Abstract
Take complex numbers \(\alpha ,\beta ,c,a_j,b_j\)\((j=0,1,2)\) such that \(c\ne 0\) and
We show that if the following functional equation of Fermat type
has meromorphic solutions of finite order, then it has only entire solutions of the form \(f(z)=Ae^{\frac{\alpha z+\beta }{3}}+Ce^{Dz},\) where A, C, D are constants, which generalizes some results due to Han and Lü.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Take complex numbers \(\alpha ,\beta ,c\) with \(c\not =0\). We will characterize all global meromorphic solutions of the following functional equation of Fermat type
under the assumption:
- (A): :
-
Take six complex numbers \(a_i,b_i\) for \(i=0,1,2\) such that
$$\begin{aligned} \mathrm{rank} \left( \begin{array}{ccc} a_{0} &{} a_{1} &{} a_{2}\\ b_{0} &{} b_{1} &{} b_{2}\\ \end{array} \right) =2. \end{aligned}$$
This kind of problems goes back to classical functional equations of Fermat type
for a positive integer n. When \(n>3\), Gross [7] proved that the Eq. (1.2) has no nonconstant meromorphic solutions on the complex plane \({\mathbb {C}}\). If \(n=2\), Gross [7] showed that all meromorphic solutions of (1.2) on \({\mathbb {C}}\) are of the form
where \(\omega \) is a nonconstant meromorphic function on \({\mathbb {C}}\). If \(n=3\), the form of meromorphic solutions of (1.2) was conjectured by Gross [7], and was completely characterized by Baker [1] (see Sect. 2). Meanwhile, we refer the reader to [8, 9] and [29].
Take two positive integers m and n. When \(m\ge 3\) and \(n\ge 3\), Montel [29] proved that the following functional equation
has no transcendental entire solutions. Further, Yang [33] showed that the Eq. (1.3) has no nonconstant entire solutions if \(\lambda =1-\frac{1}{m}-\frac{1}{n}>0\). For more detail, we refer the reader to the work of Hu, Li and Yang [20]. For some works related to partial differential equations of Fermat type, see [4, 15, 25,26,27,28].
In 1985, Hayman [19] proved that the following functional equation
has no nonconstant meromorphic (resp. entire) solutions if \(n\ge 9\) (resp. \( n\ge 6\)). For the case \(n=5\) and \(n=6\), Gundersen [11, 12] proved the existence of transcendental meromorphic solutions of the Eq. (1.4). When \(n=8\), Ishizaki [21] showed that if f, g, h are nonconstant meromorphic functions satisfying (1.4), then there must exist a small function a(z) with respect to f, g, h such that
where the left side is a Wronskian. Ng and Yeung [31] proved that (1.4) has no nontrivial meromorphic (resp. entire) solutions when \(n\ge 8 \) (resp. \(n\ge 6\)). Gundersen [10] asked that do (1.4) have nonconstant meromorphic solutions for \(n=7\) or \(n=8\)? Futhermore, Ng and Yeung [31], Yang [36] studied the existence of meromorphic solutions of the equation
where n, m, p are positive integers.
More general, if \(k(\ge 3)\) positive integers \(n_1,\ldots ,n_k\) satisfy
Hu, Li and Yang [20] proved that the following functional equation
has no nonconstant entire solutions.
Based on the observation above, these functional equations maybe have global nonconstant meromorphic solutions when the powers are lower, which further are characterized by some researchers. Hence, a natural question follows:
Problem 1.1
When the powers \(n_j\) in the Eq. (1.5) are lower, can we characterize all meromorphic solutions f of the Eq. (1.5) if \(f_j\) are replaced by differentials or differences (even their mixture) of a fixed function f?
For example, we may try to characterize all meromorphic solutions f of the following difference equation
for a fixed nonzero constant c (cf. [29, 34]). For the case \(n>m=1\), Shimomura [32] proved that the Eq. (1.6) has an entire solution of infinite order. Later, Liu et al. [23] showed that the Eq. (1.6) has no transcendental entire solutions of finite order when \(n\ne m\). Liu et al. [24] illustrated that the solutions of (1.6) are periodic functions of period 2c for \(n=m=1,\) and found that (1.6) has transcendental entire solutions of finite order for \(n=m=2\).
