Professional Documents
Culture Documents
On Statistical Approximation Properties of Q Baskakov SZ SZ Stancu Operators
On Statistical Approximation Properties of Q Baskakov SZ SZ Stancu Operators
Original Article
a
Department of Applied Mathematics & Humanities, Sardar Vallabhbhai National Institute of Technology,
Ichchhanath Mahadev Dumas Road, Surat 395 007, Gujarat, India
b
L.1627 Awadh Puri Colony Beniganj, Phase-III, Opposite Industrial Training Institute (I.T.I.), Ayodhya Main
Road, Faizabad 224 001, Uttar Pradesh, India
c
Department of Mathematics, National Institute of Technology, Silchar 788 010, Cachar, Assam, India
Keywords Abstract In the present paper, we consider Stancu type generalization of Baskakov–Szász opera-
q-integers; tors based on the q-integers and obtain statistical and weighted statistical approximation properties
q-Baskakov–Szász– of these operators. Rates of statistical convergence by means of the modulus of continuity and the
Stancu operators; Lipschitz type maximal function are also established for operators.
Rate of statistical conver-
gence; 2010 Mathematics Subject Classification: 41A10; 41A25; 41A36
Modulus of continuity;
Lipschitz type maximal Copyright 2015, Egyptian Mathematical Society. Production and hosting by Elsevier B.V.
functions This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
∞ q/1−qn
S1110-256X(15)00053-X Copyright 2015, Egyptian Mathematical Society. Production and hosting by Elsevier B.V. This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.joems.2015.07.005
JID: JOEMS
ARTICLE IN PRESS [m;September 28, 2015;12:33]
2 V.N. Mishra et al.
where x ∈ [0, ∞) and The q-analogues of the exponential function ex (see [10]),
used here is defined as
q n + k − 1 k(k−1)/2 xk
pn,k (x) = q , (1.2) ∞
∞
k (1 + x)n+k zk
q q
Eq (z) = (1 + (1 − q)q j z) = qk(k−1)/2
and j=0 k=0
[k]q !
([n]qt) k = (1 + (1 − q)z)∞
q , |q| < 1,
sqn,k (t) = E (−[n]q t) . (1.3)
[k]q ! where (1 − x)∞ = ∞
q j=0 (1 − q j x).
In case q = 1, the above operators reduce to the Baskakov–
Szász operators [2].
Later, Mishra and Sharma [3] introduced a new Stancu type 2. Moment estimates
generalization of q-Baskakov–Szász operators, which is defined
as
q/1−qn Lemma 1. [1] The following hold:
∞
Dn(α,β ) ( f ; q; x) = [n]q pqn,k (x) q−k−1 sqn,k (t) f 1. Dn (1, q; x) = 1,
k=0 0
2. Dn (t, q; x) = x + [n]qq ,
[n]q tq−k + α 1 x
× dq t, (1.4) 3. Dn (t 2 , q; x) = 1 + x2 + (1 + q(q + 2))
[n]q + β q[n]q [n]q
q2 (1 + q)
where pqn,k (x) and sqn,k (t) are Baskakov and Szász basis func- + .
[n]2q
tion respectively, defined as above. The operators Dn(α,β ) ( f ; q; x)
in (1.4) are called q-Baskakov–Szász-Stancu operators. For α = Lemma 2 ([3]). The following hold:
0, β = 0 the operators (1.4) reduce to the operators (1.1).
In the recent years several researchers have worked on 1. Dn(α,β ) (1; q; x) = 1,
[n]q x + q + α
Stancu type generalization of different operators and they have 2. Dn(α,β ) (t; q; x) = ,
obtained various approximation properties. We mention some [n]q + β
(α,β ) 2 [n]q (q[n]q + 1) 2
of important papers as [4–8]. 3. Dn (t ; q; x) = x
Before proceeding further, we recall certain notations of q- q([n]q + β )2
calculus as follows. Such notations can be found in [9,10]. We (1 + q(q + 2))[n]q + 2α[n]q
+ x
consider q as a real number satisfying 0 < q < 1. ([n]q + β )2
For q2 (1 + q) + 2qα + α 2
n + .
