Volume 2011, Article ID 928194,25pages doi:10.1155/2011/928194
Research Article
Possible Intervals for T - and M-Orders of Solutions of Linear Differential Equations in the Unit Disc
Martin Chuaqui,
1Janne Gr ¨ohn,
2Janne Heittokangas,
2and Jouni R ¨atty ¨a
21Departamento de Matem´aticas, Pontificia Universidad Cat´olica de Chile, Casilla 306, Correo 22 Santiago, 6904411 Macual Santiago, Chile
2Department of Physics and Mathematics, University of Eastern Finland, P.O. Box 111, 80101 Joensuu, Finland
Correspondence should be addressed to Janne Heittokangas,janne.heittokangas@uef.fi Received 19 May 2011; Accepted 5 July 2011
Academic Editor: Jean Michel Combes
Copyrightq2011 Martin Chuaqui et al. 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.
In the case of the complex plane, it is known that there exists a finite set of rational numbers containing all possible growth orders of solutions offkak−1zfk−1· · ·a1zfa0zf 0 with polynomial coefficients. In the present paper, it is shown by an example that a unit disc counterpart of such finite set does not contain all possibleT- andM-orders of solutions, with respect to Nevanlinna characteristic and maximum modulus, if the coefficients are analytic functions belonging either to weighted Bergman spaces or to weighted Hardy spaces. In contrast to a finite set, possible intervals forT- andM-orders are introduced to give detailed information about the growth of solutions. Finally, these findings yield sharp lower bounds for the sums ofT- andM-orders of functions in the solution bases.
1. Introduction
This research is a continuation of recent activity in the field of complex differential equations.
In particular, the present paper concerns linear differential equations of the type
fkak−1zfk−1· · ·a1zfa0zf0, 1.1 where the coefficientsa0z, . . . , ak−1zare analytic functions in the unit discD: {z:|z|<
1}of the complex planeC. A variety of publications in the existing literature illustrate that the connection between the growth of coefficient functions and the growth of solutions is relatively well understood. On the one hand, the growth estimates in1 have been proven
to be instrumental tools in estimating the growth of solutions when the growth of coefficients is known. On the other hand, proofs of the converse direction have taken advantage of the method of order reduction as well as different types of logarithmic derivative estimates.
For an analytic function inD, it is known thatT- andM-orders of growth, with respect to Nevanlinna characteristic and maximum modulus, are not equal in general. This is in contrast to the corresponding case in C. Hence, there are two distinct cases in Dto work with. First, if the growth of solutions is measured by using the T-order, then it is natural to express the other growth aspects by means of integration as well. In particular, it is reasonable to consider coefficient functions belonging to some weighted Bergman spaces and use integrated estimates for logarithmic derivatives2 . Second, if the growth of solutions is measured by using theM-order, then it is natural to express the other growth aspects by means of the maximum modulus function. In particular, it is sensible to restrict the growth of the maximum modulus of coefficient functions, which leads to weighted Hardy spaces, and work with estimates for the maximum modulus of logarithmic derivatives involving exceptional sets3 .
The main focus of this paper is in improving the lower bounds for the growth of solutions of1.1given in 2,3 and explore some consequences, which are motivated by the following observations.
By the classical results inCmaking use of Newton-Puiseux diagram, there is a finite set containing the possible growth orders of solutions of1.1assuming that coefficients are polynomials. In particular, Gundersen-Steinbart-Wang showed that this finite set consists of rational numbers obtained from simple arithmetic with the degrees of the polynomial coefficients in1.1 4, Theorem 1 . Their proof relies on classical Wiman-Valiron theory inC.
Even though a recent unit disc counterpart of Wiman-Valiron theory5 has been successfully applied to differential equations, the possible orders of solutions of 1.1 in D have been obtained only by assuming that coefficients areα-polynomial regular. Theseα-polynomial regular functions have similar growth properties than polynomials in the sense that maximal growth is attained in every direction. However, they appear to be only a relatively small subset of the Korenblum space, which characterizes finite-order solutions of 1.1 inD 6, Theorem 6.1 . Note that in the case ofC, all solutions of1.1are of finite order if and only if coefficients are polynomials7, Satz 1 .
In the present paper, it is shown by an example that a unit disc counterpart of the finite set constructed by Gundersen-Steinbart-Wang does not contain all possible orders of solutions of1.1, provided that the coefficients belong either to weighted Bergman spaces or to weighted Hardy spaces. In contrast to a finite set, we introduce possible intervals for T-orders andM-orders, giving detailed information about the growth of solutions. Finally, these findings are applied to estimate the sums ofT- andM-orders of functions in the solution bases of1.1from below.
