Results on the growth of solutions of complex linear differential equations with meromorphic coefficients

Author(s): Mansouria Saidani1, Benharrat Belaïdi1
1Department of Mathematics, Laboratory of Pure and Applied Mathematics, University of Mostaganem (UMAB), B. P. 227 Mostaganem-(Algeria)
Copyright © Mansouria Saidani, Benharrat Belaïdi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

The purpose of this paper is the study of the growth of solutions of higher order linear differential equations f(k)+(Ak1,1(z)ePk1(z)+Ak1,2(z)eQk1(z))f(k1)++(A0,1(z)eP0(z)+A0,2(z)eQ0(z))f=0 and f(k)+(Ak1,1(z)ePk1(z)+Ak1,2(z)eQk1(z))f(k1)++(A0,1(z)eP0(z)+A0,2(z)eQ0(z))f=F(z), where Aj,i(z)(0)(j=0,,k1;i=1,2), F(z) are meromorphic functions of finite order and Pj(z),Qj(z) (j=0,1,,k1;i=1,2) are polynomials with degree n1. Under some others conditions, we extend the previous results due to Hamani and Belaïdi [1].

Keywords: Order of growth; Hyper-order; Exponent of convergence of zero sequence; Differential equation; Meromorphic function.

1. Introduction and main results

Throughout this work, we assume that the reader knows the standard notations and the fundamental results of the Nevanlinna value distribution theory of meromorphic functions as the order and the hyper-order of growth, the convergence exponents of the zero-sequence and of distinct zeros, the hyper convergence exponents of the zero-sequence and the distinct zeros of a meromorphic function f, see [2-5].

We recall also the following definitions. The linear measure of a set E[0,+) is defined as m(E)=0+χE(t)dt and the logarithmic measure of a set F[1,+) is defined by lm(F)=1+χF(t)tdt, where χH(t) is the characteristic function of a set H.

For results on the growth of solutions of the complex linear differential equation f(k)+Ak1(z)ePk1(z)f(k1)++A1(z)eP1(z)f+A0(z)eP0(z)f=0, where Pj(z)=aj,nzn++aj,0 are polynomials with degree n1, aj,q (j=0,1,,k1;q=0,1,,n) are complex numbers and Aj(z) (0),  (j=0,1,,k1) are entire or meromorphic functions of finite order less than n, the reader is referred to [1,6-8].

Recently, Hamani and Belaı̈di [1] studied the order of transcendental meromorphic solutions of the homogeneous and the non-homogeneous linear differential equations (1) f(k)+hk1(z)ePk1(z)f(k1)++h1(z)eP1(z)f+h0(z)eP0(z)f=0,, (2) f(k)+hk1(z)ePk1(z)f(k1)++h1(z)eP1(z)f+h0(z)eP0(z)f=F,, and have proved the following results;

Theorem 1. [1] Let k2 be an integer and Pj(z)=ni=0aj,izi (j=0,1,,k1) be nonconstant polynomials with degree n1, where aj,0, aj,1,, aj,n (j=0,1,,k1) are complex numbers. Let hj(z) (j=0,1,,k1) be meromorphic functions ρ(hj)<n. Suppose that there exists s,d{0,1,,k1} such that hs hd0,as,n=|as,n|eiθs,  ad,n=|ad,n|eiθd,θs,θd[0,2π),θsθd then for j{0,1,,k1}{s,d},  aj,n satisfies either aj,n=cjas,n  or aj,n=cjad,n (0<cj<1). Then every transcendental meromorphic solution f whose poles are of uniformly bounded multiplicities of Eq. (1) is of infinite order and satisfies ρ2(f)=n.

Theorem 2. [1] Let k2 be an integer, hj(z),Pj(z) and an,j satisfy the hypotheses of Theorem . Let  F(0) be a meromorphic function of order ρ(F)<n. Then every transcendental meromorphic solution f whose poles are of uniformly bounded multiplicities of Equation (2)  is of infinite order and satisfies λ¯2(f)=λ2(f)=ρ2(f)=n, with at most one exceptional solution f0 of finite order.

In this paper, we continue to study the oscillation problem of solutions, we improve and extend Theorem 2 and Theorem 2 for equations of the form (3) f(k)+(Ak1,1(z)ePk1(z)+Ak1,2(z)eQk1(z))f(k1)++(A0,1(z)eP0(z)+A0,2(z)eQ0(z))f=0,, and (4) f(k)+(Ak1,1(z)ePk1(z)+Ak1,2(z)eQk1(z))f(k1)++(A0,1(z)eP0(z)+A0,2(z)eQ0(z))f=F., We obtain the following results;

Theorem 3. Let k2 be an integer and Pj(z)=aj,nzn++aj,0, Qj(z)=bj,nzn++bj,0 be polynomials with degree n1, where aj,q,bj,q (j=0,1,,k1;q=0,1,,n) are complex numbers such that aj,nbj,n0. Let Aj,i(z) (0)  (j=0,1,,k1;i=1,2) be meromorphic functions such that max{ρ(Aj,i):j=0,1,,k1;i=1,2}<n. Suppose that there exist s,d{0,1,,k1} such that As,1 Ad,10,As,2 Ad,20,as,n=|as,n|eiθs,  ad,n=|ad,n|eiθd, bs,n=|bs,n|eiφ, θs, θd, φ[0,2π), θsθd, then for j{0,1,,k1}{s,d}, an,j and bj,n satisfies either aj,n=cjas,n or aj,n=cjad,n,bj,n=cjbs,n (0<cj<1,0<cj<1). Then every transcendental meromorphic solution f whose poles are of uniformly bounded multiplicities of Equation (3)  is of infinite order and satisfies ρ2(f)=n.

Corollary 1. Let k2 be an integer and Pj(z)=aj,nzn++aj,0, Qj(z)=bj,nzn++bj,0 be polynomials with degree n1, where aj,q,bj,q (j=0,1,,k1;q=0,1,,n) are complex numbers such that aj,nbj,n0. Let Aj,i(z) (0)  (j=0,1,,k1;i=1,2) be entire functions such that max{ρ(Aj,i):j=0,1,,k1;i=1,2}<n. Suppose that there exist s,d{0,1,,k1} such that As,1 Ad,10,As,2 Ad,20,as,n=|as,n|eiθs, ad,n=|ad,n|eiθd, bs,n=|bs,n|eiφ, θs, θd, φ[0,2π), θsθd, then for j{0,1,,k1}{s,d}, aj,n and bj,n satisfies either aj,n=cjas,n or aj,n=cjad,n,bj,n=cjbs,n (0<cj<1,0<cj<1). Then every transcendental solution f of Equation (3)  is of infinite order and satisfies ρ2(f)=n.