In 2004, Yang and Li [35] proved that the following differential equation
has only transcendental entire solutions of the form
where \(P, \alpha \) are nonzero constants. However, the following differential equation
has no nonconstant global meromorphic solutions (cf. [14]).
Recently, Han and Lü [13] studied that the following difference equation
has only entire solutions of the form \(f(z)=Ae^{\frac{\alpha z+\beta }{3}}\) with a constant A satisfying \(A^{3}(1+e^{\alpha c})=1\). For more results related to differences of entire and meromorphic functions, see [5, 16,17,18, 23, 24, 32].
In this paper, we will discuss the existence of solutions to functional equations of Fermat type and obtain the following main result.
Theorem 1.2
Take complex numbers \(\alpha ,\beta ,c,a_i,b_i,i=0,1,2\) with \(c\not =0\), and assume \((\mathrm{A})\). If the Eq. (1.1) has meromorphic solutions of finite order, then it has only entire solutions of the following form
where A, C, D are constants. Moreover, if we define constants \(c_{0},c_{1}\) by \(c_{0}^{3}+c_{1}^{3}=1\), then A, D are completely determined by \(a_{i},b_{i},\alpha , c_{0},c_{1}\) as follows:
-
(1)
\(A=\frac{b_{2}c_{0}-a_{2}c_{1}}{a_{0}b_{2}-a_{2}b_{0}}\), \(C=0\) if \(a_{0}b_{1}-a_{1}b_{0}=0\), \(b_{1}a_{2}-a_{1}b_{2}=0\),
-
(2)
\(A=\frac{b_{1}c_{0}-a_{1}c_{1}}{a_{0}b_{1}-a_{1}b_{0}}\), \(C=0\) if \(a_{0}b_{1}-a_{1}b_{0}\ne 0\), \(b_{1}a_{2}-a_{1}b_{2}=0\),
-
(3)
\(A=\frac{3(b_{1}c_{0}-a_{1}c_{1})}{(b_1a_2-a_1b_2)(\alpha -3D)}\), \(C\in {\mathbb {C}}\), \(D=\frac{a_{1}b_{0}-a_{0}b_{1}}{b_{1}a_{2}-a_{1}b_{2}}\) if \(b_{1}a_{2}-a_{1}b_{2}\ne 0\).
In particular, if we take \(a_{0}=1\), \(a_{1}=a_{2}=0\), \(b_{1}=1\), \(b_{0}=b_{2}=0\) in Theorem 1.2, this is just the results in [13] and [14].
2 Preliminary
We assume that the reader is familiar with Nevanlinna theory (cf. [22, 30]) of meromorphic functions f in \({\mathbb {C}}\), such as the first main theorem of f, the second main theorem of f, the characteristic function T(r, f), the proximity function m(r, f), the counting functions N(r, f), \(\overline{N}(r,f)\), and S(r, f), where as usual S(r, f) denotes any quantity satisfying \(S(r,f)=o(T(r,f))\) as \(r\rightarrow \infty \) outside a possible exceptional set of finite logarithmic measure. Further, recall that the order of f is defined by
When \(n=3\) in the Eq. (1.2), Gross [7] showed that the Eq. (1.2) has a pair of meromorphic solutions
where \(\wp (z)\) is the Weierstrass elliptic function with periods \(\omega _{1}\) and \(\omega _{2}\), defined by
which is an even function and satisfies, after appropriately choosing \(\omega _{1}\) and \(\omega _{2}\),
Furthermore, Baker [1] provided the following important result.
Lemma 2.1
[1]. Any nonconstant meromorphic functions F(z), G(z) in the complex plane \({\mathbb {C}}\) satisfying
have the form \(F(z)=f(h(z)), G(z)=\eta g(h(z))=\eta f(-h(z))\), where f and g are elliptic functions defined by (2.1), h(z) is an entire function in \({\mathbb {C}}\) and \(\eta \) is a cube-root of the unity.
The following lemma is referred to Bergweiler [3] and Edrei and Fuchs [6].
Lemma 2.2
[3, 6]. Let f be a meromorphic function and h be an entire function in \({\mathbb {C}}\). When \(0<\rho (f),\)\(\rho (h)<\infty \), then \(\rho (f\circ h)=\infty \). When \(\rho (f\circ h)<\infty \) and h is transcendental, then \(\rho (f)=0\).
