢ x n - 1 , but proves a stronger uniform bound independent of k. It is noted in [15] that it appears to be difficult to extend the techniques used there to obtain results for general F and G. In the n = 2 case, previous results include the original theorem of Ailon and Rudnick [1] in this setting and extensions of Ostafe [15] (both with uniform bounds).

We will use Theorem 1.3 and Theorem 1.5 to solve the following quotient problem, which can be considered an analogue of the “Hadamard quotient theorem” for recurrence sequences proved by van der Poorten (see [19] and [22], and see [4] and [27] for an overview of the existing improvements). To state the result we make the following definitions: Let G ∈ ℂ ⁢ [ x 1 , … , x n ] be a nonconstant polynomial such that G ⁢ ( 0 , … , 0 ) ≠ 0 . Since G has a nonzero constant term, after arranging the index set in some order, we may write

G = a 𝐢 ⁢ ( 0 ) + ∑ j = 1 ℓ a 𝐢 ⁢ ( j ) ⁢ 𝐱 𝐢 ⁢ ( j ) ,

where a 𝐢 ⁢ ( j ) ≠ 0 for 0 ≤ j ≤ ℓ . Then for entire functions g 1 , … , g n , we let

𝔤 G := ( 1 , 𝐠 𝐢 ⁢ ( 1 ) , … , 𝐠 𝐢 ⁢ ( ℓ ) ) : ℂ → ℙ ℓ .

For functions h i : ℝ ≥ 0 → ℝ ≥ 0 , i = 1 , 2 , we write h 1 ⁢ ( r ) ≍ h 2 ⁢ ( r ) if there exist positive numbers a and b such that a ⁢ h 1 ⁢ ( r ) ≤ h 2 ⁢ ( r ) ≤ b ⁢ h 1 ⁢ ( r ) for all sufficiently large r.

Corollary 1.8.

Let F , G ∈ C ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials such that G ⁢ ( 0 , … , 0 ) ≠ 0 . Let g 1 , … , g n be entire functions such that T g G ⁢ ( r ) ≍ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .

If F ⁢ ( g 1 k , … , g n k ) / G ⁢ ( g 1 k , … , g n k ) is an entire functions for infinitely many positive integers k , or
g 1 , … , g n are entire functions without zeros and F ⁢ ( g 1 , … , g n ) / G ⁢ ( g 1 , … , g n ) is an entire function,

then there exists an index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } such that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∈ C .

Remark 1.9.

The case where ( g 1 k - 1 ) / ( g 2 k - 1 ) is entire for infinitely many positive integers k (i.e., F = - 1 + x 1 and G = - 1 + x 2 ) is treated in [10] under the assumption that T g 1 ⁢ ( r ) ≍ T g 2 ⁢ ( r ) . The case G = 1 + a 1 ⁢ x 1 + … + a n ⁢ x n , F = b 0 + b 1 ⁢ x n + 1 + … + b m ⁢ x n + m ( a i ⁢ b j ≠ 0 for 1 ≤ i ≤ n and 0 ≤ j ≤ m ) with the assumption that

max ⁡ { T g 1 ⁢ ( r ) , … , T g n ⁢ ( r ) } ≍ max ⁡ { T g n + 1 ⁢ ( r ) , … , T g n + m ⁢ ( r ) }

is studied in [8] when a i and b j are constants and in [9] when a i and b j are “small functions”, with a proof obtained by adapting the argument of [4], where Corvaja and Zannier proved a stronger version of the Hadamard quotient theorem through a sophisticated application of Schmidt’s subspace theorem. The result in [8] can be obtained from Corollary 1.8 since in this situation

T G ⁢ ( g 1 , … , g n ) ⁢ ( r ) = T 1 + a 1 ⁢ g 1 + … + a n ⁢ g n ⁢ ( r ) = T ( 1 , g 1 , … , g n ) ⁢ ( r ) + O ⁢ ( 1 )

and

max ⁡ { T g 1 ⁢ ( r ) , … , T g n ⁢ ( r ) } ≤ T ( 1 , g 1 , … , g n ) ⁢ ( r ) ≤ n ⁢ max ⁡ { T g 1 ⁢ ( r ) , … , T g n ⁢ ( r ) } .

Remark 1.10.

Part b of Corollary 1.8 follows from a more general result of Corvaja and Noguchi [3] for semi-abelian varieties (under a related growth condition).

Remark 1.11.

The growth condition on characteristic functions is essential, at least in part (b). For instance, let g ⁢ ( z ) = exp ⁡ ( 2 ⁢ π ⁢ i ⁢ z ) and f ⁢ ( z ) = exp ⁡ ( 2 ⁢ π ⁢ i ⁢ p ⁢ ( z ) ) , where p ⁢ ( x ) ∈ ℤ ⁢ [ x ] is a polynomial of degree at least 2. Then ( g - 1 ) | ( f - 1 ) , but f and g are algebraically independent.

Finally, we will consider the case when the coefficients of F and G are functions. More precisely, let 𝐠 be a holomorphic map from ℂ to ℙ n . We say a meromorphic function a is a small function with respect to 𝐠 if T a ⁢ ( r ) = o ⁢ ( T 𝐠 ⁢ ( r ) ) . Let K 𝐠 be the field containing all small functions with respect to 𝐠 . Let F , G ∈ K 𝐠 ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials. We establish results analogous to Theorem 1.3, Theorem 1.5, and Corollary 1.8 in this situation. The results in this situation go beyond the scope of Theorem 1.1. Moreover, it is of special interest as an analogy to linear recurrences over number fields (see [7]) to study the gcd of polynomials in ℂ ⁢ ( z ) ⁢ [ x 1 , … , x n ] evaluated at ( e λ 1 ⁢ z , … , e λ n ⁢ z ) , where λ 1 , … , λ n ∈ ℂ . The results are stated in Section 5.

The proofs of Theorem 1.2 and Theorem 1.3 are inspired by recent work of the first author [12] on analogous inequalities in Diophantine approximation involving greatest common divisors of multivariable polynomials evaluated at S-unit arguments. The proofs of Theorem 1.5 and Corollary 1.6 go well beyond what is possible in the arithmetic setting, as they take advantage of inequalities involving truncated counting functions and Wronskian terms, whose analogues are still largely conjectural in Diophantine approximation.

It will be useful to give a brief overview of the analogous results in Diophantine approximation involving greatest common divisors. The first general result in this direction is due to Bugeaud, Corvaja, and Zannier, who in 2003 proved:

Theorem 1.12 (Bugeaud, Corvaja, Zannier [2]).

Let a , b ∈ Z be multiplicatively independent integers. Then for every ε > 0 ,

log ⁡ gcd ⁡ ( a n - 1 , b n - 1 ) ≤ ε ⁢ n

for all but finitely many positive integers n.

Let k be a number field, M k the set of places of k, S ⊂ M k a finite set of places containing the archimedean places, 𝒪 k , S the ring of S-integers of k, and 𝒪 k , S * the group of S-units. Let 𝔾 m n be the n-dimensional algebraic torus and 𝔾 m n ⁢ ( 𝒪 k , S ) = ( 𝒪 k , S * ) n . To state more general results, for α , β ∈ k , we define the generalized logarithmic greatest common divisor and the gcd counting function (depending on a choice of S), respectively, by

(1.3) log gcd ( α , β ) = - ∑ v ∈ M k log - max { | α | v , | β | v } = h ( [ 1 : α : β ] ) - h ( [ α : β ] ) ,

(1.4) N gcd , S ⁢ ( α , β ) = - ∑ v ∈ M k ∖ S log - ⁡ max ⁡ { | α | v , | β | v } ,

where | ⋅ | v is an appropriately normalized absolute value associated to v, h is the standard (absolute logarithmic) Weil height on projective space, and log - ⁡ x = min ⁡ { 0 , log ⁡ x } (see [12] for details). After work of Corvaja and Zannier [5] in the 2-dimensional case, the first author generalized Theorem 1.12 as follows.

Theorem 1.13 (Corvaja, Zannier [5] (for case n = 2 ), Levin [12] (for case n > 2 )).

Let F , G ∈ k ⁢ [ x 1 , … , x n ] be coprime polynomials. Let ε > 0 .

There exists a finite union Z of translates of proper algebraic subgroups of 𝔾 m n such that

N gcd , S ⁢ ( F ⁢ ( u 1 , … , u n ) , G ⁢ ( u 1 , … , u n ) ) < ε ⁢ max ⁡ { h ⁢ ( u 1 ) , h ⁢ ( u 2 ) , … , h ⁢ ( u n ) }

for all ( u 1 , … , u n ) ∈ 𝔾 m n ⁢ ( 𝒪 k , S ) ∖ Z .
Suppose additionally that not both of F and G vanish at the origin ( 0 , 0 , … , 0 ) . Then there exists a finite union Z of translates of proper algebraic subgroups of 𝔾 m n such that

log ⁡ gcd ⁡ ( F ⁢ ( u 1 , … , u n ) , G ⁢ ( u 1 , … , u n ) ) < ε ⁢ max ⁡ { h ⁢ ( u 1 ) , h ⁢ ( u 2 ) , … , h ⁢ ( u n ) }

for all ( u 1 , … , u n ) ∈ 𝔾 m n ⁢ ( 𝒪 k , S ) ∖ Z .

One can also give a geometric version of Theorem 1.13.

Theorem 1.14.

Let G m n ⊂ X be a nonsingular projective toric variety of dimension n, and let Y be a closed subscheme of X of codimension at least 2, both defined over a number field k. Let A be a big divisor on X. Let S be a finite set of places of k containing the archimedean places. Let ε > 0 .

There exists a finite union Z of translates of proper algebraic subgroups of 𝔾 m n such that

N Y , S ⁢ ( P ) ≤ ε ⁢ h A ⁢ ( P ) + O ⁢ ( 1 )

for all P ∈ 𝔾 m n ⁢ ( 𝒪 k , S ) ∖ Z ⊂ X ⁢ ( k ) .
Suppose that Y is in general position with the boundary of 𝔾 m n in X . Then there exists a finite union Z of translates of proper algebraic subgroups of 𝔾 m n such that

h Y ⁢ ( P ) ≤ ε ⁢ h A ⁢ ( P ) + O ⁢ ( 1 )

for all P ∈ 𝔾 m n ⁢ ( 𝒪 k , S ) ∖ Z ⊂ X ⁢ ( k ) .

All of the described arithmetic inequalities rely on Schmidt’s Subspace Theorem in Diophantine approximation. The proofs of Theorem 1.2 and Theorem 1.3 are adapted from the proofs in [12], and use the analogue of Schmidt’s Subspace Theorem in Nevanlinna theory: Cartan’s second main theorem (in a form due to Vojta). The proofs of the asymptotic gcd inequalities of Theorem 1.5 and Corollary 1.6 combine techniques from [12] along with new ideas in order to take advantage of the stronger error terms known in the second main theorem in Nevanlinna theory.

After giving the relevant background material in the next section, in Section 3 we prove the key technical results underlying the paper. In Section 4, we apply these results to prove our main theorems. In Section 5, we review some fundamental results in Nevanlinna theory with moving targets, and we state and prove the main theorems when the coefficients of F and G are small functions.

2 Background material

2.1 Nevanlinna theory over ℂ

We will set up some notation and definitions in Nevanlinna theory for complex meromorphic functions and recall some basic results. We refer to [11, Chapter VI ] or [18, Chapter 1] for details.

Let f be a meromorphic function and z ∈ ℂ . Define v z ⁢ ( f ) := ord z ⁢ ( f ) ,

v z + ⁢ ( f ) := max ⁡ { 0 , v z ⁢ ( f ) } , v z - ⁢ ( f ) := - min ⁡ { 0 , v z ⁢ ( f ) } .

Let n f ⁢ ( ∞ , r ) (respectively, n f ( Q ) ⁢ ( ∞ , r ) ) denote the number of poles of f in { z : | z | ≤ r } , counting multiplicity (respectively, ignoring multiplicity larger than Q ∈ ℕ ). The counting function and truncated counting function of f of order Q at ∞ are defined respectively by

N f ⁢ ( ∞ , r ) := ∫ 0 r n f ⁢ ( ∞ , t ) - n f ⁢ ( ∞ , 0 ) t ⁢ 𝑑 t + n f ⁢ ( ∞ , 0 ) ⁢ log ⁡ r

= ∑ 0 < | z | ≤ r v z - ⁢ ( f ) ⁢ log ⁡ | r z | + v 0 - ⁢ ( f ) ⁢ log ⁡ r ,

and

N f ( Q ) ⁢ ( ∞ , r ) := ∫ 0 r n f ( Q ) ⁢ ( ∞ , t ) - n f ( Q ) ⁢ ( ∞ , 0 ) t ⁢ 𝑑 t + n f ( Q ) ⁢ ( ∞ , 0 ) ⁢ log ⁡ r

= ∑ 0 < | z | ≤ r min ⁡ { Q , v z - ⁢ ( f ) } ⁢ log ⁡ | r z | + min ⁡ { Q , v 0 - ⁢ ( f ) } ⁢ log ⁡ r .

