C om mun. Fac. Sci. U niv. A nk. Ser. A 1 M ath. Stat.
Volum e 68, N umb er 1, Pages 457–465 (2019) D O I: 10.1501/C om mua1_ 0000000921 ISSN 1303–5991 E-ISN N 2618-6470
http://com munications.science.ankara.edu.tr/index.php?series= A 1
A NOTE ON STATISTICAL APPROXIMATION PROPERTIES
OF COMPLEX Q-SZÁSZ- MIRAKJAN OPERATORS
DIDEM AYDIN ARI
Abstract. The complex q Szász-Mirakjan operator attached to analytic func- tions satisfying a suitable exponential type growth condition has been studied in [14]. In this paper, we consider the A-statistical convergence of complex q-Szász- Mirakjan operator.
1. Introduction
In 1996, Phillips de…ned a generalization of the Bernstein operators called q Bernstein operators by using the q binomial coe¢ cients and the q binomial the- orem [21]. In 2008, Aral introduced q Szász-Mirakjan operators and studied some approximation properties of them [12]. In 2008, Gal studied some approximation results of the complex Favard-Szász-Mirakjan operators on compact disks [17]. A di¤erent type complex q Szász-Mirakjan operator was introduced by Mahmudov in [20] for q > 1 as
Mn;q(f ; z) = P1
k=0
f [k]
[n]
1 qk(k 1)=2
[n]kzk
[k]! "q [n] q kz (1.1) for the functions which are continuous and bounded on [0; 1).
In this paper, we study some operators by taking statistical convergence instead of ordinary convergence. In 2002, Gadjiev and Orhan gave some approximation results by using statistical convergence [1]. And several authors have studied in approximation theory by using statistical convergence concept (see [3], [4] and [5]
[6], [7], [8], [13], [19]).
Now, we give some notations on q analysis given in [16] and [21].
The q integer [n] is de…ned by
Received by the editors: December 13, 2017; Accepted: January 31, 2018.
2010 Mathematics Subject Classi…cation. 40A35, 30E10.
Key words and phrases. A-Statistical approximation, complex q-Szász- Mirakjan operators, q-exponential functions, q-derivative.
c 2 0 1 8 A n ka ra U n ive rsity C o m m u n ic a tio n s Fa c u lty o f S c ie n c e s U n ive rs ity o f A n ka ra -S e rie s A 1 M a t h e m a t ic s a n d S ta t is t ic s
457
[n] := [n]q =
1 qn 1 q; q 6= 1
n; q = 1 for q > 0 and the q factorial [n]! by
[n]! := [1]q[2]q::: [n]q; n = 1; 2; :::
1; n = 0:
We give the following two q analogues of the exponential function ex which is appeared in the de…nition of the operator :
"q(x) = X1 n=0
1
[n]q!xn= 1
((1 q)x; q)1; jxj < 1
1 q; jqj < 1; (1.2)
Eq(x) = X1 n=0
qn(n2 1)
[n]q! xn= ( (1 q)x ; q)1 ; x 2 R; jqj < 1; (1.3) where (x; q)1= Q1
k=1
(1 xqk 1) (see [15]).
It is clear from (1.2) and (1.3) that "q(x)Eq( x) = 1 and
qlim!1 "q(x) = lim
q!1 Eq(x) = ex:
Let q 2 (0; 1) [ (1; 1): The q derivative of a function f (x) is de…ned as
Dqf (x) := f (x) f (qx)
(1 q)x for; x 6= 0:
Dqf (0) = lim
x!0Dqf (x); where D0qf := f; Dqnf := Dq(Dn 1q f ); n = 1; 2; :::
As a consequence of the de…nition of Dqf; we …nd Dqxn = [n]qxn 1; Dq"q(ax) = a"q(ax);
DqEq(ax) = aEq(qax):
Also the formula for the q-di¤erential of a product is
Dq(u(x)v(x)) = Dq(u(x))v(x) + u(qx)Dq(v(x)):
We know that
Dq(t; x)nq (t) = [n]q(t; x)n 1q ; where (t; x)nq =
n 1Q
k=0
(t xqk) (see [16]).
Now we de…ne the complex Szász-Mirakjan operator based on q integers.
Suppose that Rn;q := [n](1 q)bn ; where (bn) is a sequence of positive numbers such that lim
n!1bn = 1 and that DR = fz 2 C : jzj < Rg ; 1 < R < Rn;q: The complex Szász-Mirakjan operator based on q integers is obtained directly from the real version (see [12]) by taking z in place of x, namely
Snq(f ; z) = Sn(f ; q; z) (1.4)
= : Eq [n] z bn
P1 k=0
f [k]
[n]bn
([n] z)k [k]! (bn)k;
where n 2 N; 0 < q < 1 and f : [R; 1) [ DR! C has exponential growth and it has an analytical continuation into an open disk centered at the origin.