Lemma 2.3
[20]. If f is a nonconstant meromorphic function, then
where \(a_{j}\) are meromorphic functions with \(a_{p}\not \equiv 0\) and
Lemma 2.4
[5]. Let f(z) be a meromorphic function such that the order \(\rho (f)<+\infty \), and let c be a nonzero complex number. Then for each \(\varepsilon >0\), we have
where \(f_c(z)=f(z+c)\).
3 Proof of Theorem 1.2
First of all, we rewrite (1.1) into the form (2.3), where F and G are defined by
Then we claim that F(z), G(z) are constants.
We assume, to the contrary, that F(z), G(z) are not constants. By Lemma 2.1, we have
Based on ideas in [13] and [14], we confirm a fact as follows:
Claim 1:h(z) must be a nonconstant polynomial.
Solving \(\wp '(h(z))\) from the first expression of (3.2), we obtain
Substituting the above equality into (2.2), it follows that
which immediately implies the following inequality of Nevanlinna’s characteristic functions
that is, \(T(r,\wp (h))\le 2T(r,F)+O(1).\) Hence, we have \(\rho (\wp (h))\le \rho (F)\). The first expression of (3.1) means that \(\rho (F)<\infty \) since \(\rho (f)<\infty \) by the assumption. Therefore, we obtain \(\rho (\wp (h))<\infty \).
Bank and Langley [2, Equation (2.7)] gave Nevanlinna’s characteristic function of \(\wp \) as follows:
where A is the area of a parallelogram \(P_{a}\) with four vertices \(0,\omega _{1},\omega _{2},\omega _{1}+\omega _{2}\), which further yields \(\rho (\wp )=2\). Then h must be a polynomial based on Lemma 2.2. Further, by using (3.2), we know that h(z) is not a constant.
Next we distinguish three cases to prove Theorem 1.2.
Case 1:\(a_{0}b_{1}-a_{1}b_{0}=0\), \(b_{1}a_{2}-a_{1}b_{2}=0\).
Then we have \(a_{0}b_{2}-b_{0}a_{2}\ne 0\) by the rank assumption. Solving the Eqs. (3.1) and noting (3.2), it follows that
By differentiating (3.5) and noting that \((\wp ')^{2}=4\wp ^{3}-1\), \(\wp ''=6\wp ^{2}\), we have
Substituting (3.5) and (3.6) into the first equation of (3.1), we have
Let \(\{z_{j}\}^{\infty }_{j=1}\) be the zeroes of \(\wp \) satisfying \(|z_{j}|\rightarrow \infty \) as \(j\rightarrow \infty \). Then for each j, there exist \(\deg (h)\) complex numbers \(a_{j,k}\) such that \(h(a_{j,k})=z_j\), for \(\ k=1,\ldots ,\deg (h)\). Moreover, the Eq. (2.2) implies \((\wp ')^{2}(h(a_{j,k}))=(\wp ')^{2}(z_{j})=-1\), since \(\wp (z_{j})=0\).
Now, we confirm the second fact as follows:
Claim 2:\(\wp (h(a_{j,k}+c))=0\) only holds for at most finitely many \(a_{j,k}\)’s.
We assume, to the contrary, that there exists an infinite subsequence of \(\{a_{j,k}\}\), without loss of generality we may take \(\{a_{j,k}\}\) itself, such that \(\wp (h(a_{j,k}+c))=0\). Thus we have \((\wp ')^{2}(h(a_{j,k}+c))=-1\). Differentiating (3.7) and then setting \(z=a_{j,k}\), we derive a relation
Note that \(\wp '(h(a_{j,k}))=\pm i\), where i is the imaginary unit. If \(a_{2}\ne 0\), the Eq. (3.8) becomes
Since h is a nonconstant polynomial and \(\{a_{j,k}\}\) is an infinite sequence, then there are infinitely many \(a_{j,k}\) such that \(h'(a_{j,k})h'(a_{j,k}+c)\ne 0\). It follows that
or
and hence \(b_2+e^{\frac{\pm 2\pi i}{3}}\eta a_2=0.\) We may let \(b_2=-k a_2,\) where \(k=e^{\frac{\pm 2\pi i}{3}}\eta \). Note that \(a_{2}\ne 0\), then we get \( b_1=-k a_1\) because the assumption \(b_{1}a_{2}-a_{1}b_{2}=0\). Combining with the assumption \(a_{0}b_{1}-a_{1}b_{0}=0\), we obtain \(ka_{0}a_{1}+a_{1}b_{0}=0.\)
If \(a_{1}\ne 0\), \(ka_{0}a_{1}+a_{1}b_{0}=0\) implies \( b_0=-k a_0.\) It follows that \(b_i=-ka_i\) for \(i=0,1,2,\) which contradicts to the rank assumption.