Example 1. Consider the following differential equation (5) f(4)+(2izeiz2+z22e2iz2)f(3)+(2z2e2iz2iz3eiz2)f+((24iz4+12iz3)e2iz2+(4iz5+(64i)z3)eiz2)f +(10eiz2+(4iz5+8z4+6z34iz2)eiz2)f=0., Set {A0,1(z)=10, A0,2(z)=4iz5+8z4+6z34iz2,A1,1(z)=24iz4+12iz3, A1,2(z)=4iz5+(64i)z3,A2,1(z)=2z2, A2,2(z)=iz3,A3,1(z)=2iz, A2,1(z)=z22 and {P0(z)=P3(z)=iz2,P1(z)=2iz2,P2(z)=2iz2,Q0(z)=Q1(z)=Q2(z)=iz2,Q3(z)=2iz2. We have a0,2=i, a1,2=2i=as,2, a2,2=2i=ad,2, a3,2=i, we can see that {a0,2=i=12as,2, c0=12, 0<c0<1,a3,2=i=12as,2, c3=12, 0<c3<1,argas,2argad,2, and b0,2=b1,2= b2,2=i,b3,2=2i, we can see that {b0,2=i=12b3,2,b1,2=i=12b3,2,b2,2=i=12b3,2,cj=12, 0<cj<1, j=0,1,2 and max{ρ(Aj,i):j=0,,3;i=1,2}<2. Then, according to Corollary 1, every transcendental solution f of Eq. (5) satisfies ρ(f)=+ and ρ2(f)=2. We can see that f(z)=eeiz2 represents a solution of Eq. (5) that satisfies ρ(f)=+  and ρ2(f)=2.

For the case of non-homogeneous equation, we have the following result;

Theorem 4. Let k2 be an integer, Pj(z), Qj(z),  Aj,i, aj,n, bj,n, (j=0,1,,k1) satisfy the hypotheses of Theorem 3. Let F(0)  be a meromorphic function of order ρ(f)<n. Then every transcendental meromorphic solution f whose poles are of uniformly bounded multiplicities of Equation (4) satisfies λ¯(f)=λ(f)=ρ(f)=+ and λ¯2(f)=λ2(f)=ρ2(f)=n,  with at most one exceptional solution f0 of finite order.

Corollary 2. Let k2 be an integer, Pj(z), Qj(z),  Aj,i, aj,n, bj,n, (j=0,1,,k1) satisfy the hypotheses of Corollary 1. Let F(0)  be an entire function of order ρ(f)<n. Then every transcendental solution f of Equation (4) satisfies λ¯(f)=λ(f)=ρ(f)=+ and λ¯2(f)=λ2(f)=ρ2(f)=n, with at most one exceptional solution f0 of finite order.

Example 2. Consider the following differential equation (6) f(3)+((z1)ez+(z2+z+2+1z)e2z)f+((z2+z)e2z+(2z24z2z1)ez)f +((z2z+1)ez(z2+z+2+1z)ez)f=2z3+2z2+z+1., Set {A0,1(z)=z2z+1, A0,2(z)=z2z21z,A1,1(z)=z2+z, A1,2(z)=2z24z2z1,A2,1(z)=z1, A2,2(z)=z2+z+2+1z and {P0(z)=z, P1(z)=2z,P2(z)=z,Q0(z)=Q1(z)=z, Q2(z)=2z,F(z)=2z3+2z2+z+1. We have a0,1=1, a1,1=2=aS,1, a2,1=1=ad,1, we can see that {a0,1=1=12as,1, c0=12, 0<c0<1,argas,1argad,1, and b0,1=b1,1= 1,b2,1=2, we can see that {b0,1=1=12b2,1,b1,1=1=12b2,1,cj=12,0<cj<1,j=0,1. and max{ρ(Aj,i):j=0,,2;i=1,2,ρ(F)}<1. Then, according to Corollary 2, every transcendental solution f of Eq. (6) satisfies ρ(f)=+ and λ¯2(f)=λ2(f)=ρ2(f)=1 with at most one exceptional solution f0 of finite order. We can see that f(z)=z+eez represents a solution of Eq. (6) that satisfies ρ(f)=+ and λ¯2(f)=λ2(f)=ρ2(f)=1.

2. Auxiliary lemmas for the proofs of the theorems

To prove our theorems, we need the following lemmas;

Lemma 1. [9] Let Pj(z)(j=0,1,,k) be polynomials with degP0=n (n1) and degPjn (j=1,,k). Let Aj(z)  (j=0,1,,k) be meromorphic functions with finite order and max{ρ(Aj): j=0,1,,k}<n such that A0(z) 0. We denote F(z)=AkePk(z)+Ak1ePk1(z)++A1eP1(z)+A0eP0(z). If deg(P0(z)Pj(z))=n for all j=1,,k, then F is a nontrivial meromorphic function with finite order that satisfies ρ(F)=n.

Lemma 2. [10] Let f be a transcendental meromorphic function, and let α>1 and ε>0 be given constants. Then there exist a set E1 (1,+) having finite logarithmic measure and a constant B>0, that depends only on α and (n,m) (n,m positive integers with n>m0) such that for all z satisfying |z|=r[0,1]E1 , we have |f(n)(z)f(m)(z)|B(T(αr,f)r(logαr)logT(αr,f))nm.

Lemma 3. [11] Let f(z)=g(z)d(z) be a meromorphic function with ρ(f)=ρ+, where g(z) and d(z) are entire functions satisfying one of the following conditions:

  1.  g being transcendental and d  being polynomial,

  2.  g, d all being transcendental and λ(d)=ρ(d)=β<ρ(g)=ρ.

For each sufficiently large |z|=r, let zr=reiθr be a point satisfying |g(zr)| =M(r,g). Then there exist a constant δr (>0), a sequence {rm}mN, rm+ and a set E2  of finite logarithmic measure such that the estimation |f(z)f(n)(z)|rm2n(n1 {is an integer}) holds for all z satisfying |z|=rm E2, rm+  and argz= θ [θrδr, θr+δr].