Then define the counting function N f ⁢ ( r , a ) and the truncated counting function N f ( Q ) ⁢ ( r , a ) for a ∈ ℂ as

N f ⁢ ( a , r ) := N 1 / ( f - a ) ⁢ ( r , ∞ ) and N f ( Q ) ⁢ ( a , r ) := N 1 / ( f - a ) ( Q ) ⁢ ( ∞ , r ) .

The proximity function m f ⁢ ( ∞ , r ) is defined by

m f ⁢ ( ∞ , r ) := ∫ 0 2 ⁢ π log + ⁡ | f ⁢ ( r ⁢ e i ⁢ θ ) | ⁢ d ⁢ θ 2 ⁢ π ,

where log + ⁡ x = max ⁡ { 0 , log ⁡ x } for x ≥ 0 . For any a ∈ ℂ , the proximity function m f ⁢ ( a , r ) is defined by

m f ⁢ ( a , r ) := m 1 / ( f - a ) ⁢ ( ∞ , r ) .

Finally, the characteristic function is defined by

T f ⁢ ( r ) := m f ⁢ ( ∞ , r ) + N f ⁢ ( ∞ , r ) .

We recall the following version of Jensen’s formula.

Lemma 2.1.

Let f be a meromorphic function on { z : | z | ≤ r } which is not the zero function. Then

∫ 0 2 ⁢ π log ⁡ | f ⁢ ( r ⁢ e i ⁢ θ ) | ⁢ d ⁢ θ 2 ⁢ π = N f ⁢ ( r , 0 ) - N f ⁢ ( r , ∞ ) + log ⁡ | c f | ,

where c f is the leading coefficient of f expanded as Laurent series in z, i.e., f = c f ⁢ z m + ⋯ with c f ≠ 0 .

Jensen’s formula implies the first main theorem of Nevanlinna theory.

Theorem 2.2 (First Main Theorem).

Let f be a non-constant meromorphic function on C . Then for every a ∈ C , and any positive real number r,

m f ⁢ ( a , r ) + N f ⁢ ( a , r ) = T f ⁢ ( r ) + O ⁢ ( 1 ) ,

where O ⁢ ( 1 ) is independent of r.

2.2 Nevanlinna theory for Cartier divisors

We recall some notation and properties from [24, Sections 9 and 12]. Let D be a Cartier divisor on a complex variety X. A Weil function for D is a function λ D : ( X ∖ Supp ⁡ D ) ⁢ ( ℂ ) → ℝ such that for all x ∈ X ⁢ ( ℂ ) , there is an open neighborhood U of x in X, a nonzero function f ∈ K ⁢ ( X ) such that D | U = ( f ) , and a continuous function α : U ⁢ ( ℂ ) → ℝ such that

λ D ⁢ ( x ) = - log ⁡ | f ⁢ ( x ) | + α ⁢ ( x )

for all x ∈ ( U ∖ Supp ) ⁢ ( ℂ ) . We note that when D is effective λ D can be extended to a function X → ℝ ∪ { ∞ } . Let f : ℂ → X be a holomorphic map whose image is not contained in the support of the divisor D on X. The proximity function of f with respect to D is defined by

m f ⁢ ( D , r ) = ∫ 0 2 ⁢ π λ D ⁢ ( f ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ d ⁢ θ 2 ⁢ π .

Let n f ⁢ ( D , t ) (respectively, n f ( Q ) ⁢ ( D , t ) ) be the number of zeros of ρ ∘ f inside { | z | < t } , counting multiplicity, (respectively, ignoring multiplicity larger than Q ∈ ℕ ) with ρ a local defining function for D. The counting function and truncated counting function of f of order Q at ∞ are defined, respectively, by

N f ⁢ ( D , r ) = ∫ 1 r n f ⁢ ( D , t ) t ⁢ 𝑑 t ,

and

N f ( Q ) ⁢ ( D , r ) = ∫ 1 r n f Q ⁢ ( D , t ) t ⁢ 𝑑 t .

The characteristic function relative to D is defined, up to O ⁢ ( 1 ) , by

T D , f ⁢ ( r ) = m f ⁢ ( D , r ) + N f ⁢ ( D , r ) .

The following is the first main theorem.

Theorem 2.3.

Let D and D ′ be Cartier divisors on X whose supports do not contain the image of f, and suppose that D ′ is linearly equivalent to D. Then

T D ′ , f ⁢ ( r ) = T D , f ⁢ ( r ) + O ⁢ ( 1 ) .

In particular, let D be a hypersurface in ℙ n ⁢ ( ℂ ) defined by a homogeneous polynomial F of degree d. The Weil function for D can be taken as

λ D ⁢ ( 𝐱 ) = - log ⁡ | F ⁢ ( 𝐱 ) | max { | x 0 | , … , | x n | } d

for 𝐱 = [ x 0 : ⋯ : x n ] ∈ ℙ n ( ℂ ) . Let 𝐟 : ℂ → ℙ n ⁢ ( ℂ ) be a holomorphic map and let ( f 0 , … , f n ) be a reduced representation of 𝐟 , i.e., f 0 , … , f n are entire functions on ℂ without common zeros such that for all z ∈ ℂ we have 𝐟 ( z ) = [ f 0 ( z ) : ⋯ : f n ( z ) ] . The characteristic function T 𝐟 ⁢ ( r ) is defined by

T 𝐟 ⁢ ( r ) = ∫ 0 2 ⁢ π log ⁡ ∥ 𝐟 ⁢ ( r ⁢ e i ⁢ θ ) ∥ ⁢ d ⁢ θ 2 ⁢ π ,

where ∥ 𝐟 ⁢ ( z ) ∥ = max ⁡ { | f 0 ⁢ ( z ) | , … , | f n ⁢ ( z ) | } . This definition is independent, up to an additive constant, of the choice of the reduced representation of 𝐟 . In this context, the First Main Theorem reads as follows.

Theorem 2.4.

Let f : C → P n ⁢ ( C ) be a holomorphic map, and let D be a hypersurface in P n ⁢ ( C ) of degree d. If f ⁢ ( C ) ⊄ D , then for r > 0 ,

d ⁢ T 𝐟 ⁢ ( r ) = m 𝐟 ⁢ ( r , D ) + N 𝐟 ⁢ ( r , D ) + O ⁢ ( 1 ) ,

where O ⁢ ( 1 ) is bounded independently of r.

We will make use of the following elementary inequality.

Lemma 2.5.

Let g 1 , … , g n be meromorphic functions. Let

𝐠 := [ 1 : g 1 : ⋯ : g n ] : ℂ → ℙ n .

Then

T 𝐠 ⁢ ( r ) ≤ ∑ i = 1 n T g i ⁢ ( r ) + O ⁢ ( 1 ) .

Proof.

Let H be the hyperplane in ℙ n defined by x 0 = 0 . Then clearly

λ H ⁢ ( 𝐠 ⁢ ( 𝐳 ) ) = log ⁡ max ⁡ { 1 , | g 1 ⁢ ( z ) | , … , | g n ⁢ ( z ) | } ≤ ∑ i = 1 n log + ⁡ | g i ⁢ ( z ) |

and

n 𝐠 ⁢ ( H , r ) ≤ ∑ i = 1 n n g i ⁢ ( ∞ , r ) .

After integrating, it follows from the definitions that

T 𝐠 ⁢ ( r ) = m 𝐠 ⁢ ( H , r ) + N 𝐠 ⁢ ( H , r ) + O ⁢ ( 1 )

≤ ∑ i = 1 n ( m g i ⁢ ( ∞ , r ) + N g i ⁢ ( ∞ , r ) ) + O ⁢ ( 1 )

≤ ∑ i = 1 n T g i ⁢ ( r ) + O ⁢ ( 1 ) . ∎

If H 1 , … , H q , q ≥ n + 1 , are hyperplanes in ℙ n ⁢ ( ℂ ) , then we say that H 1 , … , H q are in general position if for any subset I ⊂ { 1 , … , q } , | I | = n + 1 , the intersection ⋂ i ∈ I H i is empty.

We now recall the following general form of the second main theorem from [18, Theorem A3.1.3] which was proved by Vojta in [23, Theorem 1] and Ru in [17, Theorem 2.3].

Theorem 2.6.

Let f = ( f 0 , … , f n ) : C → P n ⁢ ( C ) be a holomorphic curve whose image is not contained in any proper linear subspace, and where f 0 , … , f n are entire functions with no common zeros. Let H 1 , … , H q be arbitrary hyperplanes in P n ⁢ ( C ) . Denote by W ⁢ ( f ) the Wronskian of f 0 , … , f n . Then for any ε > 0 , we have the inequality

∫ 0 2 ⁢ π max J ⁢ ∑ k ∈ J λ H k ⁢ ( f ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ d ⁢ θ 2 ⁢ π + N W ⁢ ( f ) ⁢ ( 0 , r ) ≤ exc ( n + 1 + ε ) ⁢ T f ⁢ ( r ) + o ⁢ ( T f ⁢ ( r ) ) ,

where the maximum is taken over all subsets J of { 1 , … , q } such that the hyperplanes H j , j ∈ J , are in general position.

Finally, we recall Cartan’s second main theorem with truncated counting functions. (See [18, Theorem A3.2.2].)

Theorem 2.7.

Let H 1 , … , H q be hyperplanes in P n ⁢ ( C ) in general position, and let f = ( f 0 , … , f n ) : C → P n ⁢ ( C ) be a holomorphic curve whose image is not contained in any proper linear subspace. Then for any ε > 0 , we have the following inequality:

( q - n - 1 - ε ) ⁢ T f ⁢ ( r ) ≤ exc ∑ i = 1 q N f ( n ) ⁢ ( H i , r ) .

2.3 Counting functions and proximity functions for closed subschemes

We first recall some basic properties of Weil functions associated to closed subschemes from [20, Section 2]. Let Y be a closed subscheme on a projective variety X defined over ℂ . Then one can associate to it a function

λ Y : X ⁢ ( ℂ ) ∖ Supp ⁡ ( Y ) → ℝ

satisfying some functorial properties (up to an O ⁢ ( 1 ) ) analogous to the arithmetic case as described in [20, Theorem 2.1]. Intuitively, for each point P ∈ X ⁢ ( ℂ ) ,

λ Y ⁢ ( P ) = - log ⁡ ( distance from P to Y ) .

The following lemma will be used to define Weil functions for closed subschemes.

Lemma 2.8.

Let Y be a closed subscheme of X. Then there exist effective divisors D 1 , … , D r such that

Y = ⋂ i = 1 r D i .

Proof.

See [20, Lemma 2.2]. ∎

Definition 2.9.

Let X be a projective variety over ℂ and let Y ⊂ X be a closed subscheme of X. We define the Weil function for Y as

λ Y = min i ⁡ { λ D i } ,

where Y = ⋂ D i (such D i exist according to the above lemma).

As usual, the Weil function λ Y is well defined up to O ⁢ ( 1 ) . Let f : ℂ → X be an analytic map. The proximity function of f with respect to Y is defined by

m f ⁢ ( Y , r ) = ∫ 0 2 ⁢ π λ Y ⁢ ( f ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ d ⁢ θ 2 ⁢ π ;

the counting function of f with respect to Y is defined by

N f ⁢ ( Y , r ) = ∫ 0 r n f ⁢ ( Y , t ) - n f ⁢ ( Y , 0 ) t ⁢ 𝑑 t + n f ⁢ ( Y , 0 ) ⁢ log ⁡ r ,

where n f ⁢ ( Y , t ) is the minimum of the number of zeros of ρ i ∘ f , 1 ≤ i ≤ m , inside { | z | ≤ t } , counting multiplicity, with ρ i , 1 ≤ i ≤ m , being local defining functions of Y. We note that the definition of the counting function extends without difficulty to the case when Y is a closed subscheme of a quasi-projective variety. The characteristic function of f with respect to Y is defined by

T Y , f ⁢ ( r ) = m f ⁢ ( Y , r ) + N f ⁢ ( Y , r ) .

The proximity functions and counting functions of closed subschemes of X satisfy additivity and functoriality properties as in the classical setting of Cartier divisors.

Recall the following definition which gives an analogue of the notion of gcd in the context of meromorphic functions. Let f 1 , … , f ℓ , ℓ ≥ 2 , be meromorphic functions. We write

n ⁢ ( { f i } 1 ≤ i ≤ ℓ , r ) := ∑ | z | ≤ r min ⁡ { v z + ⁢ ( f 1 ) , … , v z + ⁢ ( f ℓ ) } ,

N gcd ⁢ ( { f i } 1 ≤ i ≤ ℓ , r ) := ∫ 0 r n ⁢ ( { f i } 1 ≤ i ≤ ℓ , r ) - n ⁢ ( { f i } 1 ≤ i ≤ ℓ , 0 ) t ⁢ 𝑑 t + n ⁢ ( { f i } 1 ≤ i ≤ ℓ , 0 ) ⁢ log ⁡ r .

Let F , G ∈ ℂ ⁢ [ x 0 , … , x n ] be coprime homogeneous polynomials and let Y be the closed subscheme of ℙ n defined by the ideal I = ( F , G ) . Let 𝐟 = ( f 0 , … , f n ) : ℂ → ℙ n ⁢ ( ℂ ) , where f 0 , … , f n are entire functions with no common zeros. The counting function of Y agrees with the gcd counting function of F ⁢ ( 𝐟 ) and G ⁢ ( 𝐟 ) , i.e.,

N 𝐟 ⁢ ( Y , r ) = N gcd ⁢ ( F ⁢ ( 𝐟 ) , G ⁢ ( 𝐟 ) , r ) + O ⁢ ( 1 ) .

