A note on a two-parameter family of operators Ab,c on weighted Bergman spaces Текст научной статьи по специальности «Математика»

Probl. Anal. Issues Anal. Vol. 8 (26), No3, 2019, pp. 125-136

DOI: 10.15393/j3.art.2019.6570

125

UDC 517.547, 517.588

S. Naik, P. K. Nath

A NOTE ON A TWO-PARAMETER FAMILY OF OPERATORS c ON WEIGHTED BERGMAN SPACES

Abstract. In this article, we prove that the two-parameter family of operators Ab'c is bounded on the weighted Bergman spaces Bpa+c-1 if a + 2 < p and unbounded if a + 2 = p.

Key words: Generalized Cesaro operator, weighted Bergman space, boundedness

2010 Mathematical Subject Classification: 47B38, 46E15

1. Introduction. Let D denote the unit disc in the complex plane C, dD its boundary, H(D) the set of all analytic functions on D and dm(-) = 1/n r drdd the normalized Lebesgue area measure on D. For 0 < p < <x, the weighted Bergman space for —1 < a < consists of functions f E H(D), such that

Bp = (a + 1)/ |f (z)|p(1 — |z|2)adm(z) =

D 1

a +W, ,w, a.

n J 0

2n

id\

Mp(r, fX1 - r2<

where Mp(r,f) = / |f(re )|vdB.

0

To define the adjoint of the generalized Cesaro operator, we need the Gaussian hypergeometric function. Let (a,n) be the shifted factorial defined by Appel's symbol

r(a + n)

(a, n) = a(a + 1)... (a + n — 1) = ^ . ;, n E N = {1, 2, 3,...}

r(a)

© Petrozavodsk State University, 2019

and (a, 0) = 1 for a = 0. The Gaussain hypergeometric function is defined by the power series expansion

F(a,b; c; z) =

n=0

(a, n)(b, n) zn (c,n)

n!

(|z| < 1),

where a, b, c are complex numbers, such that c = —m, m = 0,1, 2, 3,..., and we assume c = —m, m = 0,1, 2, 3,..., to avoid zero denominators. Clearly, F(a,b, c, z) belongs to H(D). Many properties of the hypergeometric functions are found in [1]. Asymptotic behavior of the zero-balanced (i. e., the c = a+b case) is well-known. For the non-zero balanced case, improved formulation is obtained in [4,12], whereas the geometric properties of Gaussian hypergeometric functions are considered, for example, in [9,10]. The same problems for linear and convolution operator are dealt with in [11].

Let b, c E C with Reb > 0, Re c > 0. For a function f (z) = anzn, f (z) E H(D), the two-parameter family of Cesaro averaging operators Pb'c is given by

Pb,cf (z) = E( TbTIC E bn-kak

An k=n

n=0

where

and bk are given by b0 = 1,

A!

b;c

(b,n) n)

bk

1 + b - c

Ab+l;c+l Ak-1

1 + b - c b

Ak;c

for k ^ 1. The operators Pb'c were introduced in [2] and have been studied for boundedness on various function spaces, such as Hp, BMOA, Ba [7,8], on mixed norm spaces [6], as well as on the Dirichlet space [7]. For b = 1 + a and c = 1, we obtain the generalized Cesaro operators p 1+7,1 = c7 introduced in [14]. It is known that operators CY are bounded on the Hardy space for 0 < p < x>, BMOA, and Bloch space [17] and on the Dirichlet space [16].

For b,c E C, such that Reb > 0, Rec > 0, let Ab,c be the adjoint operator of Pb,c, given by

Abcf (z) = EE

n=0

bk -nak Ab+l;c k=n Ak

c

n

z

where f (z) = ^=0 anzn E H(D) and Ablkc and bb are the same as defined for Pb'c. These operators were formally introduced in [3] and studied for boundedness on the space of Cauchy transforms. In the notation of Stempak [14], we find that

A1+7>1f = AY f.

In particular, for y = 0

00

A"f = Af = £ Ei+I a z'"

n=0 \b=n /

where AY is the adjoint operator of the generalized Cesaro operator C1 (see [17]). If y = 0, the AY is simply adjoint of the classical Cesaro operator C (see [13]). Now we recall a known result that gives an integral representation of the operator Ab'c.

Lemma 1. [3] Let b,c E C with Re b > 0, Re c > 1 and function ft,s(z) = 1 — t — s + st + tz. Then

1 1

Ab,cf (z) = mJJ sc-2(1 — t)b-1f (^t,s(z)) * F(c,1; 1; ^t,s(z))dsdt,

00

where M = (1 + b — c)(c — 1).