Lemma 4. [12] Let P(z)=(α+iβ)zn+ (α,β are real numbers,|α|+|β|0) be a polynomial with degree n1 and A(z) be a meromorphic function with ρ(A)<n. Set f(z)=A(z)eP(z), (z=reiθ), δ(P,θ)=αcosnθβsinnθ. Then for any given ε>0,  there is a set E3[1,+) having finite logarithmic measure such that for any θ[0,2π)H (H={θ[0,2π):δ(P,θ)=0})  for |z|=r[0,1]E3,r+, we have

  1.  if δ(P,θ)>0, then exp{(1ε)δ(P,θ)rn}|f(reiθ)|exp{(1+ε)δ(P,θ)rn},

  2.  if δ(P,θ)<0,  then exp{(1+ε)δ(P,θ)rn}|f(reiθ)|exp{(1ε)δ(P,θ)rn}.

Lemma 5. [13] Let φ:[0,+)R and ψ :[0,+)R be monotone nondecreasing functions such that φ(r)ψ(r) for all r(E4[0,1]),  where E4 is a set of finite logarithmic measure. Let α>1 be a given constant. Then there exists an r1=r1(α)>0 such that φ(r)ψ(αr)  for all r>r1.

Lemma 6. [11,14] Suppose that k2 and A0, A1,,Ak1,  F  (F0 or F0) are meromorphic functions such that ρ=max{ρ(Aj),ρ(F):j=0,1,k1}<. Let f be a transcendental meromorphic solution whose poles are of uniformly bounded multiplicities of the equation (7) f(k)+Ak1f(k1)++A1f+A0f=F., Then ρ2(f)ρ.

Lemma 7. [15,16]  Let  Aj(z)(0), j=0,1,,k1, F(z)0 be finite order meromorphic functions.

  1.  If f is a meromorphic solution of Eq. (7) with ρ(f)=+, then f satisfies λ¯(f)=λ(f)=ρ(f)=+.

  2.  If f is a meromorphic solution of Equation (7) with ρ(f)=+ and ρ2(f)=ρ,  then f satisfies λ¯(f)=λ(f)=ρ(f)=+, λ¯2(f)=λ2(f)=ρ2(f)=ρ.

Lemma 8. [17] Let f  be a meromorphic function of order ρ(f)=ρ<.  Then for any given ε>0, there exists a set E5(1,+) that has finite linear measure and finite logarithmic measure such that when |z|=r[0,1]E5, r+, we have |f(z)|exp(rρ+ε).

3. Proof of Theorem 3

First we prove that every transcendental meromorphic solution f of Eq. (3) is of order ρ(f)n . Assume that f is a transcendental meromorphic solution of Eq. (3) with ρ(f)<n. We can rewrite Eq. (3) in the form (8)(Ak1,1(z)ePk1(z)+Ak1,2(z)eQk1(z))f(k1)++(A0,1(z)eP0(z)+A0,2(z)eQ0(z))f=f(k).

Since max{ρ(Aj,i):j=0,1,,k1;i=1,2}<n and ρ(f)<n, then Aj,if(j) (j=0,1,,k1;i=1,2) and f(k) are meromorphic functions of finite order with ρ(Aj,if(j)) <n and ρ(f(k))<n. We have As,if(s)0(i=1,2). Indeed, if As,if(s)0, it follows that f(s)0. Then f has to be a polynomial of degree less than s. This is a contradiction. Since as,nbs,n and aj,n=cjas,n,bj,n=cjbs,n,(0<cj<1),(0<cj<1),(js),aj,n=cjas,n  or aj,n=cjad,n(0<cj<1), then aj,nbj,n, as,naj,n, bs,nbj,n and therefore deg(PsPj)=deg(QsQj)=n. Thus, by (8) and Lemma 1, we find that ρ(f(k))=n, this contradicts the fact ρ(f(k))<n. Consequently, every meromorphic solution f of Eq. (3) is transcendental with order ρ(f)n.

Assume that f is a transcendental meromorphic solution whose poles are of uniformly bounded multiplicities of Eq. (3). By Lemma 2, there exist a set E1 (1,+)  having finite logarithmic measure and a constant B>0, such that for all z satisfying |z|=r[0,1]E1, we have (9)|f(j)(z)f(z)|B(T(2r,f))k+1, j=1,2,,k, js.

By (3), it follows that the poles of f can only occur at the poles of Aj,i(z) (j=0,1,,k1;i=1,2). Note that the poles of f are of uniformly bounded multiplicities. Hence λ(1f)max{ρ(Aj,i):j=0,1,,k1;i=1,2}<n. By Hadamard factorization theorem, we know that f can be written as f(z)=g(z)d(z), where g(z) and d(z) are entire functions with λ(d)=ρ(d)=λ(1f)<nρ(f)=ρ(g). For each sufficiently large |z|=r, let zr=reiθr be a point satisfying |g(zr)| =M(r,g). By Lemma 3, there exist a constant δr (>0), a sequence {rm}mN, rm+ and a set E2 of finite logarithmic measure such that the estimation   (10)|f(z)f(s)(z)|rm2s, holds for all z satisfying |z|=rm E2, rm and argz= θ [θrδr, θr+δr].

Set z=reiθ, as,n=|as,n|eiθs, ad,n=|ad,n|eiθd, bs,n=|bs,n|eiφ, θs, θd, φ[0,2π), θsθd. Then (11) delta(Ps,θ)=|as,n|cos(nθ+θs),δ(Qs,θ)=|bs,n|cos(nθ+φ),δ(Pd,θ)=|ad,n|cos(nθ+θd). Since aj,n=cjas,n, bj,n=cjbs,n,(0<cj<1),(0<cj<1),(js) and cj,cj (j=0,1,,k1) are distinct numbers, then (12)δ(Pj,θ)=cjδ(Ps,θ) or δ(Pj,θ)=cjδ(Pd,θ), δ(Qj,θ)=cjδ(Qs,θ). Set H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0} and H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)}. For any given θ [θrδr, θr+δr](H1H2), we have δ(Ps,θ)0, δ(Pd,θ)0 and δ(Ps,θ)>δ(Pd,θ) or δ(Ps,θ)<δ(Pd,θ).

  1. δ(Ps,θ)>δ(Pd,θ). Here we also divide our proof in three subcases:  (φ=θs) or  (φ=θd) or  (φθs and φθd).