We also define gcd proximity and characteristic functions for meromorphic functions f and g, in analogy with (1.3) and (1.4). Let

m gcd ⁢ ( f , g , r ) := - ∫ 0 2 ⁢ π log - ⁡ max ⁡ { | f ⁢ ( r ⁢ e i ⁢ θ ) | , | g ⁢ ( r ⁢ e i ⁢ θ ) | } ⁢ d ⁢ θ 2 ⁢ π ,

T gcd ⁢ ( f , g , r ) := T [ 1 : f : g ] ⁢ ( r ) - T [ f : g ] ⁢ ( r ) .

We have the expected relationship between m gcd , N gcd , and T gcd .

Lemma 2.10.

Let f and g be meromorphic functions. Then

T gcd ⁢ ( f , g , r ) = m gcd ⁢ ( f , g , r ) + N gcd ⁢ ( f , g , r ) + O ⁢ ( 1 ) .

Proof.

Let h 0 be an entire function such that ( h 0 , f ⁢ h 0 , g ⁢ h 0 ) is a reduced representation of [ 1 : f : g ] , i.e., h 0 , f ⁢ h 0 , and g ⁢ h 0 are entire and have no common zeros. Similarly, let h 1 be an entire function such that [ f ⁢ h 0 / h 1 , g ⁢ h 0 / h 1 ] is a reduced representation of [ f : g ] . Then from the definitions,

T gcd ( f , g , r ) = ∫ 0 2 ⁢ π log max { | h 0 ( r e i ⁢ θ ) | , | ( f h 0 ) ( r e i ⁢ θ ) | , | ( g h 0 ) ( r e i ⁢ θ ) | } | d ⁢ θ 2 ⁢ π

- ∫ 0 2 ⁢ π log ⁡ max ⁡ { | ( f ⁢ h 0 / h 1 ) ⁢ ( r ⁢ e i ⁢ θ ) | , | ( g ⁢ h 0 / h 1 ) ⁢ ( r ⁢ e i ⁢ θ ) | } ⁢ d ⁢ θ 2 ⁢ π .

It is elementary that for any real numbers a , b > 0 ,

log ⁡ max ⁡ { 1 , a , b } max ⁡ { a , b } = - log - ⁡ max ⁡ { a , b } .

Then we find that

T gcd ⁢ ( f , g , r ) = - ∫ 0 2 ⁢ π log - ⁡ max ⁡ { | f ⁢ ( r ⁢ e i ⁢ θ ) | , | g ⁢ ( r ⁢ e i ⁢ θ ) | } ⁢ d ⁢ θ 2 ⁢ π + ∫ 0 2 ⁢ π log ⁡ | h 1 ⁢ ( r ⁢ e i ⁢ θ ) | ⁢ d ⁢ θ 2 ⁢ π

= m gcd ⁢ ( f , g , r ) + N h 1 ⁢ ( 0 , r ) + O ⁢ ( 1 ) ,

where the second line follows from the definition of m gcd and Jensen’s formula. To complete the proof, we note that from its definition, one easily finds that N h 1 ⁢ ( 0 , r ) = N gcd ⁢ ( f , g , r ) . ∎

2.4 Polynomial rings and monomial orderings

Let A = ℂ ⁢ [ x 0 , … , x n ] be the polynomial ring in n + 1 variables over ℂ . For 𝐢 = ( i 0 , … , i n ) ∈ ℕ n + 1 , we define

𝐱 𝐢 = x 0 i 1 ⁢ ⋯ ⁢ x n i n ,

and write

| 𝐢 | := i 0 + ⋯ + i n = deg ⁡ 𝐱 𝐢 .

Let

Mon = Mon ⁢ ( A ) = { 𝐱 𝐢 : 𝐢 ∈ ℕ n + 1 }

be the set of monomials of A. We note that we use the convention that ℕ is the set of nonnegative integers.

Recall that a monomial ordering on A is a total ordering > on Mon such that:

If 𝐱 𝐢 > 𝐱 𝐣 , then 𝐱 𝐢 ⁢ 𝐱 𝐤 > 𝐱 𝐣 ⁢ 𝐱 𝐤 for all 𝐢 , 𝐣 , 𝐤 ∈ ℕ n + 1 .
𝐱 𝐢 ≥ 1 for all 𝐢 ∈ ℕ n + 1 .

We describe two monomial orderings that we will use. The lexicographic ordering on A is the monomial ordering > lex such that 𝐱 𝐢 > lex 𝐱 𝐣 if the left-most nonzero entry of 𝐢 - 𝐣 is positive. Let 𝐮 ∈ ℕ n + 1 . We call 𝐮 a weight vector and define the weight order associated to 𝐮 as follows:

𝐱 𝐢 > 𝐮 𝐱 𝐣 if ⁢ 𝐮 ⋅ 𝐢 > 𝐮 ⋅ 𝐣 , or ⁢ 𝐮 ⋅ 𝐢 = 𝐮 ⋅ 𝐣 ⁢ and ⁢ 𝐱 𝐢 > lex 𝐱 𝐣 .

If > is a monomial ordering and F ∈ ℂ ⁢ [ x 0 , … , x n ] is a nonzero polynomial, we let TM ⁢ ( F ) denote the trailing monomial of F (the smallest monomial appearing in F with a nonzero coefficient). If F and G are nonzero polynomials, then TM ⁢ ( F ⁢ G ) = TM ⁢ ( F ) ⁢ TM ⁢ ( G ) . For every nonnegative integer m and subset T ⊂ ℂ ⁢ [ x 0 , … , x n ] , we let

T m = { P ∈ T : P is a homogeneous polynomial of degree m }

and let TM ⁢ ( T ) = { TM ⁢ ( F ) : F ∈ T } . We will use the following key lemma.

Lemma 2.11.

Let F 1 , F 2 ∈ k ⁢ [ x 0 , … , x n ] be coprime homogeneous polynomials of degree d, where k is a field. Let I 1 , I 2 , I 3 be the principal ideals I 1 = ( F 1 ) , I 2 = ( F 2 ) , I 3 = ( F 1 ⁢ F 2 ) . Let m ≥ d and let V = ( F 1 , F 2 ) m := k ⁢ [ x 0 , … , x n ] m ∩ ( F 1 , F 2 ) , a k-subspace of the ideal ( F 1 , F 2 ) . Let V j = ( F 1 , F 2 ) m ∩ I j , j = 1 , 2 , 3 . Let

B 1 = { F 1 ⁢ 𝐱 𝐢 : | 𝐢 | = m - d } ,

B 2 = { F 2 ⁢ 𝐱 𝐢 : | 𝐢 | = m - d } ,

B 1 ′ = { F 1 ⁢ TM ⁢ ( F 2 ) ⁢ 𝐱 𝐢 : | 𝐢 | = m - 2 ⁢ d } .

Then

B = ( B 1 ∖ B 1 ′ ) ∪ B 2

is a basis for ( F 1 , F 2 ) m . Moreover,

∑ s ∈ B j ord x i ⁢ s F j = ( m + n - d n + 1 )

for j = 1 , 2 , and

∑ s ∈ B 1 ′ ord x i ⁢ s F 1 = ( m + n - 2 ⁢ d n + 1 ) + ( m + n - 2 ⁢ d n ) ⁢ ord x i ⁢ TM ⁢ ( F 2 ) .

Proof.

Since F 1 and F 2 are coprime, we have I 1 ∩ I 2 = I 3 and V 1 ∩ V 2 = V 3 . For any finite-dimensional subspace W ⊂ k ⁢ [ x 0 , … , x n ] , it is easy to see that # ⁢ TM ⁢ ( W ) = dim ⁡ W . Since V = V 1 + V 2 and dim ⁡ V = dim ⁡ V 1 + dim ⁡ V 2 - dim ⁡ V 3 , we have

dim ⁡ V = # ⁢ B 1 + # ⁢ B 2 - # ⁢ B 1 ′ .

Let V 1 ′ be the span of B 1 ∖ B 1 ′ . Since TM ⁢ ( F 1 ⁢ 𝐱 𝐢 ) = TM ⁢ ( F 1 ) ⁢ 𝐱 𝐢 , distinct polynomials in B 1 have distinct trailing monomials, and it follows that

TM ⁢ ( V 1 ′ ) = TM ⁢ ( B 1 ∖ B 1 ′ ) = TM ⁢ ( B 1 ) ∖ TM ⁢ ( B 1 ′ ) .

If p ∈ V 1 ′ ∩ V 2 , p ≠ 0 , then p ∈ V 3 and TM ⁢ ( p ) ∈ TM ⁢ ( V 3 ) = TM ⁢ ( B 1 ′ ) , a contradiction. So V 1 ′ ∩ V 2 = 0 . Since dim ⁡ V 1 ′ + dim ⁡ V 2 = # ⁢ B 1 - # ⁢ B 1 ′ + # ⁢ B 2 = dim ⁡ V as well, it follows that B = ( B 1 ∖ B 1 ′ ) ∪ B 2 is a basis for V.

As is well known, the number of monomials of degree δ in x 0 , … , x n is ( n + δ n ) . Then

∑ | 𝐢 | = m - d ∑ i = 0 n ord x i ⁢ 𝐱 𝐢 = ∑ i = 0 n ∑ | 𝐢 | = m - d ord x i ⁢ 𝐱 𝐢 = ( m - d ) ⁢ ( m + n - d n ) .

By symmetry, we have

∑ | 𝐢 | = m - d ord x i ⁢ 𝐱 𝐢 = m - d n + 1 ⁢ ( m + n - d n ) = ( m + n - d n + 1 )

for i = 0 , … , n . This yields the first summation formula. The second summation formula follows similarly. ∎

3 Key theorem

In this section, we prove a fundamental result underlying the proofs of our main theorems.

Theorem 3.1.

Let F , G ∈ C ⁢ [ x 0 , … , x n ] be coprime homogeneous polynomials of the same degree d > 0 . Let I be the set of exponents i such that x i appears with a nonzero coefficient in either F or G. Let m ≥ d be a positive integer. Let g 0 , g 1 , … , g n be entire functions without common zeros such that the set { g 0 i 0 ⁢ ⋯ ⁢ g n i n : i 0 + ⋯ + i n = m } is linearly independent over C . Then for any ε > 0 , there exists a positive integer L such that the following holds:

M ⁢ N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r )

≤ exc c m , n , d ⁢ ∑ i = 0 n N g i ( L ) ⁢ ( 0 , r ) + ( m n + 1 ⁢ ( m + n n ) - c m , n , d - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r )

+ ( m + n - 2 ⁢ d n ) ⁢ N gcd ⁢ ( { 𝐠 𝐢 } 𝐢 ∈ I , r ) + ( M ′ ⁢ m ⁢ n + ε ⁢ m ) ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 ) ,

where g = ( g 0 , g 1 , … , g n ) : C → P n ,

c m , n , d = 2 ⁢ ( m + n - d n + 1 ) - ( m + n - 2 ⁢ d n + 1 ) , M = M m , n , d = 2 ⁢ ( m + n - d n ) - ( m + n - 2 ⁢ d n ) ,

and M ′ is an integer of order O ⁢ ( m n - 2 ) .

Proof of Theorem 3.1.

Let { ϕ 1 , … , ϕ M } be a basis of the ℂ -vector space ( F , G ) m , where M = dim ⁡ ( F , G ) m . For each point z ∈ ℂ , we construct a basis B z for the vector space V m = ℂ ⁢ [ x 0 , … , x n ] m / ( F , G ) m as follows. For 𝐢 = ( i 0 , … , i n ) ∈ ℕ n + 1 , we let

𝐠 ⁢ ( z ) 𝐢 = g 0 ⁢ ( z ) i 0 ⁢ ⋯ ⁢ g n ⁢ ( z ) i n .

Choose a monomial 𝐱 𝐢 1 ∈ ℂ ⁢ [ x 0 , … , x n ] m so that | 𝐠 ⁢ ( z ) 𝐢 1 | is minimal subject to the condition 𝐱 𝐢 1 ∉ ( F , G ) . Suppose now that 𝐱 𝐢 1 , … , 𝐱 𝐢 j have been constructed and are linearly independent modulo ( F , G ) m , but do not span ℂ ⁢ [ x 0 , … , x n ] m modulo ( F , G ) m . Then we let

𝐱 𝐢 j + 1 ∈ ℂ ⁢ [ x 0 , … , x n ] m

be a monomial such that | 𝐠 ⁢ ( z ) 𝐢 j + 1 | is minimal subject to the condition that 𝐱 𝐢 1 , … , 𝐱 𝐢 j + 1 are linearly independent modulo ( F , G ) m . In this way, we construct a basis of V m with monomial representatives 𝐱 𝐢 1 , … , 𝐱 𝐢 M ′ , where M ′ = dim ⁡ V m . Let I z = { 𝐢 1 , … , 𝐢 M ′ } . Then for each 𝐢 , | 𝐢 | = m , we have

𝐱 𝐢 + ∑ j = 1 M ′ c 𝐢 , j ⁢ 𝐱 𝐢 j ∈ ( F , G ) m

for some choice of coefficients c 𝐢 , j ∈ ℂ . Then for each such 𝐢 there is a linear forms L z , 𝐢 over ℂ such that

L z , 𝐢 ⁢ ( ϕ 1 , … , ϕ M ) = 𝐱 𝐢 + ∑ j = 1 M ′ c 𝐢 , j ⁢ 𝐱 𝐢 j .