2. Results and Motivation
The results concerning T- and M-orders of solutions of 1.1 are given, respectively, in Sections2.1-2.2and2.3-2.4. Due to the similarities of the assertions, we omit the proofs of results regardingM-orders of solutions of1.1, excluding the sketched proof ofTheorem 2.5 inSection 7.
LetMDandHDdenote the sets of all meromorphic and analytic functions inD.
For simplicity, we writeα :max{α,0}for anyα∈R,|fz||gz|if there exists a constant
C > 0 independent ofzsuch that |fz| ≤ C|gz|, andfz ∼ gzif there exist constants C1>0 andC2>0 independent ofzsuch thatC1|gz| ≤ |fz| ≤C2|gz|.
2.1. Growth of Solutions with Respect to Nevanlinna Characteristic TheT-order of growth off∈ MDis defined as
σT
f
:lim sup
r→1−
logT r, f
−log1−r, 2.1
whereTr, fis the Nevanlinna characteristic off. Forp >0 andα >−1, the weighted Bergman spaceApαconsists of thosef ∈ HDfor which
f
Apα :
D
fzp
1− |z|2 αdmz 1/p
<∞. 2.2
Functions of maximal growth in
q<α<∞Apα are distinguished by denotingf ∈ Apq, if q inf{α >−1 :f∈Apα}.
If the growth of the coefficients is expressed by means of integration, then it is natural to consider the growth of solutions of1.1with respect toT-order.
Theorem A see 2, Theorems 1 and 2 . Suppose that aj ∈ A1/k−jαj , where αj ≥ 0 for j 0, . . . , k−1, and denoteαk:0.
iLet 0 ≤ α < ∞. Then all solutions f of 1.1 satisfy σTf ≤ α if and only if maxj0,...,k−1{αj} ≤α.
iiAll nontrivial solutionsfof1.1satisfy
j1,...,kmin k
α0−αj
j αj
≤σT f
≤ max
j0,...,k−1
αj
. 2.3
iiiIf q ∈ {0, . . . , k −1}is the smallest index for which αq maxj0,...,k−1{αj}, then each solution base of 1.1 contains at leastk−qlinearly independent solutionsf such that σTf αq.
The assumptionaj∈A1/k−jαj in Theorem Aicannot be replaced byaj∈A1/k−jαj ; see 8 . We refine Theorem A and then further underscore its consequences.
Theorem 2.1. Suppose thataj∈A1/k−jαj , whereαj≥ −1 forj0, . . . , k−1, and letq∈ {0, . . . , k−1}
be the smallest index for whichαq maxj0,...,k−1{αj}. Ifs ∈ {0, . . . , q}, then each solution base of 1.1contains at leastk−slinearly independent solutionsfsuch that
js1,...,kmin
k−s
αs−αj
j−s αj
≤σT
f
≤αq, 2.4
whereαk:−1.
The case s 0 clearly reduces to Theorem Aii. If s q, then the conditionαq maxj0,...,k−1{αj}implies that
jq1,...,kmin
k−q αq−αj
j−q αj
min
jq1,...,k
k−j αq−αj
j−q αq
αq, 2.5
where the minimum is attained forj k. Hence the assertion ofTheorem 2.1fors qis contained in Theorem Aiii. Our contribution is to extend the first inequality in2.4 for s∈ {1, . . . , q−1}.Theorem 2.1is proved inSection 4, and the sharpness and the special cases k2 andk3 are further discussed inSection 3.1.
Letq ∈ {0, . . . , k−1}be the smallest index for whichαq maxj0,...,k−1{αj}. Ifαq ≤0, then all solutions in each solution base of1.1are of zeroT-order by Theorem Aii. Suppose thatαq >0. In order to state the following corollaries ofTheorem 2.1, we denote
βTs: min
js1,...,k
k−s
αs−αj
j−s αj
, s0, . . . , q, 2.6
whereαk:−1. Moreover, we define s:min
s∈
0, . . . , q
:βTs>0
. 2.7
Remark thatβTq>0, since2.5impliesαq βTq.
Corollary 2.2. Suppose thataj∈A1/k−jαj , whereαj≥ −1 forj0, . . . , k−1, and letq∈ {0, . . . , k− 1}be the smallest index for whichαqmaxj0,...,k−1{αj}>0. Then each solution base of1.1admits at mosts≤qsolutionsfsatisfyingσTf< βTs. In particular, there are at mosts≤qsolutions fsatisfyingσTf 0.