    Subcase 1.1. δ(Ps,θ)>δ(Pd,θ)>0,δ(Qs,θ)>0. If δ(Ps,θ)>0, δ(Qs,θ)>0, then we suppose δ(Ps,θ)>δ(Qs,θ) without loss of generality. Set δ3=max{δ(Pj,θ),δ(Qj,θ);js} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)}, then 0<δ3<δ(Ps,θ). Thus, by Lemma 4, for any given 0<2ε<min{δ(Ps,θ)δ3δ(Ps,θ)+δ3,δ(Ps,θ)δ(Qs,θ)δ(Ps,θ)+δ(Qs,θ)}, where c=max{cj,cj:j=0,1,,k1,js}, cs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ[θrδr, θr+δr] (13)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,1(z)ePs(z)||As,2(z)eQs(z)|exp{(1ε)δ(Ps,θ)rn}exp{(1+ε)δ(Qs,θ)rn}12exp{(1ε)δ(Ps,θ)rn}, (14)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)δ(Pj,θ)rn}+exp{(1+ε)δ(Qj,θ)rn}2exp{(1+ε)δ3rn}, j=0,1,2,,k1, js.

    From (3), we have (15)|As,1(z)ePs(z)+As,2(z)eQs(z)||ff(s)|(|f(k)f|+k1j=0,js{|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||f(j)f|}). By substituting (9),(10),(13),(14), into (15), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz= θ [θrδr, θr+δr](H1H2H3), we have 12exp{(1ε)δ(Ps,θ)rmn}2kBrm2s [T(2rm,f)]k+1exp{(1+ε)δ3rmn} which gives exp{(1ε)δ(Ps,θ)rmn}4kBrm2s [T(2rm,f)]k+1exp{(1+ε)δ3rmn}. Since 0<2ε<δ(Ps,θ)δ3δ(Ps,θ)+δ3, then we can get (16) exp{δ(Ps,θ)δ32rmn}4kBrm2s [T(2rm,f)]k+1. By Lemma 5 and (16), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. By Lemma 6 and Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.2. δ(Ps,θ)>δ(Pd,θ)>0,δ(Qs,θ)<0. We have δ(Qs,θ)<δ(Ps,θ) and δ(Qs,θ)<δ(Qj,θ)<0<δ(Ps,θ), Put d=max{cj:j=0,1,,k1,js}, ds=1. By Lemma 3, for any given ε (0<ε<12(1d1+d)), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (17)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,1(z)ePs(z)||As,2(z)eQs(z)|exp{(1ε)δ(Ps,θ)rn}exp{(1ε)δ(Qs,θ)rn}12exp{(1ε)δ(Ps,θ)rn}, (18)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)δ(Pj,θ)rn}+exp{(1ε)δ(Qj,θ)rn}2exp{(1+ε)d δ(Ps,θ)rn}, j=0,1,2,,k1,js.

    By substituting (9),(10),(17),(18), into(15), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=θ[θrδr, θr+δr] (H1H2H3), we have exp{(1ε)δ(Ps,θ)rmn}4kBrm2s [T(2rm,f)]k+1exp{(1+ε)d δ(Ps,θ)rmn}. Since 0<ε<12(1d1+d), then the last inequalities leads to (19) exp{(1d)2δ(Ps,θ)rmn}4kBrm2s [T(2rm,f)]k+1. By Lemma 5 and (19), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.3. δ(Ps,θ)>0>δ(Pd,θ),δ(Qs,θ)>0. We suppose δ(Ps,θ)>δ(Qs,θ) without loss of generality. By Lemma 3, for any given ε (0<ε<min{12(1ν1+ν),12(δ(Ps,θ)δ(Qs,θ)δ(Ps,θ)+δ(Qs,θ))}), where ν=max{cj,cj:j=0,1,,k1,js,jd}, νs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3, r+ and argz=θ[θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (20)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,1(z)ePs(z)||As,2(z)eQs(z)|exp{(1ε)δ(Ps,θ)rn}exp{(1+ε)δ(Qs,θ)rn}12exp{(1ε)δ(Ps,θ)rn}, (21)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)cjδ(Ps,θ)rn}+exp{(1+ε)cjδ(Qs,θ)rn}2exp{(1+ε)νδ(Ps,θ)rn}, j=0,1,2,,k1, js.