Here * denotes the Hadamard product (or convolution) of power series. That is, if f (z) = Y1 ^=0 anzn and g(z) = Y1 ^=0 bnzn are two analytic functions in |z| < R, then convolution between f and g is denoted by f * g and is defined by (f * g)(z) = ^=0 anbnzn. This series converges for |z| < R2. Moreover,

(f * g)(z) = 2ni i f (w)g(z/w)|z| < pR < R2. ln% J w

|w| =p

2. Preliminary results. In this section, we recall few preliminary results, which are used to state and prove the main results of this article. The adjoint operator was considered in [13] for the case (y = 0) and in [17]. In [13], Siskakis proved the following result.

Theorem 1. The operator A is bounded on the weighted Bergman space BOa if and only if a + 2 < p.

Stevic proved, in [15], a generalization of Theorem 1 for the operators AY, when y = 0:

Theorem 2. The operator AY is bounded on the weighted Bergman space Bp if and only if a + 2 < p.

The main aim of this article is to generalize Theorem 2 by finding conditions on the parameters b and c for which the operators Ab'c are bounded on the weighted Bergman spaces.

We will use the following lemma in the sequel.

Lemma 2. [5, p. 65] For each 1 < a < ro there is a positive constant C = C(a), such that

f |1 - reie|-a d0 ^ C(1 - r)-(a-1),

if 0 ^ r < 1.

Henceforth, C, K, and C1 denote positive constants, whose values are different at different occurrences.

3. Main Results. In this section, we consider the so-called convolution operator and prove its boundedness on the weighted Bergman space Ba+c-1 for c ^ 1. Also, we state and prove the main result of this paper. From now onwards, we denote F(z) = F(1, c; 1; z) for all z E D.

Lemma 3. If p E [1, ro),a> -1, c ^ 1, f E Bpa, then f * F E Bpa+{c-1)p.

Proof. Let f E Bpa. Then

1

a + 1 f , , W, „2Na

n J

0

Mp(f, r)(1 - r2)ardr < ro. (2)

Using the definition of convolution and the fact that F(1, c; 1; z) = (1-z)-for 0 < r < p < 1, we have

2n

MP(f * F, r) = 2n f |(f * F)(peid)|pdd =

1

2n

0

2n 2n

P

de

21n f F(pe^f p^) dt

2n 2n

1

2- (1 - pe«)-c/(!>-t>)dt

2n

0 0

de.

Applying Minkowski's integral inequality and Lemma 2 above, we have

2n 2n

1 Ç/ 1

Mp(f * F,r) ^ 2ny (2ny |(1 - pe^fPe^pe) p dt ^

2n J V2n

00 2n 2n

^ 2n/ t/

00

1 — pe

it

—cp

-e P

¿(0—t)

de p dt

2n 2n

— i |1 — peit|—cdt(— 2n J 1 ^ 1 V2n

0

¿c (1—p)—c+iMp(/,r :

rei(0—t)

de

From the above inequality, we find

(1 — p)(c—1)pM^(/ * F, r)(1 — r)adr ^K^ M^f,r) (1 — r)adr

where K = C/2n.

Now, taking r = p2, we have

Mp(f * F, p2)(1 — p2)a+(c—1)pdp2 ^

^ K(1 + p)(c—1)p MP (f,p)(1 — p2)adp2.

A simple calculation shows:

MPf * F, p2)(1 — p4)a+(c—1)q p2dp2 ^ K Mp (f,p) (1 — p2)ap2dp2

p

r

p

p

1

1

1

1

1

The last inequality and (2) give i

y * F,p2)(1 - p4)a+(c-1)qp2dp2 < œ. 0

This completes the proof. □

Now we give an estimate on the norm of the convolution operator )

on Ba.

Theorem 3. Let p G (0, œ),a > — 1,c ^ 1,0 : D ^ D be a non-constant analytic function. Then the operator ) = (/ * F)(0), where F = F(1,c; 1, z) on Ba+c-1(D), satisfies the following inequality:

a + 2

IIW)Ik«_. S C (MKMV II/IIBp,

C = () if a > 0 and C = () (||0||» + |0(0)|)p +3|0(0)|)-p.

Proof. We will use the method of Siskakis [13]. Let a = |0(0)| and b = «0«» and fix 0 < r < 1. By the well known consequence of the Schwarz-pick lemma on the map 01 = b-10, we have |0(z)| ^ ^o+O^, for Izl ^ r. Since a ^ b, we have (o+br) ^ ((b-a)r+2a) for all 0 < r < 1,

^ ' (b+or) ^ (b+o) '

so |0(z)| ^ bR ^ R, where R = R(r) = ((b-b+r0+2o) for all 0 < r < 1. If f * F G Ba+c-1(D), let |(u * F)(z)| be the harmonic extension |(f * F)(Rei6>)|p on |z| ^ bR; |(u * F)(z)| is continuous on |z| ^ bR and majorizes |(f * F)(z)|p there, so

|(f * F)(0(z))|p ^ |(u * F)(0(z))| for |z| ^ r.