To estimate the quantityk
j1σTfjby usingTheorem 2.1, we set γT
j
:max
βT0, . . . , βT
j
, j0, . . . , q. 2.8
EvidentlyγTj>0 forj∈ {s, . . . , q}, andγTj≤0 forj∈ {0, . . . , s−1}.
Corollary 2.3. Suppose thataj∈A1/k−jαj , whereαj≥ −1 forj0, . . . , k−1, and letq∈ {0, . . . , k− 1}be the smallest index for whichαq maxj0,...,k−1{αj} >0. Let{f1, . . . , fk}be a solution base of 1.1. Ifq0, thenk
j1σTfj kα0, while ifq≥1, then k−q
αqq−1
js
γT
j
≤k
j1
σT
fj
≤kαq. 2.9
Note that the sum in2.9is considered to be empty, if sq.
2.2. Gundersen-Steinbart-Wang Method forT-Order
We proceed to give an alternative statement ofTheorem 2.1and its corollaries by modifying the key steps in4 . This yields a natural way to define possible intervals forT-orders of solutions of1.1. As a consequence, we get a useful estimate following fromCorollary 2.3.
Setδj : αj1k−jfor allj0, . . . , k−1. Lets1∈ {0, . . . , k−1}be the smallest index satisfyingαs1 maxj0,...,k−1{αj}>0, which is equivalent to
δs1
k−s1 max
j0,...,k−1
δj k−j
>1. 2.10
If s1 cannot be found, then all solutions of 1.1 are of zero T-order by Theorem Aii.
Otherwise, for a givensm,m∈N, letsm1∈ {0, . . . , sm−1}be the smallest index satisfying δsm1−δsm
sm−sm1 max
j0,...,sm−1
δj−δsm
sm−j
>1. 2.11
Eventually this process will stop, yielding a finite list of indicess1, . . . , spsuch thatp≤kand s1> s2>· · ·> sp≥0. Further, set
BTt: δst−δst−1
st−1−st −1, t1, . . . , p, 2.12 wheres0:kandδk:0. Due to resemblance between2.12and4, Equation2.4 , it seems plausible that the possible nonzeroT-orders of solutions of1.1in the unit disc case could be found among the numbersBTt, wheret1, . . . , p. However,Example 3.1shows that this is not the case.
The following lemma allows us to view the results inSection 2.1in a new perspective.
In particular,Lemma 2.4emphasizes the connection betweenBTandγT. Lemma 2.4. One has the following:
iBT1>BT2>· · ·>BTp>0;
iiβTst BTtfor allt∈ {1, . . . , p};
iiiγTq BT1,γTj BTtfor allst≤j < st−1andt∈ {2, . . . , p}, andγTj≤0 for all j < sp. In particular,sps.
By relying onLemma 2.4,Theorem 2.1, andCorollary 2.2, we proceed to state possible intervals for T-orders of functions in solution bases of 1.1in the caseaj ∈ A1/k−jαj , where αj≥ −1 forj0, . . . , k−1. In fact, each solution base of1.1contains the following:
iat leastk−s1solutionsfsatisfyingσTf BT1;
iiat leastk−stsolutionsfsatisfyingσTf∈BTt,BT1 fort2, . . . , p;
iiiat mostspsolutionsfsatisfyingσTf∈0,BTsp.
For the following application, let{f1, . . . , fk}be a solution base of1.1. Knowing the possible intervals forT-orders, we get
k j1
σT
fj
≥k−s1BT1 · · ·
sp−1−sp
BT
p
sp·0δspsp−k. 2.13
In view ofLemma 2.4, the lower estimates in2.9and2.13are equal.
Finally, we point out a useful consequence of2.13. Ifsp 0, thenδsp sp δ0. If sp >0, thenδ0−δsp/sp ≤1 by2.11, andδsp sp ≥δ0. Hence, in both cases we can state that
k j1
σT fj
≥δspsp−k≥δ0−k≥α0k, 2.14
where the equalities hold, ifα0maxj0,...,k−1{αj}>0.
2.3. Growth of Solutions with Respect to Maximum Modulus
Alongside of theT-order, we may also define theM-order of growth off∈ HDby
σM
f
:lim sup
r→1−
loglogM r, f
−log1−r , 2.15
where Mr, f : max|z|r|fz|is the maximum modulus of f. It is well known that the inequalities
σT f
≤σM f
≤σT f
1 2.16
are satisfied for all f ∈ HD, and all possibilities allowed by 2.16 can be assumed 9, Theorems 3.5–3.7 . A functionf ∈ HDis said to belong to the weighted Hardy spaceHα∞, if there existsα≥0 such that
sup
z∈D
1− |z|2 α fz<∞. 2.17
Functions of maximal growth in
α>pHα∞ are distinguished by denotingf ∈ H∞p, if p inf{α≥ 0 :f ∈Hα∞}. Remark thatH0∞ H∞is the space of all bounded analytic functions inD. The union
α>0Hα∞is also known as the Korenblum spaceA−∞10 , and since11 H∞p is also known asGp.