    By substituting (9),(10),(20) and (21),into (15), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr] (H1H2H3), we have exp{(1ε)δ(Ps,θ)rmn}4kBrm2s [T(2rm,f)]k+1exp{(1+ε)νδ(Ps,θ)rmn}. Since 0<ε<12(1ν1+ν), then (22) exp{(1ν)2δ(Ps,θ)rmn}4kBrm2s[T(2rm,f)]k+1. By Lemma 5 and (22), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, then by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.4. δ(Ps,θ)>0>δ(Pd,θ),δ(Qs,θ)<0. We have δ(Ps,θ)>δ(Pj,θ)>0,δ(Qs,θ)<δ(Qj,θ)<0, then δ(Qs,θ)< δ(Ps,θ). Put ν=max{cj:j=0,1,,k1,js,jd},νs=1. By Lemma 3, for any given 0<ε<12(1ν1+ν), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (23)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,1(z)ePs(z)||As,2(z)eQs(z)|exp{(1ε)δ(Ps,θ)rn}exp{(1ε)δ(Qs,θ)rn}12exp{(1ε)δ(Ps,θ)rn}, (24)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)δ(Pj,θ)rn}+exp{(1ε)δ(Qj,θ)rn}2exp{(1+ε)ν δ(Ps,θ)rn}, j=0,1,2,,k1,js.By substituting (9),(10),(23),(24) into(15), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz=θ[θrδr, θr+δr](H1H2H3), we have exp{(1ε)δ(Ps,θ)rmn}4kBrm2k[T(2rm,f)]k+1exp{(1+ε)ν δ(Ps,θ)rmn}. Since 0<ε<12(1ν1+ν), then (25)exp{(1ν)2δ(Ps,θ)rmn}4kBrm2k [T(2rm,f)]k+1. By Lemma 5 and (25) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.5. 0>δ(Ps,θ)>δ(Pd,θ),δ(Qs,θ)>0. We have δ(Pd,θ)<δ(Ps,θ)<0<δ(Qs,θ). Put d=max{cj:j=0,1,,k1,js}, ds=1. By Lemma 3, for any given ε (0<ε<12(1d1+d)), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (26)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,2(z)eQs(z)||As,1(z)ePs(z)|exp{(1ε)δ(Qs,θ)rn}exp{(1ε)δ(Ps,θ)rn}12exp{(1ε)δ(Qs,θ)rn}, (27)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1ε)δ(Pj,θ)rn}+exp{(1+ε)δ(Qj,θ)rn}2exp{(1+ε)dδ(Qs,θ)rn}, j=0,1,2,,k1, js. By using a similar proof as that of subcase 1.2, since 0<ε<12(1d1+d), we can obtain for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz=θ[θrδr, θr+δr] (28)exp{(1d)2δ(Qs,θ)rmn}4kBrm2s [T(2rm,f)]k+1. So, by Lemma 5 and (28) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.6. 0>δ(Ps,θ)>δ(Pd,θ),δ(Qs,θ)<0. Set λ=min{cj,cj:j=0,1,,k1,js,jd}. By Lemma 3, for any given 0<ε<1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (29) |Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1ε)δ(Pj,θ)rn}+exp{(1ε)δ(Qj,θ)rn}2exp{(1ε)λδ(Ps,θ)rn}, j=0,1,2,,k1. From (3), we have (30) 1|ff(k)|k1j=0{|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||f(j)f|}. By substituting (9),(10),(29) into (30), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=θ [θrδr, θr+δr](H1H2H3), we have 12kBrm2k[T(2rm,f)]k+1exp{(1ε)λδ(Ps,θ)rmn} which gives (31)exp{(ε1)λδ(Ps,θ)rmn}2kBrm2k[T(2rm,f)]k+1. Since 0<ε<1 and λδ(Ps,θ)<0, then by Lemma 2.5 and (31), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 2.1. δ(Ps,θ)>δ(Pd,θ)>0. Because of as,nbs,n, we suppose |as,n|<|bs,n| without loss of generality. In this case, by (3.4) and (3.5) we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Put c=max{cj,cj:j=0,1,,k1,js,jd},cs=1. Then, 0<c<1. By Lemma 3, for any given ε with 0<ε<min{12(1c1+c),12(δ(Qs,θ)δ(Ps,θ)δ(Qs,θ)+δ(Ps,θ))}, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz= θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (32)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,2(z)eQs(z)||As,1(z)ePs(z)|exp{(1ε)δ(Qs,θ)rn}exp{(1+ε)δ(Ps,θ)rn}12exp{(1ε)δ(Qs,θ)rn}, (33)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)cδ(Ps,θ)rn}+exp{(1+ε)cδ(Qs,θ)rn}2exp{(1+ε)cδ(Qs,θ)rn}, j=0,1,2,,k1, js. By substituting (9),(10),(32),(33) into (15), since 0<ε<12(1c1+c), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr](H1H2H3), we obtain (34)exp{(1c)2δ(Qs,θ)rmn}4kBrm2s [T(2rm,f)]k+1. Thus, by Lemma 5 and (34) we get ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 2.2. δ(Ps,θ)>0>δ(Pd,θ). Because of as,nbs,n, we suppose |as,n|<|bs,n| without loss of generality. In this case, by (11)and(12) we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Put c=max{cj,cj:j=0,1,,k1,js,jd},cs=1. Using the same reasoning as in Subcase 2.1 , we can also obtain ρ(f)=+ and ρ2(f)=n.

    Subcase 2.3. 0>δ(Ps,θ)>δ(Pd,θ). We have δ(Qs,θ)<δ(Qj,θ)<δ(Ps,θ)<0, δ(Ps,θ)<δ(Pj,θ)<0. Put λ=min{cj,cj:j=0,1,,k1,js,jd}. By Lemma 3, for any given 0<ε<1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (35)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)|;|Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1ε)δ(Pj,θ)rn}+exp{(1ε)δ(Qj,θ)rn}2exp{(1ε)λδ(Ps,θ)rn}, j=0,1,2,,k1. By substituting (9),(10)and(35)into(30), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=θ[θrδr, θr+δr](H1H2H3), we have 12kBrm2k[T(2rm,f)]k+1exp{(1ε)λδ(Ps,θ)rmn} which gives (36)exp{(ε1)λδ(Ps,θ)rmn}2kBrm2k[T(2rm,f)]k+1. Since 0<ε<1 and λδ(Ps,θ)<0, then by Lemma 5 and (36) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 3.1. δ(Ps,θ)>δ(Pd,θ)>0. Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11) and (12), we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Then, 0<c<1. By Lemma 3, for any given ε with 0<ε<min12{(1c1+c),12(δ(Qs,θ)δ(Ps,θ)δ(Qs,θ)+δ(Ps,θ))}, where c=max{cj,cj:j=0,1,,k1,js,jd},cs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz= θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (32)and(33) hold. By substituting (9),(10),(32),(33)into(15), we obtain (34) for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr](H1H2H3). Since 0<ε<12(1c1+c), then by Lemma 5 and (34) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 3.2. δ(Ps,θ)>0>δ(Pd,θ). Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11)and(12), we have δ(Qs,θ)<δ(Qj,θ)<0, δ(Qs,θ)<δ(Pd,θ)<0<δ(Ps,θ). Then, 0<c<1. By Lemma 3, for any given ε (0<ε<12(1c1+c)), where c=max{cj,cj:j=0,1,,k1,js,jd},cs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz= θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (37)|As,1(z)ePs(z)+As,2(z)eQs(z)||As,2(z)ePs(z)||As,1(z)eQs(z)|exp{(1ε)δ(Ps,θ)rn}exp{(1ε)δ(Qs,θ)rn}12exp{(1ε)δ(Ps,θ)rn}, (38)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1+ε)cδ(Ps,θ)rn}+exp{(1ε)cδ(Qs,θ)rn}2exp{(1+ε)cδ(Ps,θ)rn}, j=0,1,2,,k1, js.

    By substituting (9),(10),(37)and(38)into(15), by 0<ε<12(1c1+c), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr](H1H2H3), we obtain (39)exp{(1c)2δ(Ps,θ)rmn}4kBrm2s [T(2rm,f)]k+1. Therefore, by Lemma 5 and (39) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from Equation 3, we have ρ2(f)n, so ρ2(f)=n.

    Subcase 3.3. 0>δ(Ps,θ)>δ(Pd,θ). Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11)and(12), we have δ(Qs,θ)<δ(Qj,θ)<0, δ(Qs,θ)<δ(Pd,θ)<δ(Ps,θ)<0. Put c=min{cj,cj:j=0,1,,k1}. By Lemma 3, for any given 0<ε<1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (40)|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||Aj,1(z)ePj(z)|+|Aj,2(z)eQj(z)|exp{(1ε)δ(Pj,θ)rn}+exp{(1ε)δ(Qj,θ)rn}2exp{(1ε)cδ(Ps,θ)rn}, j=0,1,2,,k1.