We note that since there are only finitely many choices of a monomial basis of V m , there are only finitely many choices of c 𝐢 , j , even as z runs through all of ℂ . Note also that

{ L z , 𝐢 ⁢ ( ϕ 1 , … , ϕ M ) : | 𝐢 | = m , 𝐢 ∉ I z }

is a basis for ( F , G ) m . From the definition of 𝐱 𝐢 1 , … , 𝐱 𝐢 M ′ , we have the key inequality

(3.1) log | L z , 𝐢 ( Φ ( 𝐠 ( z ) ) | ≤ log | 𝐠 ( z ) 𝐢 | + C ,

where Φ ⁢ ( 𝐠 ⁢ ( z ) ) = ( ϕ 1 ⁢ ( 𝐠 ⁢ ( z ) ) , … , ϕ M ⁢ ( 𝐠 ⁢ ( z ) ) ) , and the constant C is independent of z as the choice of c 𝐢 , j is finite. The map ( ϕ 1 ⁢ ( 𝐠 ) , … , ϕ M ⁢ ( 𝐠 ) ) may not be a reduced presentation of Φ ⁢ ( 𝐠 ) . Let h be an entire function such that F ⁢ ( 𝐠 ) / h and G ⁢ ( 𝐠 ) / h are entire and have no common zeros, i.e., h is a gcd of F ⁢ ( 𝐠 ) and G ⁢ ( 𝐠 ) . Let

ψ i := ϕ i ⁢ ( 𝐠 ) h , 1 ≤ i ≤ M .

As ϕ i ∈ ( F , G ) , the function ψ i is entire for 1 ≤ i ≤ M . Since F ⁢ X i m - d , G ⁢ X i m - d ∈ ( F , G ) m , 0 ≤ i ≤ n , and g 0 , … , g n have no common zero, the functions ψ i , 1 ≤ i ≤ M , have no common zero. Hence Ψ := ( ψ 1 , … , ψ M ) is a reduced form of Φ ⁢ ( 𝐠 ) .

Applying Theorem 2.6, the second main theorem, to the map Ψ ⁢ ( 𝐠 ⁢ ( z ) ) with the collection of hyperplanes in ℙ M - 1 which are the zero loci of the linear forms L z , 𝐢 , | 𝐢 | = m , 𝐢 ∉ I z , z = r ⁢ e i ⁢ θ ∈ ℂ , we get for any ε > 0 ,

(3.2) ∫ 0 2 ⁢ π ∑ | 𝐢 | = m , 𝐢 ∉ I z ( - log | L z , 𝐢 ( Φ ( 𝐠 ( r e i ⁢ θ ) ) | + log ∥ Φ ( 𝐠 ( r e i ⁢ θ ) ) ∥ ) d ⁢ θ 2 ⁢ π + N W ⁢ ( Ψ ) ( 0 , r )

≤ exc ( M + ε ) ⁢ T Φ ⁢ ( 𝐠 ) ⁢ ( r ) .

Next, we will derive a lower bound for the left-hand side of (3.2). By the definition of the characteristic function, Lemma 2.1 and the choice of h, we have

(3.3) ∫ 0 2 ⁢ π log ∥ Φ ( 𝐠 ( r e i ⁢ θ ) ) ) ∥ d ⁢ θ 2 ⁢ π = ∫ 0 2 ⁢ π log max { | ψ 1 ( r e i ⁢ θ ) , … , ψ M ( r e i ⁢ θ ) | }

+ log ⁡ | h ⁢ ( r ⁢ e i ⁢ θ ) | ⁢ d ⁢ θ 2 ⁢ π

= T Φ ⁢ ( 𝐠 ) ⁢ ( r ) + N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r ) + O ⁢ ( 1 ) .

On the other hand, since ϕ i ∈ ℂ ⁢ [ x 0 , … , x n ] m ,

log ⁡ | ϕ i ⁢ ( 𝐠 ⁢ ( z ) ) | ≤ m ⁢ log ⁡ max ⁡ { | g 0 ⁢ ( z ) | , … , | g n ⁢ ( z ) | } + O ⁢ ( 1 ) ,

and hence

(3.4) T Φ ⁢ ( 𝐠 ) ⁢ ( r ) ≤ ∫ 0 2 ⁢ π log ⁡ ∥ Φ ⁢ ( 𝐠 ⁢ ( r ⁢ e i ⁢ θ ) ) ∥ ⁢ d ⁢ θ 2 ⁢ π + O ⁢ ( 1 ) ≤ m ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 ) .

To estimate the first term of (3.2), we use the key inequality (3.1) to derive

(3.5) - ∑ | 𝐢 | = m , 𝐢 ∉ I z log | L z , 𝐢 ( Φ ( 𝐠 ( z ) ) | + c

≥ - ∑ | 𝐢 | = m , 𝐢 ∉ I z log ⁡ | 𝐠 ⁢ ( z ) 𝐢 |

≥ - ∑ | 𝐢 | = m log ⁡ | 𝐠 ⁢ ( z ) 𝐢 | + ∑ | 𝐢 | = m , 𝐢 ∈ I z log ⁡ | 𝐠 ⁢ ( z ) 𝐢 |

≥ - ∑ | 𝐢 | = m log ⁡ | 𝐠 ⁢ ( z ) 𝐢 | + M ′ ⁢ m ⁢ log ⁡ min ⁡ { | g 0 ⁢ ( z ) | , … , | g n ⁢ ( z ) | } ,

where c is a constant independent of z. Note that

∑ | 𝐢 | = m log ⁡ | 𝐠 ⁢ ( z ) 𝐢 | = m n + 1 ⁢ ( m + n n ) ⁢ ∑ i = 0 n log ⁡ | g i ⁢ ( z ) | ,

and

log ⁡ min ⁡ { | g 0 ⁢ ( z ) | , … , | g n ⁢ ( z ) | } ≥ ∑ i = 0 n log ⁡ | g i ⁢ ( z ) | - n ⁢ max ⁡ { log ⁡ | g 0 ⁢ ( z ) | , … , log ⁡ | g n ⁢ ( z ) | } .

By Lemma 2.1, Jensen’s formula, the integration of (3.5) from 0 to 2 ⁢ π over d ⁢ θ gives

(3.6) ∫ 0 2 ⁢ π ∑ | 𝐢 | = m , 𝐢 ∉ I z - log ⁡ | L z , 𝐢 ⁢ ( Φ ⁢ ( 𝐠 ⁢ ( r ⁢ e i ⁢ θ ) ) ) | ⁢ d ⁢ θ 2 ⁢ π

≥ - ( m n + 1 ⁢ ( m + n n ) - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) - M ′ ⁢ m ⁢ n ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 ) .

By (3.3) and (3.6), we can derive from (3.2) that

M ⁢ N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r ) - ( m n + 1 ⁢ ( m + n n ) - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r )

- M ′ ⁢ m ⁢ n ⁢ T 𝐠 ⁢ ( r ) + N W ⁢ ( Ψ ) ⁢ ( 0 , r )

≤ exc ε ⁢ T Φ ⁢ ( 𝐠 ) ⁢ ( r ) + O ⁢ ( 1 ) .

Using (3.4), this yields the inequality

(3.7) M ⁢ N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r )

≤ exc ( m n + 1 ⁢ ( m + n n ) - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) - N W ⁢ ( Ψ ) ⁢ ( 0 , r )

+ ( M ′ ⁢ m ⁢ n + ε ⁢ m ) ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 ) .

By direct calculation, we find that

M ′ = ( m + n n ) - M = O ⁢ ( m n - 2 ) .

Alternatively, since F and G are coprime, the ideal ( F , G ) defines a closed subset of ℙ n of codimension at least 2, and it follows from the theory of Hilbert functions and Hilbert polynomials that M ′ = O ⁢ ( m n - 2 ) .

It is clear from (3.7) that it suffices to show that there exists a large integer L (to be determined later) such that

c m , n , d ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) - ( m + n - 2 ⁢ d n ) ⁢ N gcd ⁢ ( { 𝐠 𝐢 } 𝐢 ∈ I , r ) - N W ⁢ ( Ψ ) ⁢ ( 0 , r )

≤ c m , n , d ⁢ ∑ i = 0 n N g i ( L ) ⁢ ( 0 , r ) .

The above inequality can be deduced from the inequality

c m , n , d ⁢ ∑ i = 0 n v z + ⁢ ( g i ) - ( m + n - 2 ⁢ d n ) ⁢ min 𝐢 ∈ I ⁡ v z + ⁢ ( 𝐠 𝐢 ) - v z + ⁢ ( W ⁢ ( Ψ ) )

≤ c m , n , d ⁢ ∑ i = 0 n min ⁡ { L , v z + ⁢ ( g i ) }

for all z ∈ ℂ . The inequality holds trivially if v z + ⁢ ( g i ) ≤ L for each 0 ≤ i ≤ n . Therefore, we only need to consider the case where v z + ⁢ ( g i ) > L for some 0 ≤ i ≤ n .

For z ∈ ℂ , we define a monomial ordering > 𝐠 ⁢ ( z ) on A = ℂ ⁢ [ x 0 , … , x n ] using the weight vector 𝐮 = ( v z ⁢ ( g 0 ) , … , v z ⁢ ( g n ) ) . Let

B 1 = { F 1 ⁢ 𝐱 𝐢 : | 𝐢 | = m - d } ,

B 2 = { F 2 ⁢ 𝐱 𝐢 : | 𝐢 | = m - d } ,

B 1 ′ = { F 1 ⁢ TM 𝐠 ⁢ ( z ) ⁢ ( F 2 ) ⁢ 𝐱 𝐢 : | 𝐢 | = m - 2 ⁢ d } ,

where { F 1 , F 2 } = { F , G } and TM 𝐠 ⁢ ( z ) ⁢ ( F 2 ) ≤ TM 𝐠 ⁢ ( z ) ⁢ ( F 1 ) . By Lemma 2.11,

B = ( B 1 ∖ B 1 ′ ) ∪ B 2

is a basis for ( F , G ) m . Write B = { β 1 , … , β M } . Let η j = β j ⁢ ( 𝐠 ) / h (note that the β j depend on z). From the definition of > 𝐠 ⁢ ( z ) and F 2 , it follows that

v z + ⁢ ( TM 𝐠 ⁢ ( z ) ⁢ ( F 2 ) ) = min 𝐢 ∈ I ⁡ v z + ⁢ ( 𝐠 𝐢 ) ,

where I is the set of exponents 𝐢 such that 𝐱 𝐢 appears with a nonzero coefficient in either F or G. Then by the second part of Lemma 2.11, we have for each z ∈ ℂ ,

(3.8) ∑ j = 1 M v z + ⁢ ( η j ) ≥ ∑ i = 0 n ( ∑ s ∈ B 1 ord x i ⁢ s F 1 + ∑ s ∈ B 2 ord x i ⁢ s F 2 - ∑ s ∈ B 1 ′ ord x i ⁢ s F 1 ) ⋅ v z + ⁢ ( g i )

(3.9) ≥ c m , n , d ⁢ ∑ i = 0 n v z + ⁢ ( g i ) - ( m + n - 2 ⁢ d n ) ⁢ min 𝐢 ∈ I ⁡ v z + ⁢ ( 𝐠 𝐢 ) ,

where

c m , n , d = 2 ⁢ ( m + n - d n + 1 ) - ( m + n - 2 ⁢ d n + 1 ) .

On the other hand, from the basic properties of Wronskians, we have

(3.10) v z + ⁢ ( W ⁢ ( Ψ ) ) ≥ ∑ j = 1 M v z + ⁢ ( η j ) - 1 2 ⁢ M ⁢ ( M - 1 ) .

Combining (3.8) and (3.10), we obtain that

c m , n , d ⁢ ∑ i = 0 n v z + ⁢ ( g i ) - ( m + n - 2 ⁢ d n ) ⁢ min 𝐢 ∈ I ⁡ v z + ⁢ ( 𝐠 𝐢 ) - v z + ⁢ ( W ⁢ ( Ψ ) ) ≤ 1 2 ⁢ M ⁢ ( M - 1 ) .

Let L = 1 2 ⁢ M ⁢ ( M - 1 ) ⁢ c m , n , d - 1 . The assumption that v z + ⁢ ( g i ) > L for some 0 ≤ i ≤ n implies that

1 2 ⁢ M ⁢ ( M - 1 ) = c m , n , d ⁢ L ≤ c m , n , d ⁢ ∑ i = 0 n min ⁡ { L , v z + ⁢ ( g i ) } . ∎

4 Proof of the main theorems

We recall Borel’s lemma. (See [18, Theorem A.3.3.2].)

Lemma 4.1 (Borel’s lemma).

Let f 0 , … , f n + 1 be entire functions without zeros, satisfying

f 0 + … + f n + f n + 1 = 0 .

Define an equivalence relation i ∼ j if f i / f j is constant. Then for each equivalence class S we have

∑ i ∈ S f i = 0 .