If the growth of coefficients is measured by means of maximum modulus estimates, then it is natural to consider the growth of solutions with respect toM-order.
Theorem Bsee3, Theorem 1.4 . Suppose thataj∈H∞pj1k−j, wherepj≥ −1 forj0, . . . , k− 1, and denotepk:−1.
iSuppose that
j1,...,kmin k
p0−pj
j pj
>1, 2.18
and let 1 ≤ α < ∞. Then all solutions f of 1.1 satisfyσMf ≤ α if and only if maxj0,...,k−1{pj} ≤α.
iiAll nontrivial solutionsfof1.1satisfyσMf≤maxj0,...,k−1{pj}, and
j1,...,kmin k
p0−pj
j pj
≤max σM
f ,1
. 2.19
iiiSuppose that 2.18 holds. If q ∈ {0, . . . , k − 1} is the smallest index for which pq maxj0,...,k−1{pj}, then each solution base of 1.1contains at leastk−qlinearly independent solutionsfsuch thatσMf pq.
To conclude 3, Equation 4.17 in the proof of Theorem B, the inequality 3, Equation1.9 , corresponding to2.18, must be strict. By a simple modification of the proof of Theorem B, the assumption2.18can be relaxed to
j0,...,k−1max pj
>1, 2.20
which allows us to apply Theorem Biiialso in the case that there are solutionsf satisfying σMf ≤ 1. To see that 2.20 is in fact a weaker assumption than 2.18, we refer to 2, Example 10 , which is further considered inSection 3.2. In this case
j1,...,kmin k
p0−pj
j pj
−4, max
j0,...,k−1
pj
>1. 2.21
Note that by takingjkin2.18, we obtainp0>1. Hence2.18implies2.20.
Theorem 2.5corresponds toTheorem 2.1.
Theorem 2.5. Suppose that aj ∈ H∞pj1k−j, where pj ≥ −1 forj 0, . . . , k −1, and let q ∈ {0, . . . , k−1}be the smallest index for whichpq maxj0,...,k−1{pj}. If s ∈ {0, . . . , q}, then each solution base of 1.1contains at leastk−slinearly independent solutionsfsuch that
js1,...,kmin
k−s ps−pj
j−s pj
≤max σM
f ,1
. 2.22
Note that2.22gives information onσMfonly in the case when the minimum in 2.22is strictly greater than 1. The cases0 inTheorem 2.5reduces to Theorem Bii, and the casesqreduces to Theorem Biiiwith the assumption2.20, since now
jq1,...,kmin
k−q pq−pj
j−q pj
pq max
j0,...,k−1
pj
, 2.23
where the minimum is attained for j k. For a similar argumentation, see 2.5. Our contribution is to extend2.22fors ∈ {1, . . . , q−1}. The proof ofTheorem 2.5is sketched inSection 7, and the sharpness and the the special casesk2 andk3 are further discussed inSection 3.2.
Letq∈ {0, . . . , k−1}be the smallest index for whichpq maxj0,...,k−1{pj}. Ifpq ≤1, then all solutionsfin each solution base of1.1satisfyσMf≤1 by Theorem Bii. Suppose thatpq>1. In order to state the following corollaries ofTheorem 2.5, we denote
βMs: min
js1,...,k
k−s ps−pj
j−s pj
, s0, . . . , q, 2.24
wherepk:−1. Moreover, we define s:min
s∈
0, . . . , q
:βMs>1
. 2.25
Remark thatβMq>1, since2.23impliespq βMq.
Corollary 2.6. Suppose that aj ∈ H∞p
j1k−j, wherepj ≥ −1 for j 0, . . . , k −1, and letq ∈ {0, . . . , k−1}be the smallest index for whichpq maxj0,...,k−1{pj}>1. Then each solution base of 1.1admits at mosts ≤qsolutionsf satisfyingσMf < βMs. In particular, there are at most s≤qsolutionsfsatisfyingσMf≤1.
To estimate the quantityk
j1σMfjby usingTheorem 2.5, we set γM
j
:max
βM0, . . . , βM
j
, j 0, . . . , q. 2.26
EvidentlyγMj>1 forj∈ {s, . . . , q}, andγMj≤1 forj ∈ {0, . . . , s−1}.