It follows that

2n 2n

M ((f * F)0,r) = J |(f * F)(0(rei0))|pd£ |(u * F)(0(rei0))|d0. (3) 0 0

Now, for 0 < p < 1,

2n

(u * F)(p0(0)) = ~n/ F(pei0)«(0(0)e-i0)d0,

DO

2n

|(u * F)(p0(O))| = 12-/ F(pei0)u#(0)e-i0)de

0

2n

^ 2b/|F(Pei6)||u(0(O)e-ie)|de = 0

2n

= 2n f |1 - pei0|-Cu(0(O)e-i0)de.

Finally, by Harnack's inequality and the Mean Value Theorem, we have

2n

'"«'»e-) < ^"<°> = ^2n / |f <bRe")|pde- (5)

0

From (3), (4), and (5) and using Lemma 2, we obtain

2n 2n

de <

MP((f * F)0,r) ^ ( ) I |1 - pei0|-C I |f (dReit:)|Pdt

id I-c

. I f I I — I

(2n)HbR - a

00

2n

< ^^ (bR-a)^|f)|Pde =

0

D(1 - p)-c+1 (bR + a) Mp(f R) (6)

= (2n)2 ^bR-^JMpP(f,R). (6)

Now multiply both sides of (6) by (1 - r2)ar and integrate with respect to r from 0 to 1 to get

1

J MpP((f * F)<£, r)(1 - p)c-1(1 - r2)ardr ^ 0

1

^ 77^2 i ) M(f, R)(1 - r2)ardr.

(2n)27 \bR - a

Taking p = r2, we get

1

J M((f * F)0,r)(1 - r2)a+c-1rdr * 0

1

* (2^/ (^)m(f,R)(1 - r2)"rdr.

0

Proceeding as in [13], we get i

i Mp((/ * F)0, r)(1 - r2)a+c-1rdr *

a+2 1

i /(1 - u2)aMp(/,u)udu

(2n)2 Vb + 3a/ V&- a) J K J p

0

for -1 < a < 0.

™* (^ )a( b-a )"+2»/

Hence, the conclusion follows. □

The following results, regarding the boundedness of the composition operator on the Weighted Bergman space, was proved in [13], which is a particular case of c =1 of our result, as given in Theorem 3.

Corollary. Let 0 : D ^ D be a non-constant analytic function. Then the operator T^(/) = /o0 on Bp satisfies the following inequality:

a + 2

* c, u-r + |0(O)|\ p

where C =1 if a > 0 and C = (||0|U + |0(O)|)P(||0|U + 3|0(O)|)-P, -1 < a < 0.

Now we state and prove the main result of this article.

Theorem 4. Let b, c G C with Re(b) > 0 and c ^ 1. Then the operator Ab'c is bounded on the weighted Bergman space Bp+c-1 if a + 2 < p and unbounded for a + 2 = p.

oo

Proof. Case (i) Let a

Here p > 1, because a > -1. Applying Minkowski's inequality twice and taking 0 = in Theorem 3, we obtain

l|A6'c(fm)||BP

a + c-1

1 1

K

((f * F)(<Mz)))(1 -t)b-1sc-2dsdt (1-|z|2r+c-1dm(z)

U 0 0 1 1

^ K

(f * F)(<Mz))(1 -t)b-1 dt"(1 -|z|2)a+c_1dm(zn psc-2ds^

0 U 0 1 1

^ K

(f * F)(<Mz)) " (1 -|z|2)a+c-1dm(z)) p(1 -t)b-1dtsc-2ds

0 0 U

1 1

K I I llW)lBp+c-i(1 - t)b-1dtsc-2ds ^

00

1 1 a + 2

^ KC1|f ||bs // i2 - 2s - t + p (1 - t)b-1dtsc-2ds ^

t

00

^ KCJf ||Bp2^cl^ f _+_(1 - t)b-1dt. c - 1 J t

Here K = (1 + b - c)(c - 1)(a +1). The above integral is convergent for < 1. This completes the proof.

Case (ii) Let a

Suppose f1(z) = . Using Lemma 2, we obtain

/ |i 1Z|P (1 - |z|2)adm(z) =1J(1 - r2)^ |1 - rei0|-Pderdr ^

U 0 -pi

1

2«c /"

^ — (1 - r)a+1-Pdr,

which is finite for a + 2 > p. Hence, f1 E Aja+c_ 1

1

1

n

(a) For c > 1, we have:

1 1

Ab'c(/1(z)) = K ^J(/1 * F)(1 - t - s + ts + tz)(1 - t)b—V-2dsdt:

