mjfi-10.tex August8,2016 REMARKS ON THE MIXED JOINT UNIVERSALITY FOR A CLASS OF ZETA-FUNCTIONS ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO ABSTRACT. TworemarksrelatedwiththemixedjointuniversalityforapolynomialEu- lerproductj (s)andaperiodicHurwitzzeta-functionz (s,a ;B), whena isatranscen- dentalparameter,aregiven.Oneisthemixedjointfunctionalindependence,andtheother 6 isageneralizeduniversality,whichincludesseveralperiodicHurwitzzeta-functions. 1 0 Keywords: functional independence, periodic Hurwitz zeta-function, polynomial Euler pro- 2 g duct,universality. u AMSclassification: 11M41,11M35. A 5 1. INTRODUCTION ] T In our former paper [5], we have shown a certain mixed joint universality theorem, N which is valid for an Euler product of rather general form, and a periodic Hurwitz zeta- . h function. t a Inthepresentnote,wegivetworemarks relatedwiththeresultin[5]. Thefirstremark m is on the mixed joint functional independence result. It is well known that functional [ independence properties can be deduced from universality results. We will show that 2 v such afunctionalindependenceis alsovalidinourpresent mixedjointsituation. 5 9 Thesecondremarkisonageneralizationoftheresultin[5]. Itisanimportantproblem 7 tostudyhowgeneralthemixedjointuniversalitypropertyholds. Wewillproveagenera- 3 0 lizedlimittheoremandageneralizeduniversalitytheorem,whichinvolveseveralperiodic . 1 Hurwitzzeta-functions,underacertain matrixcondition. 0 6 1 2. FUNCTIONAL INDEPENDENCE : v i The history of the problem on the functional independence of Dirichlet series goes X r back to the famous lecture of D. Hilbert [2] in 1900. He mentioned that the Riemann a zeta-function z (s)does notsatisfyany non-trivialalgebraicdifferentialequation. Recall thedefinitionofz (s). LetPbethesetofallprimenumbers,andCthesetofall complexnumbers. Fors=s +it ∈C, z (s)isgivenby ¥ −1 z (s)= (cid:229) 1 = (cid:213) 1− 1 ms (cid:18) ps(cid:19) m=1 p∈P for s >1, and can be analytically continued to the whole complex plane C except for a simplepoleat thepoints=1 withresidue1. 1 2 ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO In 1973, S.M. Voronin proved [16] the following functional independence result on z (s). LetNandN bethesetofpositiveintegers,andnon-negativeintegers,respectively. 0 Theorem A ([16]). Let N ∈N andn∈N . Thefunctionz (s)does notsatisfyanydiffer- 0 entialequation n (cid:229) sjF z (s),z ′(s),...,z (N−1)(s) ≡0 j j=0 (cid:16) (cid:17) forcontinuousfunctionsF , j=0,...,n,notall identicallyzero. j Laterthis resultwas generalized to otherzeta- and L-functions. Forthesurvey,see the monographsby A. Laurincˇikas[6]and J.Steuding [15]. Nowadaysinanalyticnumbertheory,theinvestigationofstatisticalproperties(andalso thefunctionalindependence)foracollectionofvariouszeta-functions,someofthemhave the Euler product expansion and others do not have, is a very interesting problem since an importantroleisplayed by parametersincluded inthedefinitionofthosefunctions. The first result in this direction is due to H. Mishou. In 2007, he proved [12, Theo- rem 4] that the pair of zeta-functions consisting of the Riemann zeta-function z (s) and theHurwitzzeta-function z (s,a )isfunctionallyindependent. We recall that the Hurwitz zeta-function z (s,a ) with a fixed parameter a , 0<a ≤1, isdefined bytheDirichletseries ¥ z (s,a )= (cid:229) 