Corollary 2.7. Suppose that aj ∈ H∞p
j1k−j, wherepj ≥ −1 for j 0, . . . , k −1, and letq ∈ {0, . . . , k−1} be the smallest index for which pq maxj0,...,k−1{pj} > 1. Let {f1, . . . , fk} be a solution base of 1.1. Ifq0, thenk
j1σMfj kp0, while ifq≥1, then k−q
pqq−1
js
γM j
≤k
j1
σM fj
≤kpq. 2.27
Note that the sum in2.27is considered to be empty, ifsq.
2.4. Gundersen-Steinbart-Wang Method forM-Order
We proceed to give an alternative statement ofTheorem 2.5and its corollaries by modifying the key steps in4 . This yields a natural way to define the possible intervals forM-orders of solutions of1.1. As a consequence, we get a useful estimate following fromCorollary 2.7.
Setδj : pj1k−jfor allj 0, . . . , k−1. Lets1∈ {0, . . . , k−1}be the smallest index satisfyingps1maxj0,...,k−1{pj}>1, which is equivalent to
δs1
k−s1 max
j0,...,k−1
δj
k−j
>2. 2.28
If s1 cannot be found, then all solutions f of 1.1 satisfy σMf ≤ 1 by Theorem Bii.
Otherwise, for a givensm,m∈N, letsm1∈ {0, . . . , sm−1}be the smallest index satisfying δsm1−δsm
sm−sm1 max
j0,...,sm−1
δj−δsm sm−j
>2. 2.29
Eventually this process will stop, yielding a finite list of indicess1, . . . , spsuch thatp≤kand s1> s2>· · ·> sp≥0. Further, set
BMt: δst−δst−1
st−1−st −1, t1, . . . , p, 2.30 wheres0 : kandδk : 0. ByExample 3.1, it is possible that1.1possesses a solutionf of M-order strictly greater than one such thatσMf/BMtfor allt1, . . . , p.
The following lemma, which can be proved similarly thanLemma 2.4, allows us to view the results inSection 2.3in a new perspective.
Lemma 2.8. One has the following:
iBM1>BM2>· · ·>BMp>1;
iiβMst BMtfor allt∈ {1, . . . , p};
iiiγMq BM1,γMj BMtfor allst≤j < st−1andt∈ {2, . . . , p}, andγMj≤1 for allj < sp. In particular,sps.
By relying onLemma 2.8,Theorem 2.5, andCorollary 2.6, we proceed to state possible intervals forM-orders of functions in solution bases of1.1in the caseaj ∈H∞p
j1k−j, where pj≥ −1 forj0, . . . , k−1. In fact, each solution base of1.1contains the following:
iat leastk−s1solutionsfsatisfyingσMf BM1;
iiat leastk−stsolutionsfsatisfyingσMf∈BMt,BM1 fort2, . . . , p;
iiiat mostspsolutionsfsatisfyingσMf∈0,BMsp.
For results of the same type, we refer to12, Theorem 1 and 13, Corollary 1 . To compareiandiito the estimates given in13, Corollary 1 , note that there is−1 in2.30 instead of−2 in13, Equation1.3 . Evidently, assertionsiandiiimprove the estimates
given for theM-orders of solutions in13, Corollary 1 . Moreover, by means of2.16we see thatiandiireduce to13, Corollary 1 , if we consider the growth of solutions of1.1 with respect toT-order.
For the following application, let{f1, . . . , fk}be a solution base of1.1. Knowing the possible intervals forM-orders, we get
k j1
σM fj
≥k−s1BM1 · · ·
sp−1−sp BM
p
sp·0δspsp−k. 2.31
Corresponding to the case in Section 2.2, by means of Lemma 2.8 we see that the lower estimates in2.27and2.31are equal.
Finally, we point out a practical estimate, which is a consequence of2.31. Ifsp 0, thenδspspδ0−sp. Ifsp>0, thenδ0−δsp/sp≤2 by2.29, andδspsp≥δ0−sp. Hence, in both cases we can state that
k j1
σM fj
≥δ0−sp−k≥p0k−sp. 2.32
We conclude that ifs1 0, then the equalities hold in2.32, since in this casesp s1 0.
Note that if2.18holds, then we can conclude thatsp0.
3. Sharpness Discussion
3.1. Sharpness ofTheorem 2.1We may assume that maxj0,...,k−1{αj} > 0, for otherwise all solutions are of zeroT-order. If k2, then the statement ofTheorem 2.1is contained in Theorem A, and all the assertions are sharp2, Examples 3 and 6 .
Ifk3, then we have three different cases to consider.
A1Ifα1, α2 ≤α0, then all nontrivial solutionsfof1.1satisfyσTf α0. In this case s0q.