We also recall the following result of Green [6] (See [11, Chapter VII, Theorem 4.1]).

Lemma 4.2.

Let f 0 , … , f n be entire functions satisfying

f 0 k + ⋯ + f n k = 0 .

Suppose that none of the f i are identically 0. Define an equivalence relation i ∼ j if f i / f j is constant. If k ≥ n 2 , then for each equivalence class S we have

∑ i ∈ S f i k = 0 .

We will also make use of the following result on proximity functions.

Theorem 4.3.

Let G ∈ C ⁢ [ x 1 , … , x n ] be a polynomial that does not vanish at the origin ( 0 , … , 0 ) . Suppose that g 1 , … , g n are entire functions such that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ C for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } . For all ε > 0 ,

there exists a positive integer k 0 such that for all k ≥ k 0 ,

m G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } .
if in addition each g i , 1 ≤ i ≤ n , has no zero, then

m G ⁢ ( g 1 , … , g n ) ⁢ ( 0 , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .

Proof.

By Lemma 4.2, the set

{ 𝐠 k ⁢ 𝐢 := g 1 k ⁢ i 1 ⁢ ⋯ ⁢ g n k ⁢ i n : 𝐢 = ( i 1 , … , i n ) ∈ ℕ n , | 𝐢 | ≤ d }

is linearly independent for k ≥ ( d + n n ) 2 . Since G ⁢ ( 0 , … , 0 ) ≠ 0 , G must have a nonzero constant term, and by arranging the index set in some order, we may write

G = a 𝐢 ⁢ ( 0 ) + ∑ j = 1 ℓ a 𝐢 ⁢ ( j ) ⁢ 𝐱 𝐢 ⁢ ( j ) ,

where a 𝐢 ⁢ ( j ) ≠ 0 for 0 ≤ j ≤ ℓ . Then we have

G ⁢ ( g 1 k , … , g n k ) = a 𝐢 ⁢ ( 0 ) + ∑ j = 1 ℓ a 𝐢 ⁢ ( j ) ⁢ 𝐠 k ⁢ 𝐢 ⁢ ( j ) .

We apply Theorem 2.7, Cartan’s truncated second main theorem, to the holomorphic map

𝔤 ⁢ ( k ) := ( 1 , 𝐠 k ⁢ 𝐢 ⁢ ( 1 ) , … , 𝐠 k ⁢ 𝐢 ⁢ ( ℓ ) ) : ℂ → ℙ ℓ

associated with the above expression ( k ≥ ( d + n n ) 2 ), and with the set of q = ℓ + 2 hyperplanes given by the coordinate hyperplanes of ℙ ℓ and the one defined by ∑ i = 0 ℓ a 𝐢 ⁢ ( j ) ⁢ X j . Then for any ε > 0 , we have

( 1 - ε 2 ) ⁢ T 𝔤 ⁢ ( k ) ≤ exc N G ⁢ ( g 1 k , … , g n k ) ( ℓ ) ⁢ ( 0 , r ) + ∑ j = 1 ℓ N 𝐠 k ⁢ 𝐢 ⁢ ( j ) ( ℓ ) ⁢ ( 0 , r ) .

We note that

N 𝐠 k ⁢ 𝐢 ⁢ ( j ) ( ℓ ) ⁢ ( 0 , r ) ≤ ℓ k ⁢ N 𝐠 k ⁢ 𝐢 ⁢ ( j ) ⁢ ( 0 , r ) ≤ ℓ k ⁢ T 𝐠 k ⁢ 𝐢 ⁢ ( j ) ⁢ ( r ) ≤ ℓ k ⁢ T 𝔤 ⁢ ( k ) ⁢ ( r ) ,

where the last inequality is due to the definition of characteristic functions and that

𝔤 ⁢ ( k ) = ( 1 , 𝐠 k ⁢ 𝐢 ⁢ ( 1 ) , … , 𝐠 k ⁢ 𝐢 ⁢ ( ℓ ) ) .

Then for k > max ⁡ { 2 ⁢ ℓ 2 ε , ( d + n n ) 2 } ,

(4.1) ( 1 - ε ) ⁢ T 𝔤 ⁢ ( k ) ⁢ ( r ) ≤ exc N G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r )

= T G ⁢ ( g 1 k , … , g n k ) ⁢ ( r ) - m G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) + O ⁢ ( 1 ) .

Since G ⁢ ( g 1 k , … , g n k ) is an entire function, we have

T G ⁢ ( g 1 k , … , g n k ) ⁢ ( r ) = m G ⁢ ( g 1 k , … , g n k ) ⁢ ( ∞ , r ) ≤ T 𝔤 ⁢ ( k ) ⁢ ( r ) + O ⁢ ( 1 ) .

Consequently,

m G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) ≤ exc ε ⁢ T 𝔤 ⁢ ( k ) ⁢ ( r ) ≤ exc d ⁢ ε ⁢ T ( 1 , g 1 k , … , g n k ) ⁢ ( r ) ≤ d ⁢ n ⁢ ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } .

When the g i are entire functions without zeros, we may assume that the set

{ 𝐠 𝐢 := g 1 i 1 ⁢ ⋯ ⁢ g n i n : 𝐢 = ( i 1 , … , i n ) ∈ ℕ n , | 𝐢 | ≤ d }

is linearly independent by Lemma 4.1. Then we can repeat the previous argument for k = 1 with the additional condition N g i ⁢ ( 0 , r ) = O ⁢ ( 1 ) , 1 ≤ i ≤ n , to conclude the proof. ∎

Proof of Theorem 1.2 and Theorem 1.3.

We first claim that the functions g 1 , … , g n are algebraically independent over ℂ under the assumption that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ ℂ for any index set ( i 1 , … , i n ) ∈ ℤ n ∖ { ( 0 , … , 0 ) } . If not, then they satisfy a nontrivial ℂ -linear relation, say

(4.2) ∑ 𝐢 a 𝐢 ⁢ g 1 i 1 ⁢ ⋯ ⁢ g n i n = 0 ,

where the sum is over finitely many index sets 𝐢 = ( i 1 , … , i n ) and a 𝐢 ∈ ℂ * . We may further assume that no proper sub-sum of the left-hand side of (4.2) is zero. Since g 1 , … , g n are entire functions without zeros, Borel’s lemma implies that we have a pair of indices 𝐢 = ( i 1 , … , i n ) , 𝐣 = ( j 1 , … , j n ) with 𝐢 ≠ 𝐣 such that a 𝐢 ⁢ g 1 i 1 ⁢ ⋯ ⁢ g n i n is a constant multiple of some a 𝐣 ⁢ g 1 j 1 ⁢ ⋯ ⁢ g n j n appearing on the left-hand side of (4.2), and hence g 1 i 1 - j 1 ⁢ ⋯ ⁢ g n i n - j n ∈ ℂ * , contradicting our assumption.

We prove part a of both theorems first. Consider ( ℂ * ) n ⊂ ℙ n , where we identify the vector ( x 1 , … , x n ) ∈ ( ℂ * ) n with [ 1 : x 1 : ⋯ : x n ] , and let Y ¯ be the Zariski closure of Y in ℙ n . Then N 𝐠 ⁢ ( Y , r ) = N 𝐠 ⁢ ( Y ¯ , r ) , and we may assume that Y is a closed subscheme of ℙ n (of codimension at least 2).

Let I ⊂ ℂ ⁢ [ x 0 , … , x n ] be a homogeneous ideal associated to Y. We can find homogeneous polynomials F ~ , G ~ ∈ I of the same degree d such that F ~ and G ~ are coprime. Let Y ′ be the closed subscheme defined by the ideal ( F ~ , G ~ ) . Then

N 𝐠 ⁢ ( Y , r ) ≤ N 𝐠 ⁢ ( Y ′ , r )

for all r > 0 . Therefore, we may assume that Y is defined by the ideal ( F ~ , G ~ ) after replacing Y by Y ′ . Then for any ε > 0 , Theorem 3.1 for any (large) m implies that

(4.3) M ⁢ N 𝐠 ⁢ ( Y , r ) = M ⁢ N gcd ⁢ ( F ~ ⁢ ( 𝐠 ) , G ~ ⁢ ( 𝐠 ) , r ) ≤ exc ( M ′ ⁢ m ⁢ n + ε ⁢ m ) ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 )

as the g i are entire functions with no zeros. Let ε ′ > 0 . Since we have M ′ = O ⁢ ( m n - 2 ) and M = m n n ! + O ⁢ ( m n - 1 ) , choosing m large enough, depending only on ε ′ , (4.3) implies that

(4.4) N gcd ⁢ ( F ~ ⁢ ( 𝐠 ) , G ~ ⁢ ( 𝐠 ) , r ) ≤ exc ε ′ ⁢ T 𝐠 ⁢ ( r ) .

This completes the proof of Theorem 1.2 a.

To show Theorem 1.3 a, we note that we may assume deg ⁡ F = deg ⁡ G = d since

N gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ N gcd ⁢ ( F e ⁢ ( g 1 , … , g n ) , G h ⁢ ( g 1 , … , g n ) , r ) ,

where e = deg ⁡ G and h = deg ⁡ F . Then Theorem 1.3 a follows from (4.4) by taking

F ~ ⁢ ( x 0 , … , x n ) = x 0 d ⁢ F ⁢ ( x 1 x 0 , … , x n x 0 ) ,

G ~ ⁢ ( x 0 , … , x n ) = x 0 d ⁢ G ⁢ ( x 1 x 0 , … , x n x 0 ) ,

and using Lemma 2.5.

Suppose in addition that, say, G does not vanish at the origin ( 0 , … , 0 ) . It is immediate from the definitions that

m gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ m G ⁢ ( g 1 , … , g n ) ⁢ ( 0 , r ) ,

and so Theorem 4.3 implies that

m gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .

Combined with part a above and Lemma 2.10, we obtain Theorem 1.3 b.

It remains to prove Theorem 1.2 b. Let X be a nonsingular projective toric compactification of ( ℂ * ) n . Let Y ¯ be the Zariski closure of Y in X, and suppose that Y ¯ is in general position with the boundary of ( ℂ * ) n in X. Since N 𝐠 ⁢ ( Y , r ) = N 𝐠 ⁢ ( Y ¯ , r ) , by part a it suffices to show that m 𝐠 ⁢ ( Y ¯ , r ) ≤ exc ε ⁢ T 𝐠 ⁢ ( r ) . In fact, we will show the stronger statement with Y ¯ replaced by an effective divisor D on X in general position with the boundary of ( ℂ * ) n in X. We follow the proof of [12, Theorem 4.4], which shows that there exist embeddings ϕ i : ( ℂ * ) n → 𝔸 n (which extend an automorphism of ( ℂ * ) n ) and polynomials p i ∈ ℂ ⁢ [ x 1 , … , x n ] nonvanishing at the origin, i = 1 , … , t , such that for every P ∈ ( ℂ * ) n ⊂ X ⁢ ( ℂ ) , there exists i ∈ { 1 , … , t } satisfying

λ D ⁢ ( P ) = - log ⁡ | p i ⁢ ( ϕ i ⁢ ( P ) ) | + O ⁢ ( 1 ) .

Thus,

λ D ⁢ ( P ) ≤ - ∑ i = 1 t log - ⁡ | p i ⁢ ( ϕ i ⁢ ( P ) ) | + O ⁢ ( 1 )

for all P ∈ ( ℂ * ) n ⊂ X ⁢ ( ℂ ) . By Theorem 4.3, this implies that for any ε > 0 ,

m 𝐠 ⁢ ( D , r ) ≤ ∑ i = 1 t m p i ⁢ ( ϕ i ⁢ ( g 1 , … , g n ) ) ⁢ ( 0 , r ) + O ⁢ ( 1 )

≤ exc ε ⁢ T 𝐠 ⁢ ( r ) + O ⁢ ( 1 ) ,

completing the proof. ∎

Proof of Theorem 1.5.

By replacing F and G by suitable linear combinations of F and G, we may assume that F and G have the same degree d, and that neither F nor G vanishes at the origin. We now consider the two related homogeneous polynomials

F 1 ⁢ ( x 0 , x 1 , … , x n ) = x 0 d ⁢ F ⁢ ( x 1 x 0 , … , x n x 0 ) ,

G 1 ⁢ ( x 0 , x 1 , … , x n ) = x 0 d ⁢ G ⁢ ( x 1 x 0 , … , x n x 0 ) .

Since F and G are coprime, it follows easily that F 1 and G 1 are coprime. Let g 1 , … , g n be meromorphic functions. Then there exists an entire function h 0 such that h 0 and h i := g i ⋅ h 0 , 1 ≤ i ≤ n , are entire functions without a common zero. Therefore, ( h 0 , h 1 , … , h n ) is a reduced form of the holomorphic map 𝐠 := [ 1 : g 1 : ⋯ : g n ] : ℂ → ℙ n . Assume that g 1 j 1 ⁢ … ⁢ g n j n is not constant for any ( j 1 , … , j n ) ≠ ( 0 , … , 0 ) ∈ ℤ n . Let

(4.5) k ≥ ( m + n n ) 2 .