    By substituting (9),(10)and(40)into(30), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz=θ[θrδr, θr+δr](H1H2H3), we obtain 12kBrm2k[T(2rm,f)]k+1exp{(1ε)cδ(Ps,θ)rmn} which gives (41)exp{(ε1)cδ(Ps,θ)rmn}2krm2kB[T(2rm,f)]k+1. By Lemma 5 and (41) we obtain ρ(f)=lim suprm+log+T(rm,f)logrm=+ and ρ2(f)=lim suprm+log2+T(rm,f)logrmn. In addition, by Lemma 6 and from Eq. (3), we have ρ2(f)n, so ρ2(f)=n.

  2. δ(Ps,θ)<δ(Pd,θ). Here we also divide our proof in three subcases:  (φ=θs) or  (φ=θd) or  (φθs and φθd ). Using the same reasoning as in I, we can also obtain ρ2(f)n, so ρ2(f)=n.

4. Proof of Theorem 4

First, we show that (4) can possess at most one exceptional transcendental meromorphic solution f0 of finite order. In fact, if f is another transcendental meromorphic solution of finite order of Eq. (4), then f0f is of finite order. But f0f is a transcendental meromorphic solution of the corresponding homogeneous equation of (4). This contradicts Theorem 3. We assume that f is an infinite order meromorphic solution of (4) whose poles are of uniformly bounded multiplicities. By Lemma 7 and Lemma 8 , we have λ¯(f)=λ(f)=ρ(f)=+ and λ¯2(f)=λ2(f)=ρ2(f)n.

Now, we prove that ρ2(f)n. By Lemma 3, there exists a set E1 (1,+) having finite logarithmic measure and a constant B>0 such that for all z satisfying |z|=r[0,1][E1, we have (9). Set ρ1=max{ρ(F),ρ(Aj,i(z)):j=0,1,,k1;i=1,2}. By (4), it follows that the poles of f can only occur at the poles of F and Aj,i(z),j=0,1,,k1;i=1,2. Note that the poles of f are of uniformly bounded multiplicities. Hence λ(1f)max{ρ(Aj,i(z)):j=0,1,,k1;i=1,2}=ρ1. By Hadamard factorization theorem, we know that f can be written as f(z)=g(z)d(z), where g(z) and d(z) are entire functions with λ(d)=ρ(d)=λ(1f)ρ1<ρ(f)=ρ(g)=+. For each sufficiently large |z|=r, let zr=reiθr be a point satisfying |g(zr)| =M(r,g). By Lemma 3, there exist a constant δr (>0), a sequence {rm}mN, rm+ and a set E2 of finite logarithmic measure such that the estimation (10) holds for all z satisfying |z|=rm E2, rm+ and argz= θ [θrδr, θr+δr]. Since |g(z)| is continuous in |z|=r, then there exists a constant r(>0) such that for all z satisfying |z|=r sufficiently large and argz= θ [θrδr, θr+δr], we have (42)12|g(zr)|<|g(z)|<32|g(zr)|. On the other hand, by Lemma 8, for a given ε(0<ε<nρ1), there exists a set E5(1,+) that has finite linear measure and finite logarithmic measure such that when |z|=r[0,1]E5, r+, we have (43)|F(z)|exp{rρ1+ε},|d(z)|exp{rρ1+ε}. Since |g(z)|=M(r,g)1, from (43), we obtain (44)|F(z)f(z)|=|d(z)F(z)g(z)||d(z)F(z)|M(r,g)exp{rρ1+ε}exp{rρ1+ε}=exp{2rρ1+ε}, for |z|=r[0,1]E5, r+. Set ν=min{δr,λr}. Suppose that Aj,i(z), Pj(z), Qj(z) aj,n,bj,n,(j=0,1,,k1;i=1,2)  satisfy the hypotheses of Theorem 3. Set z=reiθ, as,n=|as,n|eiθs, ad,n=|ad,n|eiθd, bs,n=|bs,n|eiφ, θs, θd, φ[0,2π), θsθd,. For any given θ [θrδr, θr+δr] (H1H2), we have δ(Ps,θ)0,δ(Pd,θ)0 and δ(Ps,θ)>δ(Pd,θ) or δ(Ps,θ)<δ(Pd,θ).

  1. δ(Ps,θ)>δ(Pd,θ). Here we also divide our proof in three cases:  (φ=θs) or  (φ=θd) or  (φθs and φθd).

    Case 1. δ(Ps,θ)>δ(Pd,θ) and (φθs and φθd).