A2Ifα0 < α1 andα2 ≤ α1, then in every solution base{f1, f2, f3}of1.1there are at least two solutionsf1andf2such thatσTfj α1for bothj 1,2, and all solutions fjsatisfy
σT
fj
≥min
3α0−2α1,3 2α0−1
2α2, α0
, j1,2,3. 3.1
In this cases0 ors1q.
A3If α0, α1 < α2, then in every solution base{f1, f2, f3} of1.1there is at least one solutionf1such thatσTf1 α2, two solutionsf1andf2such that
σT fj
≥min{2α1−α2, α1}, j1,2, 3.2 and all solutionsfjsatisfy3.1. In this cases0,s1, ors2q.
It is clear that the assertion inA1is sharp, and so are the ones inA2by2, Example 10 . Moreover,2, Example 9 shows that the assertions inA3are sharp fors0 ands2.
Example 3.2shows the sharpness of the assertions inA3fors 0,1,2. That is, in all cases there exists a solution for which the lower bound for theT-order of growth is attained.
3.2. Sharpness ofTheorem 2.5
We may assume that maxj0,...,k−1{pj} > 1, for otherwise all solutions f of 1.1 satisfy max{σMf,1}1, and we cannot conclude anything from2.22. Ifk2, then the statement ofTheorem 2.5is contained in Theorem B, and all the assertions are sharp by2, Examples 3 and 6 . In the case of2, Examples 3 and 6 , forβ >1, linearly independent solutionsf1and f2satisfyσMf1 βandσMf2 β2. Moreover,aj ∈ H∞p
j12−j, wherep0 β1 and p1 β2. Note that max{p0, p1} p1 β2 > 1, and henceq 1. An easy computation shows the sharpness fors 0 and fors q 1. In the case of2, Example 6 , forβ > 1, linearly independent solutions f1 and f2 satisfy σMf1 β and σMf2 β. Moreover, aj ∈ H∞pj12−j, where p0 β and p1 −1/2. Note now that max{p0, p1} p0 β > 1, and henceq 0. This example shows the sharpness fors q0. For another example, see 3, Example 2 .
Ifk3, then we have three different cases to consider.
B1Ifp1, p2≤p0, thens0q, and all nontrivial solutionsfof1.1satisfyσMf p0 by2.23.
B2Ifp0 < p1, andp2 ≤ p1, then in every solution base{f1, f2, f3}of1.1there are at least two solutionsf1andf2such thatσMfj p1for bothj 1,2, and all solutions fjsatisfy
max σM
fj
,1
≥min
3p0−2p1,3 2p0−1
2p2, p0
, j1,2,3. 3.3
In this cases0 ors1q.
B3If p0, p1 < p2, then in every solution base {f1, f2, f3} of 1.1 there is at least one solutionf1such thatσMf1 p2, two solutionsf1andf2such that
max σM
fj ,1
≥min
2p1−p2, p1
, j1,2, 3.4
and all solutionsfjsatisfy3.3. In this cases0,s1, ors2q.
It is clear that the assertion inB1is sharp. By2, Example 10 , we see that the asser- tion inB2corresponding tos 1 is sharp. In this case, for β > 1, linearly independent solutionsf1, f2, andf3 satisfyσMf1 σMf2 βandσMf3 0. Nowaj ∈ H∞pj13−j, wherep0 2/3β,p1 β2, andp2 0. Moreover, by2, Example 9 , we see that the assertions in B3 are sharp fors 0 and s 2. In this case for β > 1, linearly indepen- dent solutionsf1,f2, andf3 satisfyσMf1 σMf2 βandσMf3 2β. Moreover,aj ∈ H∞p
j13−j, wherep0 4/3β,p1 β, andp2 2β.Example 3.2shows the sharpness of the assertions inB3fors0,1,2. That is, in all cases there exists a solution for which equality holds in2.22.
3.3. Examples
Example 3.1shows that a unit disc counterpart of the finite set constructed by Gundersen- Steinbart-Wang does not contain growth orders of solutions of
fa1zfa0zf 0, 3.5
if coefficients belong either to weighted Bergman spaces or to weighted Hardy spaces.
Example 3.1. Letα, β∈Rbe constants satisfying 1< β < α <2β−1. Then the functions
f1z 1−zαβexp 1
1−z α
1
1z β
,
f2z 1−zαβexp 1
1z
β 3.6
are linearly independent analytic solutions of3.5, where
a0z β2
1z22β β
α3β γz 1−z1z2β − α
αβ 1−z2α
− αβ
1−z1α1z1ββ αβ 1−z2 ,
a1z 2β
1z1β− α
1−z1α α2β−1 1−z
3.7
belong toHD, andγ αβ−2/α3β∈0,1/2.