When \(a_{1}=0\), by using (3.2), we can rewrite the first equation of (3.1) as follows
By using the theory of first-order linear differential equations, the solution of (3.9) is
where C is a constant. Let \(t_j\ (j\ge 1)\) be the poles of \(\wp (z)\). Then we have \( \wp (z)=g_j(z)(z-t_{j})^{-2},\) where \(g_j\) is a holomorphic function in a neighborhood of \(t_j\) with \(g_j(t_j)\not =0\), and hence
as \(z\rightarrow t_j\). Setting \(z=h(z)\), we derive a relation
as \(h(z)\rightarrow t_j\). Note that the equation \(h'(z)=0\) only has finitely many solutions, but \(\bigcup \limits ^{\infty }_{j=1}h^{-1}(t_{j})\) is an infinite set. Thus there exist an integer j and a point \(z'\in h^{-1}(t_{j})\) such that \(h'(z')\ne 0\), that is, \(z'\) is a simple zero of \(h(z)-t_j\). Therefore f(z) has a logarithmic singular point \(z'\), which contradicts the assumption that f(z) is a meromorphic function.
Next we consider the case \(a_2=0\). Now we claim that \(a_{1}=0\). Otherwise, if \(a_{1}\ne 0\), the assumption \(b_{1}a_{2}-a_{1}b_{2}=0\) implies \(b_{2}=0\), which contradicts the rank assumption \(b_{2}a_{0}-a_{2}b_{0}\ne 0\). Since \(a_1=a_2=0\), we find \(b_{2}a_{0}\ne 0\) from \(b_{2}a_{0}-a_{2}b_{0}\ne 0\) and \(a_{0}b_{1}=0\) because the assumption \(a_{0}b_{1}-a_{1}b_{0}=0\). Thus it follows that \(a_{0}\ne 0\), \(b_{1}=0\), \(b_{2}\ne 0\). By using (3.2), we can rewrite the Eq. (3.1) as follows
Differentiating the first relation of (3.11) and noticing that \((\wp ')^{2}=4\wp ^{3}-1\), \(\wp ''=6\wp ^{2}\), it follows that
Substituting f and \(f'\) into the second equation of (3.11), we have
Therefore, by using Lemma 2.3, we find
that is \(T(r,\wp (h))=O(\log r),\) which means that \(\wp (h)\) is a rational function. This is a contradiction because \(\wp (h)\) is a transcendental meromorphic function. Hence Claim 2 is proved.
Based on Claim 2 and the fact that \(h'\) has only finitely many zeroes, there exists a positive integer J such that
Note that \(a_2\not =0\), otherwise, we can obtain a contradiction according to the above arguments. Now, return to the Eq. (3.7). We see that the coefficient of \(\wp (h(z+c))\) at \(a_{j,k}\) takes a nonzero value
when \(j>J, \ k=1,\ldots ,\deg (h)\), because
see the proof of Claim 2. Therefore, the Eq. (3.7) valued at \(a_{j,k}\) immediately yields
which further yields
where \(h_c(z)=h(z+c)\).
Note that the multiple zeros of \(\wp (h)\) occur at zeros of its derivative \(\{\wp (h)\}'=\wp '(h)h'\), that is, the zeros of \(h'\) because \(\wp '(h(a_{j,k}))=\pm i\not =0\). Hence we obtain an estimate
Further, we claim
In fact, the expression of F in (3.2) yields
Based on the factorization \(-G^{3}=F^{3}-1=(F-1)(F-\eta )(F-\eta ^{2}),\) where \(\eta \not =1\), we see that all zeros of \(F-1,F-\eta \) and \(F-\eta ^{2} \) are of multiplicities \(\ge 3\). Hence Nevanlinna’s main theorems give
which immediately implies the claim \( m(r,F)=S(r,\wp (h)).\)
Now we rewrite (3.2) into the following form
which means
Applying the lemma of logarithmic derivative, we have
and hence
Combining (3.13) with (3.14), and noticing that each pole of \(\wp (z)\) has multiplicity 2, then Nevanlinna’s first main theorem implies
where Lemma 2.4 was applied. We obtain a contradiction again.