1 (m+a )s m=0 for s > 1, and can be continued to the whole complex plane, except, for a simple pole at the point s = 1 with residue 1. In general, the function z (s,a ) has no Euler product overprimes,exceptforthecasesa =1anda = 1,whenz (s,a )isessentiallyreducedto 2 z (s). ThenMishou’sresultis thefollowingstatement. Theorem B([12]). Supposethata istranscendental. LetN ∈N,n∈N ,andF :C2N → 0 j C beacontinuousfunctionforeach j=0,...,n. Suppose n (cid:229) sj·F z (s),z ′(s),...,z (N−1)(s),z (s,a ),z ′(s,a ),...,z (N−1)(s,a ) ≡0. j j=0 (cid:0) (cid:1) Then F ≡0 for j=0,...,n. j This is the first “mixed joint” functional independence theorem. Later this result was generalized to the collection of a periodic zeta-function and a periodic Hurwitz zeta- functionbythefirst-named authorand A. Laurincˇikasin[4]. Inthispaper,wewillproverathergeneralresultonthemixedjointfunctionalindepen- dence for a class of zeta-functions, consisting of the so-called Matsumoto zeta-functions and periodicHurwitzzeta-functions. Let B = {b : m ∈ N } be a periodic sequence of complex numbers (not all zero) m 0 with minimal period k∈N, and 0<a ≤1. In 2006, A. Javtokas and A. Laurincˇikas [3] REMARKSON THEMIXEDJOINTUNIVERSALITY 3 introduced the periodic Hurwitz zeta-function z (s,a ;B). For s > 1, it is given by the series ¥ z (s,a ;B)= (cid:229) bm . (m+a )s m=0 It isknownthat z (s,a ;B)= 1 k(cid:229)−1b z s,l+a , s >1. ks l (cid:18) k (cid:19) l=0 Therefore the function z (s,a ;B) can be analytically continued to the whole complex plane, exceptforapossiblesimplepoleat thepoints=1withresidue 1k(cid:229)−1 b:= b . l k l=0 Ifb=0, thecorrespondingperiodicHurwitzzeta-functionis an entirefunction. ThefunctionalindependenceofperiodicHurwitzzeta-functionswasprovedbyA.Lau- rincˇikas in[7, Theorem1]. Nowwe recall thedefinitionofthepolynomialEulerproductsj (s),or, so-called Mat- sumotozeta-functions. e Let, for m ∈ N, g(m) ∈ N, and, for j ∈ N and 1 ≤ j ≤ g(m), f(j,m) ∈ N. Denote by p the mth prime number, and a(j) ∈ C. The zeta-function j was introduced by the m m second-named authorin[11], and itis defined bythepolynomialEulerproduct e ¥ g(m) j (s)= (cid:213) (cid:213) 1−a(j)p−sf(j,m) −1. (1) m m m=1 j=1(cid:16) (cid:17) e Supposethat a (j) b g(m)≤C p and |a |≤ p (2) 1 m m m with a positive constant C and non-negative constants a and b . In view of (2), the 1 function j (s) converges absolutely for s >a +b +1 (see Appendix), and hence in this regionitcan beexpressedas theDirichletseries e ¥ j (s)= (cid:229) ck, (3) ks k=1e e withcoefficients c . Forbrevity,denotetheshiftedversionofthefunctionj (s)by k ¥ ¥ e j (s)=j (s+a +b )= (cid:229) ck = (cid:229) ck e (4) ks+a +b ks k=1 e k=1 e with c = k−a −b c . Then, for s > 1, the series on the right-hand side of (4) converges k k absolutely. e The aim of this paper is to obtain a mixed joint functional independence of the collec- tion of zeta-functions consisting of the Matsumoto zeta-function j (s) belonging to the SteudingsubclassS, defined below,and periodicHurwitzzeta-functionsz (s,a ;B). Wesaythatthefunctionj (s)belongstotheclassSiffollowingconditionsaresatisfied: e e 4 ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO (i) thereexistsaDirichletseriesexpansion ¥ j (s)= (cid:229) a(m) ms m=1 witha(m)=O(me )foreverye >0; (ii) there existss j <1 such that j (s) can be meromorphicallycontinued