It is clear thataj ∈A1/2−jαj , whereα0 β−1 andα1 α−1. We calculate thats1 1, s2 0,BT1 α−1, andBT2 2β−α−1. Hence2β−α−1, α−1 is the only possible interval forT-orders of solutions of3.5. SinceσTf2 β−1, we conclude that theT-order of a solution does not have to be one of the endpoints.
On the other hand, it is also clear thataj ∈ H∞pj12−j, wherep0 βandp1 α. We calculate thats1 1,s2 0,BM1 α, andBM2 2β−α. Hence2β−α, α is the only possible interval forM-orders of solutions of3.5. SinceσMf2 β, we conclude that the M-order of a solution does not have to be one of the endpoints.
The following example demonstrates the sharpness of Theorems2.1and2.5in the case that they do not reduce to known results.
Example 3.2. Letβ >1, and denotegz 5/1−zβ. Then the functions
fjz 1−zβexp
gzj , j 1,2,3, 3.8
are linearly independent solutions offa2zfa1zfa0zf 0, where
a2z P2
gz
1−zQ
gz, a1z P1
gz
1−z2Q
gz, a0z β3P0
gz
1−z3Q
gz 3.9
are such that
P2ζ 54βζ8−27βζ7−24βζ6
108β54 ζ5−
82β63
ζ43βζ3
22β39 ζ2−
6β24 ζ6, P1ζ −108β2ζ1072β2ζ9
27β2−54β ζ8
27β−135β2 ζ7
24β15β2 ζ6
64β2−108β−18 ζ5
2182β−51β2 ζ4
−
3β15β2 ζ3
31β2−22β−13 ζ2
86β−14β2 ζ2β2−2, P0ζ 108ζ11−234ζ10126ζ9123ζ8−276ζ7183ζ6−104ζ540ζ4
−6ζ3−4ζ2,
Qζ −18ζ521ζ4−13ζ28ζ−2.
3.10
The zeros ofQζlie in the open disc of radius 121/18 centered at the origin by14, Lem- ma 1.3.2 . Since|gz|>|5/1−z|>5/2>39/18 for allz∈D, we conclude thata0, a1, a2 ∈ HD. In fact, the coefficientsa0, a1, and a2satisfy
a2z∼ 1
1−z 3β1
, a1z∼
1 1−z
5β2
, a0z∼
1 1−z
6β3
, 3.11
in a neighborhood ofz1, while they are bounded in a neighborhood of any boundary point in∂D\ {1}.
Note thataj ∈A1/3−jαj , whereα2 3β−1,α1 5/2β−1, andα0 2β−1. Evidently σTfj βj−1 forj 1,2,3. We deduce that there is one solutionf3such thatσTf3 α2 3β−1, two solutionsf2andf3such that
σT
f3
> σT
f2
min{2α1−α2, α1}2β−1, 3.12
and three solutionsf1,f2, andf3such that
σT
f3
> σT
f2
> σT
f1
min
3α0−2α1,3 2α0− 1
2α2, α0
β−1. 3.13
That is, in all casess 0,1,2 there exists a solution for which the lower bound in2.4is attained. Further, this example is in line withCorollary 2.2, since all solutionsf1,f2, andf3
are of strictly positiveT-order, and in this cases∗0.
NowγT0 βT0 β−1,γT1 βT1 2β−1, andγT2 βT2 3β−1. It follows that for the solution base{f1, f2, f3}equality holds in the first inequality in2.9, and for the solution base{f1f3, f2f3, f3}equality holds in the last inequality in2.9. This shows the sharpness ofCorollary 2.3.
On the other hand,aj ∈H∞p
j13−j, wherep2 3β,p1 5β/2, andp0 2β. Evidently σMfj βj forj 1,2,3. We deduce that there is one solutionf3such thatσMf3 p2 3β >1, two solutionsf2andf3such that
σM
f3
> σM
f2
min
2p1−p2, p1
2β >1, 3.14
and three solutionsf1,f2, andf3such that
σM
f3
> σM
f2
> σM
f1
min
3p0−2p1,3 2p0− 1
2p2, p0
β. 3.15
That is, in all casess 0,1,2 there exists a solution for which the lower bound in2.22is attained. Further, this example is in line withCorollary 2.6, since all solutionsf1,f2, andf3
are ofM-order strictly greater than 1, and in this cases0.
NowγM0 βM0 β,γM1 βM1 2β, andγM2 βM2 3β. It follows that for the solution base{f1, f2, f3}equality holds in2.27, and for the solution base{f1 f3, f2f3, f3}upper bound for the sum ofM-orders is attained. This shows the sharpness of Corollary 2.7.