Hence F(z) and G(z) are constants. We assume that \(F(z)=c_{0}, G(z)=c_{1}\), where \(c_{0},c_{1}\) are constants with \(c_{0}^{3}+c_{1}^{3}=1\). By (3.5), we have the solution
This proves Case 1 in Theorem 1.2.
Case 2:\(a_{0}b_{1}-a_{1}b_{0}\ne 0\), \(b_{1}a_{2}-a_{1}b_{2}=0\).
Solving the Eqs. (3.1) and noting (3.2), it follows that
By differentiating (3.15) and noting that \((\wp ')^{2}=4\wp ^{3}-1\), \(\wp ''=6\wp ^{2}\), we have
Substituting (3.15) and (3.16) into the first equation of (3.1), we have
Similar to Claim 2, we prove the following fact.
Claim 3:\(\wp (h(a_{j,k}+c))=0\) only holds for at most finitely many \(a_{j,k}\)’s.
We assume, to the contrary, that there exists an infinite subsequence of \(\{a_{j,k}\}\), without loss of generality we may take \(\{a_{j,k}\}\) itself, such that \(\wp (h(a_{j,k}+c))=0\). Thus we have \((\wp ')^{2}(h(a_{j,k}+c))=-1\). Differentiating (3.17) and then setting \(z=a_{j,k}\), we derive a relation
Note that \(\wp '(h(a_{j,k}))=\pm i\), where i is the imaginary unit. If \(a_{2}\ne 0\), the Eq. (3.18) becomes
Since h is a nonconstant polynomial and \(\{a_{j,k}\}\) is an infinite sequence, then there are infinitely many \(a_{j,k}\) such that \(h'(a_{j,k})h'(a_{j,k}+c)\ne 0\). It follows that
Let \(t_j\ (j\ge 1)\) be the poles of \(\wp (z)\) satisfying \(|t_{j}|\rightarrow \infty \) as \(j\rightarrow \infty \) and take \(b_{j,k}\in {\mathbb {C}}\) satisfying \(h(b_{j,k})=t_j\) for \(k=1,\ldots ,\deg (h)\). Then there exists an integer \(j_0\) such that when \(j>j_0\), \(b_{j,k}\) are simple zeros of \(h(z)-t_j\) and \(h(z+c)-t_j\) has only simple zeros. Thus, the unique term \( 2(b_{1}+a_{1}\eta )\wp ^{3}(h(z))h'(z)/{\sqrt{3}} \) with poles of multiplicity 6 at \(b_{j,k}\ (j>j_0)\) must vanish, that is, \(b_{1}+a_{1}\eta =0\). Combining with (3.19), we obtain \(a_1=b_1=0\). This contradicts the assumption \(a_0b_1-a_1b_0\not =0\).
If \(a_{2}=0\), then we have \(a_{1}b_{2}=0\) by the assumption \(b_{1}a_{2}-a_{1}b_{2}=0\). Hence we distinguish two cases depending on whether \(a_{1}\) is zero or not as follows.
If \(a_{1}=0\), then we have \(a_{0}b_{1}\ne 0\) by the assumption \(a_{0}b_{1}-a_{1}b_{0}\ne 0\). Now, by using (3.2), we can rewrite the Eq. (3.1) as follows
Differentiating the first relation of (3.20) and noticing that \((\wp ')^{2}=4\wp ^{3}-1\), \(\wp ''=6\wp ^{2}\), it follows that
Substituting f and \(f'\) into the second equation of (3.20), we have
Differentiating (3.22) and then setting \(z=a_{j,k}\), we derive a relation
Note that \(\wp '(h(a_{j,k}))=\pm i\), where i is the imaginary unit. If \(b_{2}\ne 0\), (3.23) derives a contradiction. Hence we have \(b_2=0\). Now we can rewrite the second equation of (3.20) as follows
Substituting the first relation of (3.20) into the above equation, we have
Differentiating the above equation and then setting \(z=a_{j,k}\), we derive a relation
Note that \(\wp '(h(a_{j,k}))=\pm i\) and \(\wp '(h(a_{j,k}+c))=\pm i\), where i is the imaginary unit. The above equation immediately implies one and only one of the following four situations
where
Obviously, \(A_1, A_2\) are nonzero constants by the assumption \(a_{0}b_{1}\ne 0\). Since h is a nonconstant polynomial and \(\{a_{j,k}\}\) is an infinite sequence, the relations (3.25) immediately yield functional equations
which further mean that one of the following four equalities
holds by comparing the leading coefficient.