to thehalf- planes >s j ; (iii) thereexistsaconstantc≥0such that j (s +it)=O(|t|c+e ) foreveryfixed s >s j and e >0; (iv) thereexiststheEulerproductexpansionoverprimenumbers,i.e., j (s)= (cid:213) (cid:213) l 1−aj(p) −1; (cid:18) ps (cid:19) p∈Pj=1 (v) thereexistsaconstantk >0 such that lim 1 (cid:229) |a(p)|2=k , x→¥ p (x) p≤x wherep (x) denotesthenumberofprimes p, p≤x. This class was introduced by J. Steuding in [15], and is a subclass of the class of Matsumotozeta-functions. Forj ∈S, let s ∗ betheinfimumofall s forwhich 1 1 T |j e(s +it)|2dt ∼ (cid:229)¥ |a(m)|2 2T Z−T m=1 m2s holdsforanys ≥s . Then itisknownthat 1 ≤s ∗ <1 (see[15, Theorem2.4]). 1 2 Westatethefirst mainresultofthispaperin thefollowingtheorem. Theorem 1. Suppose that a is a transcendental number, and the function j (s) belongs to the class S. Let N ∈ N, n ∈ N , and the function F : C2N → C be continuous for 0 j j=0,1,...,n. If e n (cid:229) sj·F j (s),j ′(s),...,j (N−1)(s),z (s,a ;B),z ′(s,a ;B),...,z (N−1)(s,a ;B) j j=0 (cid:0) (cid:1) isidenticallyzero,thenF ≡0 for j=0,1,...,n. j 3. PROOF OF THEOREM 1 To the proof of the mixed joint functional independence for the functions j (s) and z (s,a ;B), we need two further propositions: the mixed joint universality theorem and theso-calleddenseness lemma. REMARKSON THEMIXEDJOINTUNIVERSALITY 5 3.1. Mixedjointuniversalityofthefunctionsj (s)andz (s,a ;B). TheproofofTheo- rem 1 is based on themixed joint universalitytheorem in the Voronin sense for the func- tions j (s) and z (s,a ;B). It was obtained by the authors in [5, Theorem 2.2]. We will givethe statement of this universalitytheorem as a lemma. Let R be the set of real num- bers, and D(a,b)={s∈C:a <s <b} for any a<b. For any compact subset K ⊂C, denotebyHc(K)thesetofall C-valuedcontinuousfunctionsdefined on K,holomorphic in the interior of K. By Hc(K) we mean the subset of Hc(K) consisting of all elements 0 which arenon-vanishingonK. Lemma 2 ([5]). Suppose that j ∈ S, and a is a transcendental number. Let K be a 1 compact subset of D(s ∗,1), and K be a compact subset of D(1,1), both with connected 2 e 2 complements. Supposethat f ∈Hc(K )and f ∈Hc(K ). Then, foreverye >0, 1 0 1 2 2 1 liminf m t ∈[0,T]: sup|j (s+it )− f (s)|<e , 1 T→¥ T (cid:26) s∈K1 sup|z (s+it ,a ;B)− f (s)|<e >0, 2 (cid:27) s∈K2 where m {A} denotes theLebesgue measureof themeasurablesetA⊂R. Note that for the proof of Lemma 2 we use the joint mixed limit theorem in the sense of weakly convergent probability measures for the Matsumoto zeta-functions j (s) and the periodic Hurwitz zeta-function z (s,a ;B). It was proved by the authors in [5, Theo- rem 2.1]. 3.2. A denseness lemma. FortheproofofTheorem1, weneed adenseness lemma. Definethemapu:R→C2N bytheformula u(t) = j (s +it),j ′(s +it),...,j (N−1)(s +it), (cid:0) z (s +it,a ;B),z ′(s +it,a ;B),...,z (N−1)(s +it,a ;B) (cid:1) with 1 <s <1. 2 Lemma 3. Supposethata istranscendental. Then theimageofR byu isdenseinC2N. Proof. Wewillgiveasketch,sincetheprooffollowsinthesamewayas Lemma13from [4](see, also, [12,Theorem 3]). Wecan find asequence{t m :t m ∈R}, limm→¥ t m =¥ ,such thattheinequalities e |j (j)(s +it )−s |< m 1j 2N and e |z (j)(s +it ,a ;B)−s |< m 2j 2N 6 ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO hold for every e > 0 and arbitrary complex numbers s , l = 1,2, j = 0,...,N−1. To lj showthis,we considerthepolynomial s ·sN−1 s ·sN−2 s l,N−1 l,N−2 l0 p (s)= + +...