4. Proof of Theorem 2.1
The following lemma on the order reduction procedure originates fromC.
Lemma Csee4, Lemma 6.4 . Letf0,1, f0,2, . . . , f0,mbe m≥2 linearly independent meromorphic solutions of
yka0,k−1zyk−1· · ·a0,0zy0, k≥m, 4.1
wherea0,0z, . . . , a0,k−1zare meromorphic functions inD. For 1≤p≤m−1, set
fp,j
fp−1,j1 fp−1,1
, j1, . . . , m−p. 4.2
Thenfp,1, fp,2, . . . , fp,m−pare linearly independent meromorphic solutions of
yk−pap,k−p−1zyk−p−1· · ·ap,0zy0, 4.3
where
ap,jz k−p1
nj1
n j1
ap−1,nzfp−1,1n−j−1z
fp−1,1z 4.4
forj 0, . . . , k−p−1. Herean,k−nz≡1 for alln0, . . . , p.
Lemma Dsee15, Lemma Eb . Letkandjbe integers satisfyingk > j≥0, and letε >0. Iff is meromorphic inDsuch thatσTf<∞, andfj/≡0, then
D
fkz fjz
1/k−j
1− |z|σTfεdmz<∞. 4.5
4.1. Cases1
Let k ≥ 3, q ≥ 2, s 1, and βT1 > 0, since otherwise there is nothing to prove.
In particular, if α1 ≤ 0, then 2.4 is trivial, since by taking j k in 2.6, we obtain βT1≤α1≤0. Let{f0,1, f0,2, . . . , f0,k}be a solution base of1.1, and assume on the contrary to the assertion that there exists12 linearly independent solutionsf0,1andf0,2such that max{σTf0,1, σTf0,2}:σ < βT1. Then the meromorphic functiong: f0,1/f0,2satisfies σTg≤σ. Moreover, Lemma C implies thatgsatisfies
gk−1a1,k−2zgk−2· · ·a1,0zg0, 4.6
where
a1,jz a0,j1z k
nj2
n j1
a0,nzf0,1n−j−1z
f0,1z 4.7
forj0,1, . . . , k−2, anda0,kz≡1. Therefore
|a0,1z| ≤ |a1,0z|k
n2
n 1
|a0,nz|
f0,1n−1z f0,1z
, 4.8
where
|a1,0z| ≤
gk−1z gz
|a1,k−2z|
gk−2z gz
· · ·|a1,1z|
gz gz
, 4.9
sincegsatisfies4.6. Putting the last two inequalities together, we obtain
|a0,1z|k−1
j1
a1,jz
gjz gz
k
n2
|a0,nz|
f0,1n−1z f0,1z
. 4.10
Letε >0. Raising both sides to the power 1/k−1and integrating over the discD0, rof radiusr∈0,1with respect to1− |z|2α1−εdmz, we obtain
D0,r|a0,1z|1/k−1
1− |z|2 α1−εdmz
k−1
j1
D
a1,jz1/k−1
gjz gz
1/k−1
1− |z|2 α1−εdmz
k
n2
D|a0,nz|1/k−1
f0,1n−1z f0,1z
1/k−1
1− |z|2 α1−εdmz.
4.11
To deal with the second sum in4.11, consider
In:
D|a0,nz|1/k−1
f0,1n−1z f0,1z
1/k−1
1− |z|2 α1−εdmz, n2, . . . , k. 4.12
By Lemma D,
Ik
D
f0,1k−1z f0,1z
1/k−1
1− |z|2 α1−εdmz<∞ 4.13
for ε > 0 being small enough since σTf0,1 ≤ σ < βT1 ≤ α1. Moreover, by H ¨older’s inequality, with indicesk−1/k−nandk−1/n−1, we have
In≤
D|a0,nz|1/k−n
1− |z|2 αnεdmz
k−n/k−1
·
⎛
⎜⎝
D
f0,1n−1z f0,1z
1/n−1
1− |z|2 ω1ndmz
⎞
⎟⎠
n−1/k−1 4.14
for alln2, . . . , k−1, where
ωsn: k−sαs−αn
n−s αn−2k−n−s
n−s ε. 4.15
The first member in the product is finite sincea0,n ∈ A1/k−nαn for alln 2, . . . , k−1 by the assumption, and so is the second one forε >0 small enough since
σT
f0,1
≤σ < βT1≤ k−1α1−αn
n−1 αn, n2, . . . , k−1, 4.16 by the antithesis. Thusk
n2Inis finite forε >0 being small enough.