Thus we obtain \(h'(z)=h'(z+c)\), which implies \(h(z)=az+b\), where \(a(\ne 0)\) and b are constants. Now the equations \(\wp (h(a_{j,k}))=0,\ \wp (h(a_{j,k}+c))=0\) become \( \wp (aa_{j,k}+b)=0,\ \wp (aa_{j,k}+b+ac)=0,\) that is, \(\{aa_{j,k}+b,aa_{j,k}+b+ac\}\subset \{z_j\}_{j=1}^\infty \). Hence we have
since \(\wp \) has only two distinct zeros in the parallelogram \(P_{a}\) and \(\wp \) is a function of double periods \(\omega _1\) and \(\omega _2\).
If \(ac=m\omega _1+n\omega _2\) for some \(m,n\in {\mathbb {Z}}\), the Eq. (3.24) becomes
since \(\wp (az+b)=\wp (az+b+ac)\), and hence \(a_{0}\eta +b_{0}+b_{1}e^{\frac{\alpha c}{3}}=0,\ a_{0}\eta -b_{0}-b_{1}e^{\frac{\alpha c}{3}}=0,\) because \(\wp '(az+b)\) is a transcendental meromorphic function. We can obtain \(a_{0}=0\), which is a contradiction.
When \(ac\in P_a\), we rewrite (3.24) into the form
It is obvious that the function on the left-hand side of (3.28) has pole at \(z=-\frac{b}{a}\), but the function on the right-hand side of (3.28) take a finite value. This leads to a contradiction.
Therefore, we must have \(a_1\not =0\). By our assumptions \(a_{2}=0\) and \(b_{1}a_{2}-b_{2}a_{1}=0\), then \(b_{2}=0\). Now, by using (3.2), we can rewrite the Eq. (3.1) as follows
Solving the Eq. (3.29), it follows that
which derives an equation
Differentiating the above equation and then setting \(z=a_{j,k}\), we derive a relation
Note that \(\wp '(h(a_{j,k}))=\pm i\) and \(\wp '(h(a_{j,k}+c))=\pm i\), where i is the imaginary unit. The above equation immediately implies one and only one of (3.25), where
Note that h is a nonconstant polynomial and \(\{a_{j,k}\}\) is an infinite sequence, immediately yields the corresponding functional Eq. (3.26), which further means that one of the following four equalities
holds by comparing the leading coefficient.
In the following, we only consider the case \(A_1=B_1\) in (3.32), and omit the others due to the similarity of their arguments. Note that \(\wp '(h(a_{j,k}))=\pm i\). W. l. o. g., we assume \(\wp '(h(a_{j,k}))= i\). Now, if \(A_1\) is zero. Further, \(a_{j,k}\) is also a zero of
Assume that \(a_{j,k}\) is a zero of \(\wp (h(z))\) with multiplicity l. It follows from (3.30) that \(a_{j,k}\) may be a pole of f(z) and \(f(z+c)\) of multiplicity \(\le l-1\). However, it follows from (3.29) that \(a_{j,k}\) is a pole of \(a_{0}f(z)+a_{1}f(z+c)\) with multiplicity l. This is a contradiction.
Hence we have \(A_1\not =0\). Similarly, if \(\wp '(h(a_{j,k}))=- i\), we also obtain \(A_2\not =0\). Any way, we also have \(B_1\not =0, B_2\not =0\). Thus we obtain \(h'(z)=h'(z+c)\) again from (3.26), that is, \(h(z)=az+b\), where \(a(\ne 0)\) and b are constants. Next we can derives a contradiction according to the arguments between (3.26) and (3.28). Hence Claim 3 is proved.
Based on Claim 3, we can obtain a contradiction according to the proof of Case 1 after Claim 2.
Hence \(F=c_0\) and \(G=c_1\) are constants such that \(c_0^3+c_1^3=1\). By solving (3.15), we obtain a solution
This is just Case 2 in Theorem 1.2.