Throughout the paper we call the operator (1.4) as complex q Szász-Mirakjan operator.
It is clear that by using divided di¤erences Snq(f ; z) can be expressed as Snq(f ; z) = Sn(f; q; z) = P1
j=0
qj(j21)f 0;bn[1]
[n] ; :::;bn[j]
[n] zj; (1.5) similar to the real version of the q Szász-Mirakjan operators (see [12]), where fh
0;bn[n][1]; :::;bn[n][j]i
denotes the divided di¤erence of f on the knots 0;bn[n][1]; :::;bn[n][j]:
2. Statistical Convergence of Snqn(f ; z)
First of all, we recall some de…nitions and notations which we use in this study.
Let A = (ajn) be a nonnegative regular matrix. The A density of K N given by
A(K) := lim
j
P1 n2K
ajn;
whenever the limit exists. A sequence x = (xn) is called A statistically convergent to a number L if for every " > 0;
A(fn 2 N : jxn Lj "g) = 0: (2.1) It is not di¢ cult to see that (2.1) is equivalent to
jlim!1
P1 n:jxn Lj "
ajn= 0; for every " > 0:
This limit expression is denoted by stA lim xn = L ( see in [2], [9],[10], [11]).
Now, we give a lemma which we use in the proof of Theorem 1.
Lemma 1. Let DR= fz 2 C : jzj < Rg ; 1 < R < Rn;q; where Rn;q = [n]bn
q(1 q) and
f : [R; 1) [ DR! C
be continuous in [R; 1) [ DR, analytic in DR, namely f (z) = P1
k=0
ckzk for all z 2 DR and there exist M; C; B > 0 and A 2 R1; 1 , with the property jckj M Ak!k
for all k = 0; 1; :::(which implies jf(z)j M eAjzj for all z 2 DRand jf(x)j CeBx for all x 2 [R; 1)): Then Snq(f ; z) is well de…ned and analytic as function of z in DR (see [14]).
Theorem 1. Suppose that the conditions of Lemma 1 are satis…ed. Suppose also that A be a nonnegative regular summability matrix, q = qn is a sequence such that 0 < qn< 1 and stA lim qn = 1 and stA lim[n]bn
qn = 0:
i. Let 1 r < B1 be arbitrary …xed. Then for all jzj r, we have stA lim jSnqn(f ; z) f (z)j = 0:
ii. For the simultaneous approximation by complex q Szász-Mirakjan operator, we have
stA lim D(p)qn (Snqn(f ; z)) D(p)qnf (z) = 0 where Cr1;A is given as in the case (i) :
Proof. i. From [14], by taking ek(z) = zk; it is clear that Tn;k(z) := Snqn(ek; z) is a polynomial of degree k, k = 0; 1; 2; ::: and
Tn;0(z) = 1; Tn;1(z) = z for all z 2 C Also, using q derivative of Tn;k(z) for z 6= 0, we get
DqTn;k(z)
= [n]qn
zbn Tn;k+1(z) [n]qn
bn
Eq [n]q
n qn z bn
P1 j=0
[j]qn
[n]qnbn
!k
[n]qnqnz j
[j]qn! (bn)j (2.2) for all z 2 C; k = 0; 1; 2; :::: Therefore, we obtain
Tn;k(z) = zTn;k 1(qnz) + zbn
[n]q
n
Dq(Tn;k 1(z)) : The last equality implies that
Tn;k(z) zk
= zbn
[n]qnDq Tn;k 1(z) zk 1 + z Tn;k 1(qnz) (qnz)k 1 +[k 1]
qn
[n]q
n
bnzk 1 zk(1 qn) [k 1]qn: (2.3)
From the Bernstein inequality in Dr= fz 2 C: jzj rg ; we have jDq(Pk(z)j kPk0k k
rkPkkr; (2.4)
where k:kr= max
z2Dr jf(z)j (see [18, p. 55]): From (2.3) and (2.4), we obtain that Tn;k(z) zk
rbn
[n]qn Tn;k 1(z) zk 1 rk 1 r
+r Tn;k 1(qnz) (qnz)k 1 + rk 1[k 1]qn
[n]qn bn+ rk[k 1]qnj1 qnj : By passing to norm we reach to
Tn;k(z) zk
r
(k 1)bn
[n]q
n
+ r
!