00 1 1

= K J J F (1,c; 1; 1 - t - s + ts + tz)(1 - t)b-1sc-2dsdt

00 1 1

= K /7 (1 - t)b'1sc'\ dsdt,

00

(t + s - ts - tz)c

where K = (1 + b - c)(c - 1). Now we find 1 1

A/» = K// ^ty dsdt =

00 1 1

OO

K / (1 - t)b-1 t—csc—2 £ ^sn dsdt =

0 0 n=0

~ S 1 1

!n 1 i\ra Wi -Ara+b-1+— (ra+c)^4 / „ra+c—2,

n!

n=0 0 0

1 1

— (c)^ ^n 1 (1 - t)-+b—1

(-1)- /(1 - t)n+b—1t—(n+c)dW sn+c—2ds

= K £ ^(-1)^ (1 -(-+c) dtl sn+c—2ds. n=0 0 0

Since n + c > 1, the first integral diverges. Hence, Ab'c(/1(z)) is unbounded.

(b) For c = 1, from (1) we have

ro ro , ^ ^ / ^ ^ bk—nak \

Ab,1(/(*)) = ^^ A-TTJz-

Ab

г=0 k=n ^fc

For /1(z) = , we find

ro ro , ro , <TO <TO ,

A/)) = £ (£ -r) z- = £ A+ra + £ (£ ^ )z-

n=0 k=n —k k=0 —k n=1 k=n —k

Hence, we have

~ = V (1 + 6 - i)^*1 = b + k - 1

fc=0 k fc=0 ^ k fc=0 which is divergent. □

Remark. It is still an open question, whether the operator Ab'c is bounded on the weighted Bergman space B^+c-1 for a + 2 > p.

Conflicts of Interests: The authors declare that there is no conflict of interests regarding the publication of this paper.

References

[1] Andrews G.E., Askey R., Roy R. Special Functions, Cambridge University Press, Cambridge, 1999.

[2] Agrawal M. R., Howlett P. G., Lucas S. K. Naik S. and Ponnusamy S. Boundedness of generalized Cesaro averaging operators on certain function spaces, J. Comput. Appl. Math., 1998, vol. 126, pp. 3553-3560.

[3] Borgohain D., Naik S. Generalized Cesaro operators on the spaces of Cauchy transforms, Acta Sci. Math.(Szeged), 2017, vol. 83, pp. 143-154.

[4] Balasubramanian R., Ponnusamy S. On Ramanujan' asymptotic expansions and inequalities for hypergeometric functions, Proc. Indian Acad. Sci. (Math. Sci.), 1998, vol. 108, no. 2, pp. 95-108.

[5] Duren P. L. Theory of Hp spaces, Academic Press, New York, 1981.

[6] Naik S. Generalized Cesaro operators on mixed norm spaces, J. Ind. Acad. Math., 2009, vol. 31, pp. 295-306.

[7] Naik S. Generalized Cesaro averaging operators on certain function spaces, Ann. Polinicci Math., 2010, vol. 98, pp. 189-199.

[8] Naik S. Cesaro type operators on spaces of analytic functions, Filomat, 2011, vol. 25, pp. 85-97.

[9] Ponnusamy S. Close-to-convexity properties of Gaussian hypergeometric functions, J. Comput. Appl. Math., 1997, vol. 88, pp. 327-337.

[10] Ponnusamy S .Hypergeometric transforms of functions with derivative in a half plane, J. Comput. Appl. Math., 1998, vol. 96, pp. 35-49.

[11] Ponnusamy S., R0nning F. Duality for Hadamard products applied to certain integral transforms, Complex Variables: Theory and Appl., 1997, vol. 32, pp. 263-287.

[12] Ponnusamy S., Vuorinen M.Asymptotic expansions and inequalities for hyperge- ometric functions, Mathematika, 1997, vol. 44, pp. 278-301.

[13] Siskakis A. Semigroups of composition operators in Bergman spaces, Bull. Austral.Math.Soc., 1987, vol. 35 , pp. 397-406.

[14] Stempak K. Cesaro averaging operators, Proc. Royal Soc. Edinburg, 1994, vol. 124 (A), pp. 121-126.

[15] SteviC S.A note on the generalized Cesaro operator on Bergman Space, Indian J. Math. 2004, vol. 46, no. 1, pp. 129-136.

[16] SteviC S. The generalized Cesaro operator on Dirichlet spaces, Studia Sci. Math. Hungar., 2003, vol. 40, pp. 83-94.

[17] Xiao J. Cesaro type operators on Hardy, BMOA and Bloch spaces, Arch. Math., 1997, vol. 68, pp. 398-406.

Received June 27, 2019. In revised form, September 26, 2019. Accepted October 01, 2019. Published online October 16, 2019.

Department of Applied Sciences Gauhati University Guwahati, Assam, India-781 014 S. Naik

E-mail: spn20@yahoo.com; P. K. Nath

E-mail: pankaj.kumar0246@gmail.com