Case 3:\(b_{1}a_{2}-a_{1}b_{2}\ne 0\).
By using the theory of first-order linear differential equations, the above equation has the solution
where C is a constant.
If \(b_{1}+a_{1}\eta \ne 0\), letting \(t_j\ (j\ge 1)\) be the poles of \(\wp (z)\), then we have \(\wp (z)=g_j(z)(z-t_{j})^{-2},\) where \(g_j\) is a holomorphic function in a neighborhood of \(t_j\) with \(g_j(t_{j})\ne 0\), and hence
as \(z\rightarrow t_{j}\). Setting \(z=h(z)\), we derive a relation
as \(h(z)\rightarrow t_j\). Note that the equation \(h'(z)=0\) has only finitely many solutions, and \(\bigcup \nolimits ^{\infty }_{j=1}h^{-1}(t_{j})\) is an infinite set. There exist an integer j and a point \(z'\in h^{-1}(t_{j})\) such that \(h'(z')\ne 0\), that is, \(z'\) is a simple zero of \(h(z)-t_j\). It follows from (3.34) that f(z) has a logarithmic singular point \(z'\), which contradicts the assumption that f(z) is a meromorphic function.
If \(b_{1}+a_{1}\eta =0\), we have \(b_{1}-a_{1}\eta \ne 0\). Otherwise, if \(b_{1}-a_{1}\eta =0\), that is, \(b_1=0\) and hence \(a_1=0\), which contracted the assumption \(b_1a_2-a_1b_2\not =0\). Now, we can rewrite the Eq. (3.33) as follows
Similarly, there exist an integer j and a point \(z''\in h^{-1}(z_{j})\) such that \(h'(z'')\ne 0\), that is, f(z) has a logarithmic singular point \(z''\) since \(\wp \) has only simple zeros. We obtain a contradiction again.
Hence \(F=c_0\) and \(G=c_1\) are constants such that \(c_0^3+c_1^3=1\). Thus (3.33) admits a solution
where \(D=\frac{a_{1}b_{0}-a_{0}b_{1}}{b_{1}a_{2}-a_{1}b_{2}}\) is a constant. This is just Case 3 in Theorem 1.2. The proof of Theorem 1.2 is completed.
References
Baker, I.N.: On a class of meromorphic functions. Proc. Am. Math. Soc. 17(4), 819–822 (1966)
Bank, S.B., Langley, J.K.: On the value distribution theory of elliptic functions. Mon. Math. 98(1), 1–20 (1984)
Bergweiler, W.: Order and lower order of composite meromorphic functions. Mich. Math. J. 36, 135–146 (1989)
Chang, D.C., Li, B.Q.: Description of entire solutions of eiconal type equations. Can. Math. Bull. 55, 249–259 (2012)
Chiang, Y.M., Feng, S.: On the Nevanlinna characteristic of \(f(z +\eta )\) and difference equations in the complex plane. Ramanujan J. 16, 105–129 (2008)
Edrei, A., Fuchs, W.H.J.: On the zeros of \(f(g(z))\) where \(f\) and \(g\) are entire functions. J. Anal. Math. 12, 243–255 (1964)
Gross, F.: On the equation \( f^{n}+ g^{n}=1\). I. Bull. Am. Math. Soc. 72(1), 86–88 (1966)
Gross, F.: On the functional equation \(f^{n}+ g^{n}= h^{n}\). Am. Math. Mon. 73(10), 1093–1096 (1966)
Gross, F.: On the equation \( f^{n}+ g^{n}=1\). II. Bull. Am. Math. Soc. 74, 647–767 (1968)
Gundersen, G.G.: Research questions on meromorphic functions and complex differential equations. Comput. Methods Funct. Theory 17, 195–209 (2017)
Gundersen, G.G.: Meromorphic solutions of \(f^{5}+g^{5}+h^{5}\equiv 1\). Complex Var. Theory Appl. 43, 293–298 (2001)
Gundersen, G.G.: Meromorphic solutions of \(f^{6}+g^{6}+h^{6}\equiv 1\). Analysis (Munich) 18, 285–290 (1998)
Han, Q., Lü, F.: On the functional equation \(f^{n} (z)+ g^{n} (z)= e^{\alpha z+\beta }\). arXiv:1612.06842v1 (2016)
Han, Q., Lü, F.: On the Fermat-type equation \(f^3(z)+f^3(z+c)=1\). Aequationes Math. 91(1), 129–136 (2017)