Tn;k 1(z) zk 1 r+ rkk 1 qn+ bn
[n]q
n
! : By using mathematical induction with respect to k; the above recurrence formula gives that
Tn;k(z) zk
r
(k + 1)!rk
2 1 qn+ bn [n]qn
!
for all k 2 and …xed an arbitrary n n0: There exists an n0 such that for all n > n0; then [n]bn
qn < 1: Assume that it is true for k: Since [k]qn (k + 1) is satis…ed for all 0 < qn< 1; the recurrence formula reduces to
Tn;k+1(z) zk+1
r
1 qn+ bn [n]qn
!rk+1 2
(
(k + 1)!k bn
[n]qn + (k + 1)! + 2(k + 1) )
for all k 2 and for all n > n0: By this inequality, it follows Tn;k+1(z) zk+1
r
(k + 2)!
2 rk+1 1 qn+ bn
[n]qn
! : for k 2 and for all n > n0:
Now, we show that
Snqn(f ; z) = P1
k=0
ckSnqn(ek; z) = P1
k=0
ckTn;k(z) (2.5) for all z 2 DR: For any m 2 N, let us de…ne
fm(z) = Pm j=0
cjzj if jzj r < R and fm(x) = f (x) if x 2 (r; 1):
From the hypothesis on f; it is clear that for any m 2 N, jfm(x)j CmeBmx for all x 2 [0; 1): Ratio test implies that for each …xed m; n 2 N and z;
jSnqn(fm; z)j Cm Eq [n]qn z bn
P1 k=0
[n]qn
k
jzjk [k]q
n! (bn)k eBm
[k]qn [n]qnbn
< 1:
Therefore, Snqn(fm; z) is well de…ned. Now, we set
fm;k(z) = ckek(z) if jzj r and fm;k(x) = f (x)
m + 1 if x 2 (r; 1):
It is clear that each fm;k is of exponential growth on [0; 1) and that fm(z) =
Pm k=0
fm;k(z):
Since Snqn is linear, we have Sqnn(fm; z) =
Pm k=0
ckSqnn(ek; z) for all jzj r;
which proves that
mlim!1Snqn(fm; z) = Snqn(f ; z) for any …xed n 2 N and jzj r: But this is immediate from
mlim!1kfm f kr= 0 and from the inequality
jSnqn(fm) Snqn(f )j Eq [n]qn z bn
"q [n]qnjzj
bn kfm f kr
Mr;nkfm f kr,
for all jzj r: Consequently the statement (2.5) is satis…ed.
In this way, from the hypothesis on ck, this implies for all jzj r jSnqn(f ; z) f (z)j
1 qn+ bn
[n]qn
!
Cr;B; (2.6)
where
Cr;B=M A 2
P1 k=2
(k + 1) (rA)k 1 is …nite for all 1 r < B1: Note that the series P1
k=2
uk+1and its derivative P1
k=2
(k + 1)uk are uniformly and absolutely convergent in any compact disk included in the open unit disk.
As stA lim qn= 1 there exists n1(") and K1 N of density 1 such that 1 qn< "
for all n 2 K1 and n > n1("): On the other hand, since stA lim[n]bn
qn = 0 there exists n2(") and K2 N of density 1 such that [n]bn
qn < " for all n 2 K2 and n > n2("): Now de…ne K = K1\ K2: Note that A(K1\ K2) = 1 and for all " > 0 and for all n > n0(") = max fn1; n2g
1 qn+ bn
[n]qn < " (2.7)
Hence (2.7) and (2.6) imply that
stA lim jSnqn(f ; z) f (z)j = 0:
ii. Let be the circle of radius r1 > r with centered 0, since for any jzj r and v 2 ; we have jv zj r1 r; by Cauchy’s formulas it follows that for all jzj r and n 2 N
D(p)qn (Snqn(f ; z)) D(p)qnf (z) Snq(p)n (f ; z) f(p)(z)
= p!
2
Z Snqn(f ; v) f (v) (v z)p+1 dv
1 qn+ bn
[n]qn
! Cr1;B
p!