Then { ( h 0 k ) i 0 ⁢ ⋯ ⁢ ( h n k ) i n : i 0 + ⋯ + i n = m } is linearly independent over ℂ by Lemma 4.2. Let 𝐡 k = ( h 0 k , … , h n k ) : ℂ → ℙ n . Let I be the set of exponents 𝐢 such that 𝐱 𝐢 appears with a nonzero coefficient in either F 1 or G 1 . Note that

F 1 ⁢ ( 𝐡 k ) = h 0 k ⁢ d ⁢ F ⁢ ( g 1 k , … , g n k ) ,

G 1 ⁢ ( 𝐡 k ) = h 0 k ⁢ d ⁢ G ⁢ ( g 1 k , … , g n k ) ,

and so for any z ∈ ℂ ,

v z ⁢ ( F 1 ⁢ ( 𝐡 k ) ) = v z ⁢ ( h 0 k ⁢ d ) + v z ⁢ ( F ⁢ ( g 1 k , … , g n k ) ) ≥ min 𝐢 ∈ I ⁡ v z ⁢ ( 𝐡 k ⁢ 𝐢 ) + v z ⁢ ( F ⁢ ( g 1 k , … , g n k ) ) ,

v z ⁢ ( G 1 ⁢ ( 𝐡 k ) ) = v z ⁢ ( h 0 k ⁢ d ) + v z ⁢ ( G ⁢ ( g 1 k , … , g n k ) ) ≥ min 𝐢 ∈ I ⁡ v z ⁢ ( 𝐡 k ⁢ 𝐢 ) + v z ⁢ ( G ⁢ ( g 1 k , … , g n k ) ) ,

since x 0 d is a monomial appearing nontrivially in F 1 and G 1 as F and G do not vanish at the origin. Then

v z ⁢ ( F 1 ⁢ ( 𝐡 k ) ) - min 𝐢 ∈ I ⁡ v z ⁢ ( 𝐡 k ⁢ 𝐢 ) ≥ v z + ⁢ ( F ⁢ ( g 1 k , … , g n k ) ) ,

v z ⁢ ( G 1 ⁢ ( 𝐡 k ) ) - min 𝐢 ∈ I ⁡ v z ⁢ ( 𝐡 k ⁢ 𝐢 ) ≥ v z + ⁢ ( G ⁢ ( g 1 k , … , g n k ) ) ,

and it follows that

N gcd ⁢ ( F 1 ⁢ ( 𝐡 k ) , G 1 ⁢ ( 𝐡 k ) , r ) - N gcd ⁢ ( { 𝐡 k ⁢ 𝐢 } 𝐢 ∈ I , r ) ≥ N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) .

Then by Theorem 3.1, for any ε > 0 there exists a positive integer L such that for all positive integers k satisfying (4.5), we have the following:

M ⁢ N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r )

≤ M ⁢ ( N gcd ⁢ ( F 1 ⁢ ( 𝐡 k ) , G 1 ⁢ ( 𝐡 k ) , r ) - N gcd ⁢ ( { 𝐡 k ⁢ 𝐢 } 𝐢 ∈ I , r ) )

≤ exc c m , n , d ⁢ ∑ i = 0 n N h i k ( L ) ⁢ ( 0 , r ) + ( m n + 1 ⁢ ( m + n n ) - c m , n , d - M ′ ⁢ m ) ⁢ ∑ i = 0 n N h i k ⁢ ( 0 , r )

+ ( ( m + n - 2 ⁢ d n ) - M ) ⁢ N gcd ⁢ ( { 𝐡 k ⁢ 𝐢 } 𝐢 ∈ I , r ) + ( M ′ ⁢ m ⁢ n + ε ⁢ m ) ⁢ T 𝐡 k ⁢ ( r ) + O ⁢ ( 1 ) ,

where

c m , n , d = 2 ⁢ ( m + n - d n + 1 ) - ( m + n - 2 ⁢ d n + 1 ) , M = 2 ⁢ ( m + n - d n ) - ( m + n - 2 ⁢ d n ) ,

and M ′ is an integer of order O ⁢ ( m n - 2 ) . Elementary computations give that

( m + n n ) = m n n ! + ( n + 1 ) ⁢ m n - 1 2 ⁢ ( n - 1 ) ! + O ⁢ ( m n - 2 ) ,

c m , n , d = m n + 1 ( n + 1 ) ! + m n 2 ⁢ ( n - 1 ) ! + O ⁢ ( m n - 1 ) ,

M = m n n ! + O ⁢ ( m n - 1 ) .

Then

m n + 1 ⁢ ( m + n n ) - c m , n , d = O ⁢ ( m n - 1 ) ,

( m + n - 2 ⁢ d n ) - M = O ⁢ ( m n - 1 ) .

Furthermore,

∑ i = 0 n N h i k ( L ) ⁢ ( 0 , r ) ≤ L k ⁢ ∑ i = 0 n N h i k ⁢ ( 0 , r ) ,

and

∑ i = 0 n N h i k ⁢ ( 0 , r ) = ∑ i = 0 n N 𝐡 k ⁢ ( H i , r ) ≤ ( n + 1 ) ⁢ T 𝐡 k ⁢ ( r ) = ( n + 1 ) ⁢ T 𝐠 k ⁢ ( r ) ,

where H i is the coordinate hyperplane defined by x i = 0 . Then we find that after choosing m sufficiently large, for all sufficiently large k (depending on m),

(4.6) N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ T 𝐠 k ⁢ ( r ) .

This concludes the proof of Theorem 1.5 a. If in addition g 1 , … , g n are entire functions and, say, G does not vanish at the origin, then

m gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ m G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) ,

and so Theorem 4.3 implies that

m gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) }

for k large enough. Together with (4.6), we reach the conclusion of b. ∎

Proof of Corollary 1.6.

If f i ⁢ g j ∉ ℂ for any ( i , j ) ≠ ( 0 , 0 ) ∈ ℤ 2 , then the assertion follows from Theorem 1.5 for the meromorphic functions f and g with the polynomials F = x 1 - 1 and G = x 2 - 1 . Suppose that f i ⁢ g j = c ∈ ℂ for some ( i , j ) ∈ ℤ 2 ∖ { ( 0 , 0 ) } . If f k - 1 and g k - 1 have no common zero for all k sufficiently large, then the gcd inequality (1.1) holds trivially. Otherwise, we may find z 0 ∈ ℂ such that f k ⁢ ( z 0 ) = g k ⁢ ( z 0 ) = 1 for some k. This implies that c k = 1 and hence f i ⁢ k ⁢ g j ⁢ k = 1 , contradicting the assumption that f and g are multiplicatively independent. ∎

The proof of Corollary 1.4 is similar (with k = 1 ) to the proof of Corollary 1.6, and so we omit it.

Proof of Corollary 1.8.

We prove part a. The proof of part b is similar.

Suppose that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ ℂ for any index set ( i 1 , … , i n ) ∈ ℤ n + 1 ∖ { ( 0 , … , 0 ) } and that F ⁢ ( g 1 k , … , g n k ) / G ⁢ ( g 1 k , … , g n k ) is an entire function. Then

N G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) = N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ,

and by Theorem 1.5 a, for any ε > 0 , there exists a positive integer k 0 such that

(4.7) N G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) }

if k ≥ k 0 . On the other hand, using the same notation as in the proof of Theorem 4.3 and recalling the first part of (4.1),

( 1 - ε ) ⁢ T 𝔤 ⁢ ( k ) ⁢ ( r ) ≤ exc N G ⁢ ( g 1 k , … , g n k ) ⁢ ( 0 , r )

for k > k 1 := max ⁡ { 2 ⁢ ℓ 2 ε , ( d + n n ) 2 } . Together with (4.7), we get

T 𝔤 ⁢ ( k ) ⁢ ( r ) ≤ exc 2 ⁢ ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } + O ⁢ ( 1 )

for k ≥ max ⁡ { k 0 , k 1 } . Hence,

T 𝔤 G ⁢ ( r ) ≤ exc 2 ⁢ ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } + O ⁢ ( 1 ) ,

contradicting the assumption that T 𝔤 G ⁢ ( r ) ≍ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } . ∎

5 GCD with moving targets

5.1 Statement of the main results

Let 𝐠 be a holomorphic map from ℂ to ℙ n . Let K 𝐠 be the set containing all meromorphic functions a such that T a ⁢ ( r ) = o ⁢ ( T 𝐠 ⁢ ( r ) ) . By the basic properties of characteristic functions, K 𝐠 is a field.

Theorem 5.1.

Let g 1 , … , g n be entire functions without zeros and g = ( 1 , g 1 , … , g n ) . Let F , G ∈ K g ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials. Assume that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ K g for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } . Let ε > 0 .

Then

N gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .
If F and G are not both identically zero at the origin ( 0 , … , 0 ) and the coefficients of F and G are entire functions in K 𝐠 , then

T gcd ⁢ ( F ⁢ ( g 1 , … , g n ) , G ⁢ ( g 1 , … , g n ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } .

Theorem 5.2.

Let g 1 , … , g n be meromorphic functions and let g = [ 1 : g 1 : … : g n ] be a holomorphic map from C to P n . Let F , G ∈ K g ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials such that not both of them are identically zero at ( 0 , … , 0 ) . If g 1 i 1 ⁢ ⋯ ⁢ g n i n ∉ K g for any index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } , then for any ε > 0 , there exists k 0 such that for k ≥ k 0

one has

N gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) } ,
one has

T gcd ⁢ ( F ⁢ ( g 1 k , … , g n k ) , G ⁢ ( g 1 k , … , g n k ) , r ) ≤ exc ε ⁢ max 1 ≤ i ≤ n ⁡ { T g i k ⁢ ( r ) }

if g 1 , … , g n are entire functions and the coefficients of F and G are entire functions in K 𝐠 .

Let g 1 , … , g n be entire functions and 𝐠 = ( 1 , g 1 , … , g n ) . Let G ∈ K 𝐠 ⁢ [ x 1 , … , x n ] be a nonconstant polynomial such that G ⁢ ( 0 , … , 0 ) is not identically zero. Since G has a nonzero constant term, after arranging the index set in some order, we may write

G = a 𝐢 ⁢ ( 0 ) + ∑ j = 1 ℓ a 𝐢 ⁢ ( j ) ⁢ 𝐱 𝐢 ⁢ ( j ) ∈ K 𝐠 ⁢ [ x 1 , … , x n ] ,

where a 𝐢 ⁢ ( j ) ≠ 0 for 0 ≤ j ≤ ℓ . We then let 𝔤 G := ( 1 , 𝐠 𝐢 ⁢ ( 1 ) , … , 𝐠 𝐢 ⁢ ( ℓ ) ) : ℂ → ℙ ℓ .

Corollary 5.3.

Let g 1 , … , g n be entire functions and g = ( 1 , g 1 , … , g n ) . Assume that T g G ⁢ ( r ) ≍ max 1 ≤ i ≤ n ⁡ { T g i ⁢ ( r ) } . Let F , G ∈ K g ⁢ [ x 1 , … , x n ] be nonconstant coprime polynomials with coefficients that are entire functions. Assume that G ⁢ ( 0 , … , 0 ) is not identically zero.

If F ⁢ ( g 1 k , … , g n k ) / G ⁢ ( g 1 k , … , g n k ) are entire functions for infinitely many positive integers k , or
g 1 , … , g n are entire functions without zeros and F ⁢ ( g 1 , … , g n ) / G ⁢ ( g 1 , … , g n ) is an entire function,

then there exists an index set ( i 1 , … , i n ) ∈ Z n ∖ { ( 0 , … , 0 ) } such that g 1 i 1 ⁢ ⋯ ⁢ g n i n ∈ K g .

5.2 Nevanlinna theory with moving targets

Let 𝐟 = ( f 0 , … , f n ) be a holomorphic map from ℂ to ℙ n , where f 0 , f 1 , … , f n are holomorphic functions without a common zero. Let a 0 , … , a n ∈ K 𝐟 , and let L := a 0 ⁢ X 0 + … + a n ⁢ X n . Then L defines a hyperplane H in ℙ n ⁢ ( K 𝐟 ) . We note that H ⁢ ( z ) is the hyperplane determined by the linear form

L ⁢ ( z ) = a 0 ⁢ ( z ) ⁢ X 0 + … + a n ⁢ ( z ) ⁢ X n

for z ∈ ℂ that is not a common zero of a 0 , … , a n , or a pole of any a k , 0 ≤ k ≤ n . The definition of the Weil function, proximity function and counting function can be easily extended to moving hyperplanes. For example,

λ H ⁢ ( z ) ⁢ ( P ) = - log ⁡ | ( h ⁢ a 0 ) ⁢ ( z ) ⁢ x 0 + ⋯ + ( h ⁢ a n ) ⁢ ( z ) ⁢ x n | max ⁡ { | x 0 | , … , | x n | } ⁢ max ⁡ { | ( h ⁢ a 0 ) ⁢ ( z ) | , … , | ( h ⁢ a n ) ⁢ ( z ) | } ,

where h is a meromorphic function such that h ⁢ a 0 , … , h ⁢ a n are entire functions without common zeros, P = ( x 0 , … , x n ) ∈ ℙ n ⁢ ( ℂ ) and z ∈ ℂ . It is clear that

λ H ⁢ ( z ) ⁢ ( P ) = - log ⁡ | a 0 ⁢ ( z ) ⁢ x 0 + ⋯ + a n ⁢ ( z ) ⁢ x n | max ⁡ { | x 0 | , … , | x n | } ⁢ max ⁡ { | a 0 ⁢ ( z ) | , … , | a n ⁢ ( z ) | }

for z ∈ ℂ which is not a common zero of a 0 , … , a n , or a pole of any a k , 0 ≤ k ≤ n . The first main theorem for a moving hyperplane H can be stated as

T 𝐟 ⁢ ( r ) = N 𝐟 ⁢ ( H , r ) + m 𝐟 ⁢ ( H , r ) + o ⁢ ( T 𝐟 ⁢ ( r ) ) .