    Subcase 1.1. δ(Ps,θ)>δ(Pd,θ)>0,δ(Qs,θ)>0. If δ(Ps,θ)>0, δ(Qs,θ)>0, then we suppose δ(Ps,θ)>δ(Qs,θ) without loss of generality. Set δ3=max{δ(Pj,θ),δ(Qj,θ);js} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)}, then 0<δ3<δ(Ps,θ). Thus by Lemma 3, for any given ε with 0<ε<min{12(δ(Ps,θ)δ3δ(Ps,θ)+δ3),12(δ(Ps,θ)δ(Qs,θ)δ(Ps,θ)+δ(Qs,θ)),nρ1}, where c=max{cj,cj:j=0,1,,k1,js},cs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ[θrδr, θr+δr](H1H2H3), we have (13)and(14) hold. From (4), we can write (45)|As,1(z)ePs(z)+As,2(z)eQs(z)||ff(s)|(|F(z)f|+|f(k)f|+k1j=0,js{|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||f(j)f|}). By substituting (9),(10),(13),(14)and(44)into(45), for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and   argz=θ[θrδr, θr+δr](H1H2H3), we have (46)exp{(1ε)δ(Ps,θ)rmn}4(k+1)Brm2s[T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)δ3rmn}. Since 0<ε<min{12(δ(Ps,θ)δ3δ(Ps,θ)+δ3),nρ1}, then by Lemma 5 and (46), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. By Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.2. δ(Ps,θ)>δ(Pd,θ)>0,δ(Qs,θ)<0. We have δ(Qs,θ)<δ(Ps,θ) and δ(Qs,θ)<δ(Qj,θ)<0<δ(Ps,θ). Put d=max{cj:j=0,1,,k1,js},ds=1 By Lemma 3, for any given ε (0<ε<min{12(1d1+d),nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (17)and(18) hold. By substituting (9),(10),(17),(18)and(44)into(45) for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz= θ [θrδr, θr+δr](H1H2H3), we get (47)exp{(1ε)δ(Ps,θ)rmn}4(k+1)Brm2k [T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)d δ(Ps,θ)rmn}. Since 0<ε<min{12(1d1+d),nρ1}, by Lemma 5 and (47), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.3. δ(Ps,θ)>0>δ(Pd,θ),δ(Qs,θ)>0. We suppose δ(Ps,θ)>δ(Qs,θ) without loss of generality. By Lemma 3, for any given ε (0<ε<min{12(1ν1+ν),12(δ(Ps,θ)δ(Qs,θ)δ(Ps,θ)+δ(Qs,θ)),nρ1}), where ν=max{cj,cj:j=0,1,,k1,js,jd},νs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3, r+ and argz=θ[θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (20)and(21) hold. By substituting (9),(10),(20),(21)and(44)into(45) for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz= θ [θrδr, θr+δr](H1H2H3), we get (48)exp{(1ε)δ(Ps,θ)rmn}4(k+1)Brm2s [T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)νδ(Ps,θ)rmn}. Since 0<ε<min{12(1ν1+ν),nρ1}, then by Lemma 5 and (48), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, then by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.4. δ(Ps,θ)>0>δ(Pd,θ),δ(Qs,θ)<0. We have δ(Ps,θ)>δ(Pj,θ)>0,δ(Qs,θ)<δ(Qj,θ)<0, then δ(Qs,θ)< δ(Ps,θ). Put ν=max{cj:j=0,1,,k1,js,jd},νs=1. By Lemma 3, for any given ε (0<ε<min{12(1ν1+ν),nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (3.16) and (3.17) hold. By substituting (9),(10),(23),(24)and(44)into(45) for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz= θ [θrδr, θr+δr](H1H2H3), we have (49)exp{(1ε)δ(Ps,θ)rmn}4(k+1)Brm2k [T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)ν δ(Ps,θ)rmn}. Since 0<ε<min{12(1ν1+ν),nρ1}, then by Lemma 5 and (49), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.5. 0>δ(Ps,θ)>δ(Pd,θ),δ(Qs,θ)>0. We have δ(Pd,θ)<δ(Ps,θ)<0<δ(Qs,θ). Put d=max{cj:j=0,1,,k1,js},ds=1. By Lemma 3, for any given ε (0<ε<min{12(1d1+d),nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3, r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (26) and (27) hold. Using a similar proof as that of Subcase 1.5 of Theorem 3 by (45), we can obtain for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr] (50)exp{(1ε)δ(Qs,θ)rmn}4(k+1)Brm2s [T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)dδ(Qs,θ)rmn}. Since 0<ε<min{12(1d1+d),nρ1}, by Lemma 5 and (50), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 1.6. 0>δ(Ps,θ)>δ(Pd,θ),δ(Qs,θ)<0. Set (51)λ=min{cj,cj:j=0,1,,k1,js,jd}. By Lemma 3, for any given ε (0<ε<min{12,nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3, r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0,δ(Qs,θ)=δ(Ps,θ)} are finite sets, we have (29) holds. From (4), we have (51)1|ff(k)|(|F(z)f(z)|+k1j=0{|Aj,1(z)ePj(z)+Aj,2(z)eQj(z)||f(j)f|}). By substituting (9),(10),(29)and(44)into(51) for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=θ [θrδr, θr+δr](H1H2H3), we get 12(k+1)Brm2k[T(2rm,f)]k+1exp{2rρ1+ε}exp{(1ε)λδ(Ps,θ)rmn} which gives (52)exp{(ε1)λδ(Ps,θ)rmn}2(k+1)Brm2kexp{2rρ1+ε}[T(2rm,f)]k+1. Since 0<ε<min{12,nρ1} and δ(Ps,θ)<0, by Lemma 5 and (52), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Case 2. δ(Ps,θ)>δ(Pd,θ) and φ=θs. 

    Subcase 2.1. δ(Ps,θ)>δ(Pd,θ)>0. Because of as,nbs,n, we suppose |as,n|<|bs,n| without loss of generality. In this case, by (11) and (12), we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Put c=max{cj,cj:j=0,1,,k1,js,jd},cs=1. Then, 0<c<1. By Lemma 3, for any given ε with 0<ε<min{12(1c1+c),12(δ(Qs,θ)δ(Ps,θ)δ(Qs,θ)+δ(Ps,θ)),nρ1}, there is a set E3[1,+) having finite logar measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz= θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (32) and (33) hold. By substituting (9),(10),(32),(33)and(44)into(45), we obtain for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz= θ [θrδr, θr+δr](H1H2H3) (53)exp{(1ε)δ(Qs,θ)rmn}4(k+1)Brm2s [T(2rm,f)]k+1exp{2rρ1+ε}exp{(1+ε)cδ(Qs,θ)rmn}. Since 0<ε<min{12(1c1+c),nρ1}, then by Lemma 5 and (53) we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 2.2. δ(Ps,θ)>0>δ(Pd,θ). Because of as,nbs,n, we suppose |as,n|<|bs,n| without loss of generality. In this case, by (11) and (12) we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Put c=max{cj,cj:j=0,1,,k1,js,jd},cs=1. Using the same reasoning as in Subcase 2.1 , we can also obtain ρ(f)=+ and ρ2(f)=n.

    Subcase 2.3. 0>δ(Ps,θ)>δ(Pd,θ). We have δ(Qs,θ)<δ(Qj,θ)<δ(Ps,θ)<0, δ(Ps,θ)<δ(Pj,θ)<0. Put λ=min{cj,cj:j=0,1,,k1,js,jd}. By Lemma 3, for any given ε (0<ε<min{12,nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (35) holds. By (9),(10),(35),(44)and(51), we have for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=θ [θrδr, θr+δr](H1H2H3), 12(k+1)Brm2k[T(2rm,f)]k+1exp(2rρ1+ε)exp{(1ε)λδ(Ps,θ)rmn}, which gives (54)exp{(ε1)λδ(Ps,θ)rmn}2(k+1)Bkrm2kexp(2rρ1+ε)[T(2rm,f)]k+1. Since 0<ε<nρ1 and δ(Ps,θ)<0, then by Lemma 5 and (54), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Case 3. δ(Ps,θ)>δ(Pd,θ) and φ=θd. 