+ , l =1,2. lN (N−1)! (N−2)! 0! Then, for j=0,...,N−1 and l =1,2,wehave (j) p (0)=s . lN lj Now,inviewofLemma2andrepeatinganalogousargumentsasintheproofofLemma13 from[4],wecanprovetheexistenceoftheabovesequence{t }andobtainthattheimage m ofR bythemapu isdensein C2N. (cid:3) 3.3. Proof of Theorem 1. Now we are ready to complete the proof of Theorem 1. The essentialideaisduetoVoronin (see, forexample,[17]). Wefirst provethatF ≡0. n Supposethat F 6≡0. It followsthat thereexistsapoint n a=(s ,s ,...,s ,s ,s ,...,s )∈C2N 10 11 1,N−1 20 21 2,N−1 such that F (a)6= 0. From the continuity of the function F , we find a bounded domain n n G⊂C2N such thata∈G,and, forall s∈G, |F (s)|≥c>0. (5) n By Lemma3, thereexistsa sequence{t m :t m ∈R},limm→¥ t m =¥ , suchthat j (s +it),j ′(s +it),...,j (N−1)(s +it), (cid:0) z (s +it,a ;B),z ′(s +it,a ;B),...,z (N−1)(s +it,a ;B) ∈G. (cid:1) Butthistogetherwith(5)contradictstothehypothesisofthetheoremift issufficiently m large. Hence, F ≡0. n Similarlywe can showF ≡0,...,F ≡0, inductively. Theproofiscomplete. n−1 0 4. A GENERALIZATION Themixedjointuniversalityand themixedjoint functionalindependencetheorem can beobtainedinthefollowingmoregeneral situation. Suppose that a be a real number with 0 < a < 1, l(j) ∈ N, j = 1,...,r, and l = j j l(1)+...+l(r). Foreach j and l, 1≤ j≤r,1≤l ≤l(j),letB ={b ∈C:m∈N } jl mjl 0 be a periodic sequence of complex numbers (not all zero) with the minimal period k , jl and let z (s,a ;B ) be the corresponding periodic Hurwitz zeta-function. Denote by k j jl j the least common multiple of periods k ,...,k . Let B be the matrix consisting of j1 jl(j) j REMARKSON THEMIXEDJOINTUNIVERSALITY 7 coefficients b from theperiodicsequences B , l =1,...,l(j),m=1,...,k ,i.e., mjl jl j b b ... b 1j1 1j2 1jl(j) b b ... b B = 2j1 2j2 2jl(j) . j ... ... ... ... b b ... b kjj1 kjj2 kjjl(j) The functional independence for the above collection of periodic Hurwitz zeta-func- tions was proved by A. Laurincˇikas in [7, Theorem 3] under a certain matrix condi- tion. The proof is based on the joint universality theorem among periodic Hurwitz zeta- functionsprovedbyJ. Steudingin[14]. Fortheproofof mixedjointfunctionalindependence, we may adoptthe methoddeve- loped in series of works by A. Laurincˇikas and his colleagues (see, for example, [1], [8], [13]). Then itispossibletoobtainthefollowinggeneralizationofTheorem 1. Theorem4. Supposethata ,...,a arealgebraicallyindependentoverQ,rankB =l(j), 1 r j 1 ≤ j ≤ r, and j (s) belongs to the class S. Let the function F : CN(l +1) → C be a j continuousfunctionforeach j=0,...,n. Supposethatthefunction e n (cid:229) sj·F j (s),j ′(s),...,j (N−1)(s), j (cid:18) j=0 z (s,a ;B ),z ′(s,a ;B ),...,z (N−1)(s,a ;B ),..., 1 11 1 11 1 11 z (s,a ;B ),z ′(s,a ;B ),...,z (N−1)(s,a ;B ),..., 1 1l(1) 1 1l(1) 1 1l(1) z (s,a ;B ),z ′(s,a ;B ),...,z (N−1)(s,a ;B ),..., r r1 r r1 r r1 z (s,a ;B ),z ′(s,a ;B ),...,z (N−1)(s,a ;B ) r rl(r) r rl(r) r rl(r) (cid:19) is identicallyzero. Then F ≡0for j=0,...,n. j Thistheoremisaconsequenceofthefollowingmixedjointuniversalitytheorem,which is a generalization of Lemma2. This theorem is also an analogue of a result of J. Genys, R. Macaitiene˙, S. Racˇkauskiene˙ and D. Šiaucˇiu¯nas [1, Theorem 3], which treats the case