Han, Q.: On complex analytic solutions of the partial differential equation \((u_{z_{1}} )^{m} + (u_{z_{2}} )^{n} = u^{m}\). Houst. J. Math. 35, 277–289 (2009)
Halburd, R.G., Korhonen, R.J.: Nevanlinna theory for the difference operator. Ann. Acad. Sci. Fenn. Math. 31, 463–478 (2006)
Halburd, R.G., Korhonen, R.J.: Difference analogue of the lemma on the logarithmic derivative with applications to difference equations. J. Math. Anal. Appl. 314(2), 477–487 (2006)
Halburd, R.G., Korhonen, R.J.: Meromorphic solutions of difference equations, integrability and the discrete P ainlevé equations. J. Phys. A. 40, 1–38 (2007)
Hayman, W.: Waring’s problem für analytische Funktionen. Bayer. Akad. Wiss. Math. Natur. Kl. Sitzungsber. 1984, 1–13 (1985)
Hu, P.C., Li, P., Yang, C.C.: Unicity of meromorphic mappings. Adv. Complex Anal. Appl. Kluwer Academic Publishers (2003)
Ishizaki, K.: A note on the functional equation \(f^{n}+g^{n}+ h^{n}=1\) and some complex differential equations. Comput. Methods Funct. Theory 2(1), 67–85 (2003)
Laine, I.: Nevanlinna Theory and Complex Differential Equations. Walter de Gruyter, Berlin (1993)
Liu, K., Yang, L.Z., Liu, X.L.: Existence of entire solutions of nonlinear difference equations. Czechoslov. Math. J. 61, 565–576 (2011)
Liu, K., Cao, T.B., Cao, H.Z.: Entire solutions of Fermat type differential-difference equations. Arch. Math. (Basel) 99, 147–155 (2012)
Li, B.Q.: On meromorphic solutions of \(f^{2} + g^{2} = 1\). Math. Z. 258, 763–771 (2008)
Li, B.Q., Ye, Z.: On meromorphic solutions of \(f^{3} + g^{3} = 1\). Arch. Math. (Basel) 90, 39–43 (2008)
Li, B.Q.: Entire solutions of \((u_{z_{1}} )^{m} + (u_{z_{2}} )^{n} = e^{g}\). Nagoya Math. J. 178, 151–162 (2005)
Li, B.Q.: Entire solutions of eiconal type equations. Arch. Math. (Basel) 89, 350–357 (2007)
Montel, P.: Leçons sur les familles normales de fonctions analytiques et leurs applications, pp. 135–136. Collect. Borel, Paris (1927)
Nevanlinna, R.: Analytic Functions. Springer, Berlin (1970)
Ng, T.W., Yeung, S.K.: Entire holomorphic curves on a Fermat surface of low degree. arXiv:1612.01290 (2016)
Shimomura, S.: Entire solutions of a polynomial difference equation. J. Fac. Sci. Univ. Tokyo Sect. IA Math. 28, 253–266 (1981)
Yang, C.C.: A generalization of a theorem of P. Montel on entire functions. Proc. Am. Math. Soc. 26, 332–334 (1970)
Yang, C.C., Laine, I.: On analogies between nonlinear difference and differential equations. Proc. Jpn. Acad. Ser. A. 86, 10–14 (2010)
Yang, C.C., Li, P.: On the transcendental solutions of a certain type of nonlinear differential equations. Arch. Math. 82, 442–448 (2004)
Yang, L.Z., Zhong, Z., Zhang, J.L.: Non-existence of meromorphic solutions of a Fermat type functional equation. Aequationes Math. 76, 140–150 (2008)
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Mohammad Sal Moslehian.
This work of both authors was partially supported by NSFC of China (Nos. 11461070, 11271227), PCSIRT (No. IRT1264), and the Fundamental Research Funds of Shandong University (No. 2017JC019).
Rights and permissions
About this article
Cite this article
Hu, PC., Wang, Q. On Meromorphic Solutions of Functional Equations of Fermat Type. Bull. Malays. Math. Sci. Soc. 42, 2497–2515 (2019). https://doi.org/10.1007/s40840-018-0613-1
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40840-018-0613-1