    Subcase 3.1. δ(Ps,θ)>δ(Pd,θ)>0. Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11)and(12), we have δ(Qs,θ)>δ(Qj,θ)>0, δ(Qs,θ)>δ(Ps,θ)>δ(Pj,θ)>0. Then, 0<c<1. By Lemma 3, for any given ε with 0<ε<min{12(1c1+c),12(δ(Qs,θ)δ(Ps,θ)δ(Qs,θ)+δ(Ps,θ)),nρ1}, where c=max{cj,cj:j=0,1,,k1,js,jd},cs=1, there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3, r+ and argz=θ[θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (32)and(33) hold. By substituting (9),(10),(32),(33)and(44)into(45), we obtain (55)exp{(1ε)δ(Qs,θ)rmn}4(k+1)Brm2s[T(2rm,f)]k+1exp(2rρ1+ε)exp{(1+ε)cδ(Qs,θ)rmn} for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz=  θ [θrδr, θr+δr](H1H2H3). Since 0<ε<min{12(1c1+c),nρ1}, then by Lemma 5 and (55), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 3.2. δ(Ps,θ)>0>δ(Pd,θ). Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11)and(12), we have δ(Qs,θ)<δ(Qj,θ)<0, δ(Qs,θ)<δ(Pd,θ)<0<δ(Ps,θ). Then, 0<c<1. By Lemma 3, for any given ε (0<ε<min{12(1c1+c),nρ1}), where c=max{cj,cj:j=0,1,,k1,js,jd},cs=1, there is a set E3[1,+) having finite logar measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz= θ [θrδr, θr+δr] H1H2H3, where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (37)and(38) hold. Substituting (9),(10),(37),(38)and(44)into(45), we obtain for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and  argz= θ [θrδr, θr+δr](H1H2H3) (56)exp{(1ε)δ(Ps,θ)rmn}4(k+1)Brm2s[T(2rm,f)]k+1exp{(1+ε)cδ(Ps,θ)rmn}. Since 0<ε<min{12(1c1+c),nρ1}, then by Lemma 5 and (56), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

    Subcase 3.3. 0>δ(Ps,θ)>δ(Pd,θ). Because of ad,nbs,n, we suppose |ad,n|<|bs,n| without loss of generality. In this case, by (11) and (12), we have δ(Qs,θ)<δ(Qj,θ)<0, δ(Qs,θ)<δ(Pd,θ)<δ(Ps,θ)<0. Put c=min{cj,cj:j=0,1,,k1}. By Lemma 3, for any given ε (0<ε<min{12,nρ1}), there is a set E3[1,+) having finite logarithmic measure such that for all z satisfying |z|=r[0,1]E3,r+ and argz=θ [θrδr, θr+δr](H1H2H3), where H1={θ[0,2π):δ(Ps,θ)=0,δ(Pd,θ)=0}, H2={θ[0,2π):δ(Ps,θ)=δ(Pd,θ)} and H3={θ[0,2π):δ(Qs,θ)=0} are finite sets, we have (40) holds. By substituting (9),(10),(40)and(44)into(51), we obtain for all z satisfying |z|=rm[0,1](E1E2E3), rm+ and argz= θ [θrδr, θr+δr](H1H2H3) 12(k+1)Brm2k[T(2rm,f)]k+1exp(2rρ1+ε)exp{(1ε)cδ(Ps,θ)rmn} which gives (57)exp{(ε1)cδ(Ps,θ)rmn}2(k+1)Brm2k[T(2rm,f)]k+1exp(2rρ1+ε). By Lemma 5 and (57), we obtain ρ(f)=lim suprm+logT(rm,f)logrm=+ and ρ2(f)=lim suprm+loglogT(rm,f)logrmn. In addition, by Lemma 6 and from (4), we have ρ2(f)n, so ρ2(f)=n.

  2. δ(Ps,θ)<δ(Pd,θ). Here we also divide our proof in three subcases:  (φ=θs) or  (φ=θd) or  (φθs and φθd ). Using the same reasoning as in I, we can obtain ρ2(f)n, so ρ2(f)=n.

References

  1. Hamani, K., & Belaı̈di, B. (2017). On the hyper-order of transcendental meromorphic solutions of certain higher order linear differential equations. Opuscula Mathematica, 37(6), 853-874.

  2. Hayman, W. K. (1964). Meromorphic functions (Vol. 78). Oxford Mathematical Monographs, Clarendon Press, Oxford.

  3. Kwon, K. H. (1996). On the growth of entire functions satisfying second order linear differential equations. Bulletin of the Korean Mathematical Society, 33(3), 487-496.

  4. Laine, I. (1993). Nevanlinna Theory and Complex Differential Equations. Walter de Gruyter & Co., Berlin.

  5. Yang, C. C., & Yi, H. X. (2004). Uniqueness theory of meromorphic functions (Vol. 557). Springer Science & Business Media.

  6. Belaı̈di, B., & Abbas, S. (2008). On the hyper order of solutions of a class of higher order linear differential equations. Analele Ştiinţifice ale Universităţii “Ovidius” Constanţa. Seria: Matematică, 16(2), 15-30.

  7. Belaı̈di, B. (2007). On the meromorphic solutions of linear differential equations. Journal of Systems Science and Complexity, 20, 41-46.

  8. Tu, J., & Yi, C. F. (2008). On the growth of solutions of a class of higher order linear differential equations with coefficients having the same order. Journal of Mathematical Analysis and Applications, 340(1), 487-497.

  9. Andasmas, M., & Belaı̈di, B. (2013). On the order and hyper-order of meromorphic solutions of higher order linear differential equations. Hokkaido Mathematical Journal, 42(3), 357-383.

  10. Gundersen, G. G. (1988). Estimates for the logarithmic derivative of a meromorphic function, plus similar estimates. Journal of the London Mathematical Society, 2(1), 88-104.

  11. Habib, H., & Belaı̈di, B. (2011). On the growth of solutions of some higher order linear differential equations with entire coefficients. Electronic Journal of Qualitative Theory of Differential Equations, 2011(93), 115–134.

  12. Hamani, K., & Belaidi, B. (2013). On the hyper-order of solutions of a class of higher order linear differential equations. Bulletin of the Belgian Mathematical Society-Simon Stevin, 20(1), 27-39.

  13. Gundersen, G. G. (1988). Finite order solutions of second order linear differential equations. Transactions of the American Mathematical Society, 305(1), 415-429.

  14. Chen, W., & Xu, J. (2009). Growth of meromorphic solutions of higher-order linear differential equations. Electronic Journal of Qualitative Theory of Differential Equations, 2009(1), 1-13.

  15. Belaı̈di, B. (2008). Growth and oscillation theory of solutions of some linear differential equations. Matematicki vesnik, 60(234), 233-246.

  16. Zong-xuan, C. (1994). Zeros of meromorphic solutions of higher order linear differential equations. Analysis, 14(4), 425-438.

  17. Chen, Z. X. (1996). The zero, pole and order of meromorphic solutions of differential equations with meromorphic coefficients. Kodai Mathematical Journal, 19(3), 341-354.