that j (s)isreplaced byz (s). Theorem5. Supposethata ,...,a arealgebraicallyindependentoverQ,rankB =l(j), 1 r j 1≤ j ≤r, and j (s) belongs to the class S. Let K be a compact subset of D(s ∗,1), K 1 2jl be compact subsets of D(1,1), all of which with connected complements. Suppose that 2 e f ∈Hc(K )and f ∈Hc(K ). Then, foreverye >0, 1 0 1 2jl 2jl 1 liminf m t ∈[0,T]: sup|j (s+it )− f (s)|<e , 1 T→¥ T (cid:26) s∈K1 max max sup |z (s+it ,a ;B )− f (s)|<e >0. j jl 2jl 1≤j≤r1≤l≤l(j)s∈K2jl (cid:27) 8 ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO Remark1. Considerthecasewhenalll(j)=1,1≤ j≤r. WriteB =B . Inthiscase, j1 j the condition rankB =l(j) trivially holds. Therefore the joint universalityand the func- j tional independence among the functions j (s), z (s,a ,B ),...,z (s,a ,B ) are valid 1 1 r r withoutany matrixcondition,onlyundertheassumptionthat a ,...,a are algebraically 1 r independentoverQ. WhenA.Laurincˇikasstartedhisstudyontheuniversalityofperiodic Hurwitz zeta-functions, he assumed various matrix-type conditions, but finally, the joint universalityamongperiodicHurwitzzeta-functionswithoutanymatrixconditionwases- tablished by A. Laurincˇikas and S. Skerstonaite˙ in [9, Theorem 3]. Our Theorems 4 and 5 includethe“mixed”generalizationofthistheorem. 5. PROOF OF THEOREMS 4 AND 5 In the proof of Theorems 4 and 5, the crucial role is played by a mixed joint limit theorem in the sense of weakly convergent probability measures in the space of analytic functions. 5.1. A generalized mixed joint limit theorem. Let D be an open subset of D(s ∗,1), 1 andD beanopensubsetofD(1,1). ForanysetS,byB(S)wedenotethesetofallBorel 2 2 subsets of S. For any region D, denote by H(D) the set of all holomorphic functions on D. LetH betheCartesian productofl +1 suchspaces, i.e. H =H(D )×H(D )×...×H(D ). 1 2 2 l | {z } Moreover,let ¥ W = (cid:213) g and W = (cid:213) g , 1 p 2 m p∈P m=0 whereg =g forall p∈P,g =g forall m∈N , andg ={s∈C:|s|=1}, anddefine p m 0 W =W ×W ×...×W 1 21 2r where W =W for all j =1,...,r. Then by the Tikhonovtheorem W is a compact topo- 2j 2 logicalAbeliangroupalso. Thenwehavetheprobabilityspace(W ,B(W ),m ). Herem H H is the product of Haar measures m ,m ,...,m , where m is the probability Haar H1 H21 H2r H1 measure on (W ,B(W )) and m is the probability Haar measure on (W ,B(W )), 1 1 H2j 2j 2j j=1,...,r. Letw (p) betheprojectionofw ∈W to g , and,forevery m∈N, define 1 1 1 p w (m)= (cid:213) w (p)a 1 1 pa km with respect to the factorization of m to primes. Denote by w (m) the projection of 2j w ∈W to g ,m∈N , j=1,...,r. 2j 2j m 0 REMARKSON THEMIXEDJOINTUNIVERSALITY 9 For brevity, write a = (a ,...,a ), B = B ,...,B ,···,B ,···,B , s = 1 r 11 1l(1) r1 rl(r) (s ,s ,...,s ,...,s ,...,s )∈Cl +(cid:0)1, and (cid:1) 1 211 21l(1) 2r1 2rl(r) Z(s,a ;B)=(j (s ),z (s ,a ;B ),...,z (s ,a ;B ), 1 211 1 11 21l(1) 1 1l(1) ...,z (s ,a ;B ),...,z (s ,a ;B )). 