We will reformulate the second main theorem with moving targets stated in [18, Theorem A4.2.1] to suit our purpose. Let a j ⁢ 0 , … , a j ⁢ n ∈ K 𝐟 , and let L j := a j ⁢ 0 ⁢ X 0 + … + a j ⁢ n ⁢ X n . Without loss of generality, we will normalize the linear forms L j , 1 ≤ j ≤ q , such that for each 1 ≤ j ≤ q , there exists 0 ≤ j ′ ≤ n such that a j ⁢ j ′ = 1 . Let t be a positive integer and let V ⁢ ( t ) be the complex vector space spanned by the elements

{ ∏ a j ⁢ k n j ⁢ k : n j ⁢ k ≥ 0 , ∑ n j ⁢ k = t } ,

where the product and sum runs over 1 ≤ j ≤ q and 0 ≤ k ≤ n . Let 1 = b 1 , … , b u be a basis of V ⁢ ( t ) and b 1 , … , b w a basis of V ⁢ ( t + 1 ) . It is clear that u ≤ w . Moreover, we have (see [25, Lemma 6])

lim inf t → ∞ ⁡ dim ⁡ V ⁢ ( t + 1 ) dim ⁡ V ⁢ ( t ) = 1 .

The following formulation of the second main theorem with moving targets follows from the proof of [18, Theorem A4.2.1] by adding the Wronskian term when applying the second main theorem.

Theorem 5.4.

Suppose that f = ( f 0 , … , f n ) : C → P n ⁢ ( C ) is a holomorphic curve, where f 0 , f 1 , … , f n are entire functions without common zero. Let H j , 1 ≤ j ≤ q , be arbitrary (moving) hyperplanes given by L j := a j ⁢ 0 ⁢ X 0 + … + a j ⁢ n ⁢ X n , where a j ⁢ 0 , … , a j ⁢ n ∈ K f . Denote by W the Wronskian of { h ⁢ b m ⁢ f k : 1 ≤ m ≤ w , 0 ≤ k ≤ n } , where h is a meromorphic function such that h ⁢ b 1 , … , h ⁢ b w are entire functions without common zero. If f is linearly non-degenerate over K f , then for any ε > 0 , we have the following inequality:

∫ 0 2 ⁢ π max J ⁢ ∑ k ∈ J λ H k ⁢ ( r ⁢ e i ⁢ θ ) ⁢ ( 𝐟 ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ d ⁢ θ 2 ⁢ π + 1 u ⁢ N W ⁢ ( 0 , r ) ≤ exc ( w u ⁢ ( n + 1 ) + ε ) ⁢ T 𝐟 ⁢ ( r ) + o ⁢ ( T 𝐟 ⁢ ( r ) ) ,

where the maximum is taken over all subsets J of { 1 , … , q } such that H j ⁢ ( r ⁢ e i ⁢ θ ) , j ∈ J , are in general position.

The following two lemmas are moving targets versions of Borel’s lemma and Green’s theorem. The proofs can be found in [9].

Lemma 5.5.

Let f 0 , … , f n be entire functions with no zeros and f := ( f 0 , … , f n ) a holomorphic map from C to P n ⁢ ( C ) . Suppose that f 0 , … , f n are linearly dependent over K f . Then for each f i , there exists j ≠ i such that f i / f j ∈ K f .

Lemma 5.6.

Let f 0 , … , f n be nonzero entire functions without a common zero and let f = ( f 0 , … , f n ) be a holomorphic map from C to P n . Assume that for an integer k ≥ n 2 the following holds:

a 0 ⁢ f 0 k + … + a n ⁢ f n k = 0 ,

where a i ≠ 0 ∈ K f , 0 ≤ i ≤ n . Then for each f i , there exists j ≠ i such that ( f i / f j ) k ∈ K f .

5.3 Key theorem

In this subsection we state and prove Theorem 5.7, the analogue of Theorem 3.1 in the moving targets setting. After establishing Theorem 5.7, the proofs of Theorem 5.1, Theorem 5.2 and Corollary 5.3 are developed in a manner similar to their constant analogues, and we therefore omit the proofs.

The key result is the following analogue of Theorem 3.1.

Theorem 5.7.

Let g 0 , g 1 , … , g n be entire functions without common zeros and let g = ( g 0 , g 1 , … , g n ) . Let F , G ∈ K g ⁢ [ x 0 , x 1 , … , x n ] be coprime homogeneous polynomials of the same degree d > 0 . Let I be the set of exponents i such that x i appears with a nonzero coefficient in either F or G, and let m ≥ d be a positive integer. Moreover, suppose that the set { g 0 i 0 ⁢ … ⁢ g n i n : i 0 + ⋯ + i n = m } is linearly independent over K g . Then for any ε > 0 , there exists a positive integer L such that the following holds:

M ⁢ N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r )

≤ exc c m , n , d ⁢ ∑ i = 1 n N g i ( L ) ⁢ ( 0 , r ) + ( m n + 1 ⁢ ( m + n n ) - c m , n , d - M ′ ⁢ m ) ⁢ ∑ i = 1 n N g i ⁢ ( 0 , r )

+ ( m + n - 2 ⁢ d n ) ⁢ N gcd ⁢ ( { 𝐠 𝐢 } 𝐢 ∈ I , r ) + ( M ′ ⁢ m ⁢ n + ε ⁢ m + M ⁢ ε 2 ) ⁢ T 𝐠 ⁢ ( r ) + o ⁢ ( T 𝐠 ⁢ ( r ) ) ,

where

c m , n , d = 2 ⁢ ( m + n - d n + 1 ) - ( m + n - 2 ⁢ d n + 1 ) , M = 2 ⁢ ( m + n - d n ) - ( m + n - 2 ⁢ d n ) ,

and M ′ is an integer of order O ⁢ ( m n - 2 ) .

The basic ideas used to prove the theorem are similar to the ideas used in the proof of Theorem 3.1. We will make explicit the important differences in the moving target case and omit whatever is identical or obvious from the proof of Theorem 3.1.

Proof of Theorem 5.7.

Let ( F , G ) be the ideal generated by F and G in K 𝐠 ⁢ [ x 0 , … , x n ] and ( F , G ) m := K 𝐠 ⁢ [ x 0 , … , x n ] m ∩ ( F , G ) . We choose { ϕ 1 , … , ϕ M } to be a basis of the K 𝐠 -vector space ( F , G ) m consisting of elements of the form F ⁢ 𝐱 𝐢 and G ⁢ 𝐱 𝐣 .

For each z ∈ ℂ , we construct a basis B z for V m = K 𝐠 ⁢ [ x 0 , … , x n ] m / ( F , G ) m as follows. For 𝐢 = ( i 0 , … , i n ) ∈ ℕ n + 1 , we let 𝐠 ⁢ ( z ) 𝐢 = g 0 ⁢ ( z ) i 0 ⁢ ⋯ ⁢ g n ⁢ ( z ) i n . Choose a monomial 𝐱 𝐢 1 so that | 𝐠 ⁢ ( z ) 𝐢 1 | is minimal subject to the condition 𝐱 𝐢 1 ∉ ( F , G ) . Suppose now that 𝐱 𝐢 1 , … , 𝐱 𝐢 j have been constructed. Then we let

𝐱 𝐢 j + 1 ∈ K 𝐠 ⁢ [ x 0 , … , x n ] m

𝐱 𝐢 + ∑ j = 1 M ′ b 𝐢 , j , z ⁢ 𝐱 𝐢 j ∈ ( F , G ) m

for some choice of coefficient b 𝐢 , j , z ∈ K 𝐠 . Then for each such 𝐢 there is a linear form L z , 𝐢 over K 𝐠 such that

(5.1) L z , 𝐢 ⁢ ( ϕ 1 , … , ϕ M ) = 𝐱 𝐢 + ∑ j = 1 M ′ b 𝐢 , j , z ⁢ 𝐱 𝐢 j .

We note that there are only finitely many choice of b 𝐢 , j , z even as z runs through all of ℂ , and the set { L z , 𝐢 ⁢ ( ϕ 1 , … , ϕ M ) : | 𝐢 | = m , 𝐢 ∉ I z } is a basis for ( F , G ) m . From the definition of 𝐱 𝐢 1 , … , 𝐱 𝐢 M ′ , we have the key inequality

(5.2) log | L z , 𝐢 ( Φ ( 𝐠 ( z ) ) | ≤ log | 𝐠 ( z ) 𝐢 | + log ∥ L z , 𝐢 ( z ) ∥ + O ( 1 ) ,

where Φ ⁢ ( 𝐠 ⁢ ( z ) ) = ( ϕ 1 ⁢ ( 𝐠 ⁢ ( z ) ) , … , ϕ M ⁢ ( 𝐠 ⁢ ( z ) ) ) ,

∥ L z , 𝐢 ⁢ ( z ) ∥ = max 0 ≤ j ≤ M ′ ⁡ | b 𝐢 , j , z ⁢ ( z ) | ,

and b 𝐢 , 0 , z = 1 . The map ( ϕ 1 ⁢ ( 𝐠 ) , … , ϕ M ⁢ ( 𝐠 ) ) may not be a reduced presentation of Φ ⁢ ( 𝐠 ) . Let h be a meromorphic function such that F ⁢ ( 𝐠 ) / h and G ⁢ ( 𝐠 ) / h are entire and have no common zeros, i.e.,

(5.3) N h ⁢ ( 0 , r ) = N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r ) .

Moreover, since g i , 0 ≤ i ≤ n , are entire functions, the poles of h comes from the poles of the coefficients of F and G, and hence

(5.4) N h ⁢ ( ∞ , r ) ≤ o ⁢ ( T 𝐠 ⁢ ( r ) ) .

Let ψ i := ϕ i ⁢ ( 𝐠 ) / h , 1 ≤ i ≤ M . By the choice of ϕ i , the function ψ i is entire for 1 ≤ i ≤ M . Furthermore, since F ⁢ X i m - d , G ⁢ X i m - d ∈ ( F , G ) m , 0 ≤ i ≤ n , and g 0 , … , g n have no common zero, the functions ψ i , 1 ≤ i ≤ M , have no common zero. Hence Ψ := ( ψ 1 , … , ψ M ) is a reduced form of Φ ⁢ ( 𝐠 ) .

Let V be the complex vector space spanned by 1, the coefficients of F and G, and all possible (finitely many) choices of b 𝐢 , j , z in (5.1). Let t be a large positive integer and let V ⁢ ( t ) be the finite-dimensional vector space spanned by products of t elements in V.

Let b 1 = 1 , b 2 , … , b u be a basis of V ⁢ ( t ) and b 1 , … , b w a basis of V ⁢ ( t + 1 ) . We will choose t sufficiently large so that

w u ≤ 1 + ε 2 ⁢ m .

Denote by W the Wronskian of { a ⁢ b m ⁢ ψ k : 1 ≤ m ≤ w , 1 ≤ k ≤ M } , where a is an entire function such that a = a ⁢ b 1 , a ⁢ b 2 , … , a ⁢ b w are entire and have no common zeros. Applying Theorem 5.4 to the holomorphic map Φ ⁢ ( 𝐠 ) with the reduced form Ψ = ( ψ 1 , … , ψ M ) and the set of linear forms { L z , 𝐢 : | 𝐢 | = m , 𝐢 ∉ I z } , we have the following inequality:

(5.5) ∫ 0 2 ⁢ π ∑ | 𝐢 | = m , 𝐢 ∉ I z λ L z , 𝐢 ⁢ ( Ψ ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ ( Ψ ⁢ ( r ⁢ e i ⁢ θ ) ) ⁢ d ⁢ θ 2 ⁢ π + 1 u ⁢ N W ⁢ ( 0 , r )

≤ exc ( w u ⁢ M + ε ) ⁢ T Ψ ⁢ ( r ) + o ⁢ ( T Ψ ⁢ ( r ) ) ,

where

λ L z , 𝐢 ⁢ ( Ψ ⁢ ( r ⁢ e i ⁢ θ ) ) ( Ψ ( r e i ⁢ θ ) ) = - log | L z , 𝐢 ( Φ ( 𝐠 ( z ) ) | + log ∥ Φ ( 𝐠 ( z ) ) ∥ + log ∥ L z , 𝐢 ( z ) ∥

if z = r ⁢ e i ⁢ θ is not a pole of any coefficient of L z , 𝐢 . We note that such z is in a discrete subset of ℂ , and hence its radius r ∈ ( 0 , ∞ ) is in a set of finite Lebesgue measure. In the following computation, we will only consider z which is not a pole of any b 𝐢 , j , z from (5.1). Next, we will derive a lower bound for the first term of the left-hand side of (5.5). By the definition of the characteristic function, Lemma 2.1, (5.3) and (5.4), we have

∫ 0 2 ⁢ π log ∥ Φ ( 𝐠 ( z ) ) ) ∥ d ⁢ θ 2 ⁢ π = ∫ 0 2 ⁢ π log max { | ψ 1 ( z ) , … , ψ M ( z ) | } + log | h ( z ) | d ⁢ θ 2 ⁢ π

= T Ψ ⁢ ( r ) + N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r ) + o ⁢ ( T 𝐠 ⁢ ( r ) ) .