2
2 r1
(r1 r)p+1
= 1 qn+ bn [n]qn
! Cr1;B
p!r1 (r1 r)p+1: Similarly we get from hypothesis that for all " > 0 there exists a subset K N of density 1 and n0= n0(") such that D(p)qn (Snqn(f ; z)) D(p)qnf (z) < " for all n > n0
and n 2 K:
Remark 1. Consider the matrix method C = (cjn) which is called Cesåro matrix and de…ned as
cjn=
1 j; 0;
n j
otherwise :
In this case A-statistical convergence reduces to statistical convergence. Now de…ne a sequence q = (qn) as
qn =
1 n;
n n+1;
n = m2 (m 2 N) otherwise :
It is obvious that q is not convergent but it is statistically convergent to 1:
Remark 2. Let dn be a sequence of positive numbers such that dn ! 1 and
dn
[n]qn ! 0 as n ! 1. Note that the sequence de…ned as bn = n;
dn;
n = m2 (m 2 N) otherwise ;
bn
[n]qn is statistically convergent to zero.
Note that these examples do not satisfy the hypothesis of Theorem 2.3 in [14], but they satisfy the hypothesis of our theorem.
Remark 3. If we take A = I identity matrix, we get ordinary convergence. There- fore when we take A = I;we have Theorem 2.3 in [14].
References
[1] Gadjiev A.D. and Orhan, C., Some approximation theorems via statistical convergence, Rocky Mt. J. Math., vol. 32, no. 1, pp. 129-138, 2002.
[2] Connor, J. S., On strong matrix summability with respect to a modulus and statistical convergence, Canad. Math. Bull. 32 (1989), 194-198.
[3] Duman O. and Orhan, C., Statistical Approximation in the space of locally integrable func- tions, Publ. Math., vol.63, no. 1-2, pp. 133-144, 2003.
[4] Duman O. and Orhan, C., Rates of a-statistical convergence of operators in the space of locally integrable functions, Appl. Math. Lett., vol.21, no.5, pp.431-435, 2008.
[5] Dirik, F., Duman O. and Demirci, K., Approximation in statistical sense to B-continuous functions by positive linear operators, Studia Scientiarum Mathematicarum Hungarica 47 (2010) 289-298.
[6] Erku¸s E. and Duman, O., A Korovkin type approximation theorem in statistical sense, Studia Sci. Math. Hungarica 43 (2006), 285-294.
[7] Sakao¼glu ·I. and Ünver, M., Statistical Approximation for multivariable integrable functions, Miskolc Math. Notes, vol.13, no.2, pp. 485-491, 2012.
[8] Duman, O., Özarslan M.A. and Do¼gru, O., On integral type generalizations of positive linear operators, Studia Math. 174 (2006), 1-12.
[9] Freedman A. R. and Sember, J. J., Densities and summability , Paci…c J. Math. 95 (1981), 293-3005.
[10] Fast, H., Sur la convergence statistique, Colloq. Math. 2 (1951), 241-244.
[11] Miller, H. I., A measure theorretical subsequence characterization of statistical convergence, Trans. Amer. Math. Soc. 347 (1995), 1811-1819.
[12] Aral, A., A generalization of Szász-Mirakjan operators based on q integer; Mathematical and Computer Modelling 47, (2008), 1052-1062.
[13] Aral A. and Duman, O., A Voronovskaya-type formula for SMK operators via statistical convergence, Mathematica Slovaca 61 (2011) 235-244
[14] Ayd¬n, D., On Complex q-Szàsz-Mirakjan Operators Commun. Fac.Sci. Univ. Ank. Series A1. Volume 61, Number 2, (2012), 51-66.
[15] Gasper, G. and Rahman, M., Basic Hypergeometric Series, Cambridge University Press, Cambridge, 1990.
[16] Ernst, T., The history of q calculus and a new method, U.U.D.M Report 2000, 16, Depart- ment of Mathematics, Upsala University.
[17] Gal, S. G., Approximation and geometric properties of complex Favard-Szász-Mirakjan op- erators in compact diks, Comput. Math. Appl., 56, (2008), 1121-1127.
[18] Gal, S. G., Approximation by Complex Bernstein and Convolution Type Operators, World Scienti…c Publishing Co, USA, 2009.
[19] Söylemez, D. and Unver, M., Korovkin Type Theorems for Cheney-Sharma Operators via Summability Methods, Results in Mathematics, Volume 72, (2017),1601-1612.
[20] Mahmudov, N. I., Approximation properties of complex q Szász-Mirakjan operators in com- pact disks. Computers & Mathematics with Applications, (2010), 1784-1791.
[21] Phillips, G. M., Interpolation and Approximation by Polynomials, Springer-Verlag, 2003.
Current address : K¬r¬kkale Universty Faculty of Arts and Sciences, Department of Mathemat- ics, Yah¸sihan,K¬r¬kkale Turkey
E-mail address : didemaydn@hotmail.com
ORCID Address: http://orcid.org/0000-0002-5527-8232