2r1 r r1 2rl(r) r rl(r) Let w =(w ,w ,...,w )∈W . Define theH-valued random element Z(s,a ,w ;B) on 1 21 2r theprobabilityspace(W ,B(W ),m )by theformula H Z(s,a ,w ;B)=(j (s ,w ),z (s ,a ,w ;B ),...,z (s ,a ,w ;B ), 1 1 211 1 21 11 21l(1) 1 21 1l(1) ...,z (s ,a ,w ;B ),...,z (s ,a ,w ;B )), 2r1 r 2r r1 2rl(r) r 2r rl(r) where j (s,w )= (cid:229)¥ cmw 1(m), s∈D(s ∗,1), 1 ms m=1 and z (s,a ,w ;B )= (cid:229)¥ bmjlw 2j(m), s∈D 1,1 , j=1,...,r, l =1,...,l(r), j 2j jl (m+a )s 2 m=0 j (cid:0) (cid:1) respectively. Theseseriesareconvergentforalmostallw ∈W andw ∈W , j=1,...,r. 1 1 2j 2j DenotebyP thedistributionoftherandomelementZ(s,a ,w ;B),i.e., Z P (A)=m (w ∈W :Z(s,a ,w ;B)∈A), A∈B(H). Z H Nowwearereadytostateourmixedjointlimittheoremforthefunctionsj (s),z (s,a ;B ), 1 11 ...,z (s,a ;B )as alemma. r rl(r) Lemma 6. Suppose that the numbers a ,...,a are algebraically independent over Q, 1 r and j ∈S. Then theprobabilitymeasureP defined by T e 1 P (A)= meas{t ∈[0,T]:Z(s+it ,a ;B)∈A}, A∈B(H), (6) T T converges weakly toP as T →¥ . Z When r =1 and l(1)=1, this lemma is Theorem 2.1 from [5]. On the other hand, the same type of limit theorem with j replaced by the Riemann zeta-function is given in [1, Theorem4]. Theproofoftheabovelemmaisquitesimilartothatofthoseresults,sohere wegiveaverybriefoutline. Let j (s), j (s,w ), z (s,a ;B), z (s,a ,w ;B) be the same as in [5, Section 3], and n n 1 n n 2 define b b Z (s,a ;B)=(j (s ),z (s ,a ;B ),...,z (s ,a ;B ), n n 1 n 211 1 11 n 21l(1) 1 1l(1) ...,z (s ,a ;B ),...,z (s ,a ;B )) n 2r1 r r1 n 2rl(r) r rl(r) 10 ROMAKACˇINSKAITE˙ ANDKOHJIMATSUMOTO and Z (s,a ,w ;B)=(j (s ,w ),z (s ,a ,w ;B ),...,z (s ,a ,w ;B ), n n 1 1 n 211 1 21 11 n 21l(1) 1 21 1l(1) ...,z (s ,a ,w ;B ),...,z (s ,a ,w ;B )), b n b2r1 r 2r r1b n 2rl(r) r 2r rl(rb) where w = (w ,w ,...,w ) ∈ W . bThe probability measuresbP , P and P on H 1 21 2r T,n T,n T are defined similarly as P in (6), with replacing Z(s+it ,a ;B) by Z (s+it ,a ;B), b b b bT b n b Z (s+it ,a ,w ;B) and Z(s+it ,a ,w ;B), respectively. As an analogueof [5, Lemma3] n or [1, Lemma 2], we can show that both measures P and P converge weakly to a T,n T,n b b certain probability measure P as T →¥ . (In the course of the proof, a key is Lemma 1 n b from [1], which is based on the assumption that a ,...,a are algebraically independent 1 r overQ.) Next we need an approximation lemma in the mean value sense, similar to [5, Lem- ma 3.3]. In the proof of [5, Lemma 3.3], we used the result of [3]. The desired approxi- mation can be shown by using, instead of [3], mean value results given in [10, (2.3), (2.5)]. Then we can prove that both P and P converge weakly to a certain probability mea- T T sureP. Thisisananalogueof[5,Lemma3.4]or[1,Lemma5]. Finallywecanshowthat b P=P , by theusual ergodicargument. ThiscompletestheproofofLemma6. Z 5.2. Completion ofthe proof. Now wecompletetheproofof Theorems 4 and 5. Here- after we assume that a ,...,a are algebraically independent over Q, rankB = l(j), 1 r j 1≤ j ≤r, and j (s)belongsto theclass S. Let Sj betheset ofall f ∈H(D1)whiech isnon-vanishingon D1, orconstantly≡0 on D . 1 l Lemma 7. Thesupportof themeasurePZ isSj ×H(D2) . This can be shown just analogously to [1, Theorem 5]. Then Theorem 5 follows from Lemmas6 and 7by thestandard argument;seeagain[1]. Next,as a generalizationofLemma3, wecan showthat theimageof themapu:R→ C(l +1)N bytheformula u(t)= j (s +it),j ′(s +it),...,j (N−1)(s +it), (cid:18) z (s +it,a ;B ),z ′(s +it,a ;B ),...,z (N−1)(s +it,a ;B ),..., 1 11 1 11 1 11 z (s +it,a ;B ),z ′(s +it,a ;B ),...,z (N−1)(s +it,a ;B ),..., 1 1l(1) 1 1l(1) 1 1l(1) z (s +it,a ;B ),z ′(s +it,a ;B ),...,z (N−1)(s +it,a ;B ),..., r r1 r r1 r r1 z (s +it,a ;B ),z ′(s +it,a ;B ),...,z (N−1)(s +it,a ;B ) r rl(r) r rl(r) r rl(r) (cid:19)