On the other hand, since ϕ i = ∑ I α I ⁢ x I ∈ K 𝐠 ⁢ [ x 0 , … , x n ] m ,

log ⁡ | ϕ i ⁢ ( 𝐠 ⁢ ( z ) ) | ≤ m ⁢ log ⁡ max ⁡ { | g 0 ⁢ ( z ) | , … , | g n ⁢ ( z ) | } + ∑ I log + ⁡ | α I ⁢ ( z ) | + O ⁢ ( 1 ) ,

and hence

(5.6) ∫ 0 2 ⁢ π log ⁡ ∥ Φ ⁢ ( 𝐠 ⁢ ( z ) ) ∥ ⁢ d ⁢ θ 2 ⁢ π ≤ m ⁢ T 𝐠 ⁢ ( r ) + o ⁢ ( T 𝐠 ⁢ ( r ) ) .

Now, we use the key inequality (5.2), the estimates in (3.5) and the two equations following it to derive

(5.7) ∑ | 𝐢 | = m , 𝐢 ∉ I z ( - log ⁡ | L z , 𝐢 ⁢ ( Φ ⁢ ( 𝐠 ⁢ ( z ) ) ) | + ∥ L z , 𝐢 ⁢ ( z ) ∥ )

≥ - ∑ | 𝐢 | = m , 𝐢 ∉ I z log ⁡ | 𝐠 ⁢ ( z ) 𝐢 | + O ⁢ ( 1 )

≥ - m n + 1 ⁢ ( m + n n ) ⁢ ∑ i = 0 n log ⁡ | g i ⁢ ( z ) | + M ′ ⁢ m ⁢ ∑ i = 0 n log ⁡ | g i ⁢ ( z ) |

- M ′ ⁢ m ⁢ n ⁢ max ⁡ { log ⁡ | g 0 ⁢ ( z ) | , … , log ⁡ | g n ⁢ ( z ) | } + O ⁢ ( 1 ) ,

for z ∈ ℂ that is not a pole of any b 𝐢 , j , z from (5.1). By Lemma 2.1, Jensen’s formula, the integration of (5.7) from 0 to 2 ⁢ π over d ⁢ θ gives

∫ 0 2 ⁢ π ∑ | 𝐢 | = m , 𝐢 ∉ I z ( - log ⁡ | L z , 𝐢 ⁢ ( Φ ⁢ ( 𝐠 ⁢ ( z ) ) ) | + ∥ L z , 𝐢 ⁢ ( z ) ∥ ) ⁢ d ⁢ θ 2 ⁢ π

≥ exc - ( m n + 1 ⁢ ( m + n n ) - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) - M ′ ⁢ m ⁢ n ⁢ T 𝐠 ⁢ ( r ) + o ⁢ ( T 𝐠 ⁢ ( r ) ) .

Together with (5.5) and (5.6), we have

M ⁢ N gcd ⁢ ( F ⁢ ( 𝐠 ) , G ⁢ ( 𝐠 ) , r ) + 1 u ⁢ N W ⁢ ( 0 , r )

≤ exc ( ( w u - 1 ) ⁢ M + M ′ ⁢ n + ε ) ⁢ m ⁢ T 𝐠 ⁢ ( r )

+ ( m n + 1 ⁢ ( m + n n ) - M ′ ⁢ m ) ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) + o ⁢ ( T 𝐠 ⁢ ( r ) ) .

Since F and G are coprime, the ideal ( F , G ) defines a closed subset of 𝔸 n of codimension at least 2. As is well known from the theory of Hilbert functions and Hilbert polynomials, this implies that M ′ = O ⁢ ( m n - 2 ) .

It then suffices to show that there exists a large integer L such that

c m , n , d ⁢ ∑ i = 0 n N g i ⁢ ( 0 , r ) - ( m + n - 2 ⁢ d n ) ⁢ N gcd ⁢ ( { 𝐠 𝐢 } 𝐢 ∈ I , r ) - 1 u ⁢ N W ⁢ ( 0 , r )

≤ c m , n , d ⁢ ∑ i = 0 n N g i ( L ) ⁢ ( 0 , r ) .

This inequality is deduced in a manner analogous to the corresponding part of the proof of Theorem 3.1, and so we omit the remaining details. ∎

Funding source: National Science Foundation

Award Identifier / Grant number: DMS-1352407

Funding source: Ministry of Science and Technology, Taiwan

Award Identifier / Grant number: 106-2115-M-001-001-MY2 and 108-2115-M-001-001-MY2

Funding statement: The first author was supported in part by NSF grant DMS-1352407. The second author was supported in part by Taiwan’s MoST grant 106-2115-M-001-001-MY2 and 108-2115-M-001-001-MY2.

References

[1] N. Ailon and Z. Rudnick, Torsion points on curves and common divisors of a k - 1 and b k - 1 , Acta Arith. 113 (2004), no. 1, 31–38. 10.4064/aa113-1-3Search in Google Scholar

[2] Y. Bugeaud, P. Corvaja and U. Zannier, An upper bound for the G.C.D. of a n - 1 and b n - 1 , Math. Z. 243 (2003), no. 1, 79–84. 10.1007/s00209-002-0449-zSearch in Google Scholar

[3] P. Corvaja and J. Noguchi, A new unicity theorem and Erdös’ problem for polarized semi-abelian varieties, Math. Ann. 353 (2012), no. 2, 439–464. 10.1007/s00208-011-0692-xSearch in Google Scholar

[4] P. Corvaja and U. Zannier, Finiteness of integral values for the ratio of two linear recurrences, Invent. Math. 149 (2002), no. 2, 431–451. 10.1007/s002220200221Search in Google Scholar

[5] P. Corvaja and U. Zannier, A lower bound for the height of a rational function at S-unit points, Monatsh. Math. 144 (2005), no. 3, 203–224. 10.1007/s00605-004-0273-0Search in Google Scholar

[6] M. L. Green, Some Picard theorems for holomorphic maps to algebraic varieties, Amer. J. Math. 97 (1975), 43–75. 10.2307/2373660Search in Google Scholar

[7] N. Grieve and J. T.-Y. Wang, Greatest common divisors with moving targets and linear recurrence sequences, preprint (2019), https://arxiv.org/abs/1902.09109. 10.1090/tran/8220Search in Google Scholar

[8] J. Guo, The quotient problem for entire functions, Canad. Math. Bull. 62 (2019), no. 3, 479–489. 10.4153/S0008439518000097Search in Google Scholar

[9] J. Guo, The quotient problem for entire functions with moving targets, Houston J. Math., to appear. Search in Google Scholar

[10] J. Guo and J. T.-Y. Wang, Asymptotic gcd and divisible sequences for entire functions, Trans. Amer. Math. Soc. 371 (2019), no. 9, 6241–6256. 10.1090/tran/7435Search in Google Scholar

[11] S. Lang, Introduction to complex hyperbolic spaces, Springer, New York 1987. 10.1007/978-1-4757-1945-1Search in Google Scholar

[12] A. Levin, Greatest common divisors and Vojta’s conjecture for blowups of algebraic tori, Invent. Math. 215 (2019), no. 2, 493–533. 10.1007/s00222-018-0831-zSearch in Google Scholar

[13] J. Noguchi and J. Winkelmann, Nevanlinna theory in several complex variables and Diophantine approximation, Grundlehren der Mathematischen Wissenschaften 350, Springer, Berlin 2014. 10.1007/978-4-431-54571-2Search in Google Scholar

[14] J. Noguchi, J. Winkelmann and K. Yamanoi, The second main theorem for holomorphic curves into semi-abelian varieties. II, Forum Math. 20 (2008), no. 3, 469–503. Search in Google Scholar

[15] A. Ostafe, On some extensions of the Ailon–Rudnick theorem, Monatsh. Math. 181 (2016), no. 2, 451–471. 10.1007/s00605-016-0911-3Search in Google Scholar

[16] H. Pasten and J. T.-Y. Wang, GCD bounds for analytic functions, Int. Math. Res. Not. IMRN (2017), no. 1, 47–95. 10.1093/imrn/rnw028Search in Google Scholar

[17] M. Ru, On a general form of the second main theorem, Trans. Amer. Math. Soc. 349 (1997), no. 12, 5093–5105. 10.1090/S0002-9947-97-01913-2Search in Google Scholar

[18] M. Ru, Nevanlinna theory and its relation to Diophantine approximation, World Scientific Publishing, River Edge 2001. 10.1142/4508Search in Google Scholar

[19] R. Rumely, Notes on van der Poorten’s proof of the Hadamard quotient theorem. I, II, Séminaire de Théorie des Nombres, Paris 1986–87, Progr. Math. 75, Birkhäuser, Boston (1988), 349–409. Search in Google Scholar

[20] J. H. Silverman, Arithmetic distance functions and height functions in Diophantine geometry, Math. Ann. 279 (1987), no. 2, 193–216. 10.1007/BF01461718Search in Google Scholar

[21] J. H. Silverman, Generalized greatest common divisors, divisibility sequences, and Vojta’s conjecture for blowups, Monatsh. Math. 145 (2005), no. 4, 333–350. 10.1007/s00605-005-0299-ySearch in Google Scholar

[22] A. J. van der Poorten, Solution de la conjecture de Pisot sur le quotient de Hadamard de deux fractions rationnelles, C. R. Acad. Sci. Paris Sér. I Math. 306 (1988), no. 3, 97–102. Search in Google Scholar

[23] P. Vojta, On Cartan’s theorem and Cartan’s conjecture, Amer. J. Math. 119 (1997), no. 1, 1–17. 10.1353/ajm.1997.0009Search in Google Scholar

[24] P. Vojta, Diophantine approximation and Nevanlinna theory, Arithmetic geometry, Lecture Notes in Math. 2009, Springer, Berlin (2011), 111–224. 10.1007/978-3-642-15945-9_3Search in Google Scholar

[25] J. T.-Y. Wang, An effective Roth’s theorem for function fields, Rocky Mountain J. Math. 26 (1996), no. 3, 1225–1234. 10.1216/rmjm/1181072046Search in Google Scholar

[26] K. Yamanoi, Holomorphic curves in abelian varieties and intersections with higher codimensional subvarieties, Forum Math. 16 (2004), no. 5, 749–788. 10.1515/form.2004.035Search in Google Scholar

[27] U. Zannier, Diophantine equations with linear recurrences. An overview of some recent progress, J. Théor. Nombres Bordeaux 17 (2005), no. 1, 423–435. 10.5802/jtnb.499Search in Google Scholar

Received: 2019-01-06

Revised: 2019-08-09

Published Online: 2019-11-09

Published in Print: 2020-10-01

This work is licensed under the Creative Commons Attribution 4.0 International License.

Greatest common divisors of analytic functions and Nevanlinna theory on algebraic tori

Abstract

1 Introduction

Theorem 1.1 (Noguchi, Winkelmann, Yamanoi [14, Theorem5.1] (see also [13, Sectiom 6.5])).

Theorem 1.2.

Theorem 1.3.

Corollary 1.4.

Theorem 1.5.

Corollary 1.6.

Remark 1.7.

Corollary 1.8.

Remark 1.9.

Remark 1.10.

Remark 1.11.

Theorem 1.12 (Bugeaud, Corvaja, Zannier [2]).

Theorem 1.13 (Corvaja, Zannier [5] (for case n = 2 ), Levin [12] (for case n > 2 )).

Theorem 1.14.

2 Background material

2.1 Nevanlinna theory over ℂ

Lemma 2.1.

Theorem 2.2 (First Main Theorem).

2.2 Nevanlinna theory for Cartier divisors

Theorem 2.3.

Theorem 2.4.

Lemma 2.5.

Proof.

Theorem 2.6.

Theorem 2.7.

2.3 Counting functions and proximity functions for closed subschemes

Lemma 2.8.

Proof.

Definition 2.9.

Lemma 2.10.

Proof.

2.4 Polynomial rings and monomial orderings

Lemma 2.11.

Proof.

3 Key theorem

Theorem 3.1.

Proof of Theorem 3.1.

4 Proof of the main theorems

Lemma 4.1 (Borel’s lemma).

Lemma 4.2.

Theorem 4.3.

Proof.

Proof of Theorem 1.2 and Theorem 1.3.

Proof of Theorem 1.5.

Proof of Corollary 1.6.

Proof of Corollary 1.8.

5 GCD with moving targets

5.1 Statement of the main results

Theorem 5.1.

Theorem 5.2.

Corollary 5.3.

5.2 Nevanlinna theory with moving targets

Theorem 5.4.

Lemma 5.5.

Lemma 5.6.

5.3 Key theorem

Theorem 5.7.

Proof of Theorem 5.7.

References

Journal and Issue

Articles in the same Issue