1−q ([n]q + β )2
, q = 1,
[n]q = 1−q
n, q = 1,
3. Korovkin type statistical approximation properties
and
[n]q [n − 1]q [n − 2]q ...[1]q , n = 1, 2, . . . , The idea of statistical convergence goes back to the first edition
[n]q ! =
1, n = 0. (published in Warsaw in 1935) of the monograph of Zygmund
Then for q > 0 and integers n, k, k ≥ n ≥ 0, we have [11]. Formerly the concept of statistical convergence was intro-
duced by Steinhaus [12] and Fast [13] and later reintroduced by
[n + 1]q = 1 + q[n]q and [n]q + qn [k − n]q = [k]q . Schoenberg [14]. Statistical convergence, while introduced over
We observe that nearly 50 years ago, has only recently become an area of active
research. Different mathematicians studied properties of statis-
(1 + x)nq = (−x; q)n tical convergence and applied this concept in various areas.
(1 + x)(1 + qx)(1 + q2 x) · · · (1 + qn−1 x), n = 1, 2, . . . , In approximation theory, the concept of statistical conver-
=
1, n = 0. gence was used in the year 2002 by Gadjiev and Orhan [15].
They proved the Bohman–Korovkin type approximation theo-
Also, for any real number α, we have
rem for statistical convergence. It was shown that the statistical
(1 + x)∞ versions are stronger than the classical ones.
(1 + x)αq =
q
.
(1 + qα x)∞
q
Korovkin type approximation theory also has many useful
connections, other than classical approximation theory, in other
In special case, when α is a whole number, this definition coin- branches of mathematics (see Altomare and Campiti in [16]).
cides with the above definition. Let us recall the concept of a limit of a sequence extended
The q-Jackson integral and q-improper integral defined as to a statistical limit by using the natural density δ of a set K of
a
∞ positive integers:
f (x)dq x = (1 − q)a f (aqn )qn
0 n=0 δ(K ) = lim n−1 {the number k ≤ n such that k ∈ K}
n
and
∞/A n n whenever the limit exists (see [17], p. 407). So, the sequence
∞
q q x = (xk ) is said to be statistically convergent to a number L,
f (x)dq x = (1 − q)a f ,
0 n=0
A A meaning that for every > 0,
Please cite this article as: V.N. Mishra et al., On statistical approximation properties of q-Baskakov–Szász–Stancu operators,
Journal of the Egyptian Mathematical Society (2015), http://dx.doi.org/10.1016/j.joems.2015.07.005
JID: JOEMS
ARTICLE IN PRESS [m;September 28, 2015;12:33]
On statistical approximation properties of q-Baskakov–Szász–Stancu operators 3
It is denoted by st − lim xn = L. (q + α) β
n ≤δ k≤n: + x .
In [18] Doğru and Kanat defined the Kantorovich-type mod- [k]q + β [k]q + β
ification of Lupaş operators as follows:
By using (3.2), we get
n [k+1]
R̃n ( f ; q; x) = [n + 1]
[n+1]
f (t) dq t
n (q + α) β
k st − lim + x = 0.
k=0
[k]
[n+1]
n [n]q + β [n]q + β
−k k(k−1)/2 k
q q x (1 − x)(n−k) So, we have
× . (3.1)
(1 − x + qx) · · · (1 − x + qn−1 x)
(q + α) β
Doğru and Kanat [18] proved the following statistical δ k≤n: + x = 0,
[n]q + β [n]q + β
Korovkin-type approximation theorem for operators (3.1).
then
Theorem 1. Let q = (qn ), 0 < q < 1, be a sequence satisfying the
following conditions: st − lim Dn(α,β ) (t; qn ; x) − x = 0.
n
holds.
[n]q (q[n]q + 1)
≤ − 1ν 2
In [19] Doğru gave some examples so that (qn ) is statistically q([n]q + β )2
convergent to 1 but it may not be convergent to 1 in the ordinary (1 + q(q + 2))[n]q + 2α[n]q
case. + ν
([n]q + β )2
Now, we consider a sequence q = (qn ), qn ∈ (0, 1), such that 2
q (1 + q) + 2qα + α 2
+
lim qn = 1. (3.3) ([n]q + β )2
n→∞
[n]q (q[n]q + 1)
The condition (3.3) guarantees that [n]qn → ∞ as n → ∞. ≤ μ2 −1
q([n]q + β )2
Theorem 2. Let Dn(α,β ) be the sequence of the operators (1.4) and
(1 + q(q + 2))[n]q + 2α[n]q
the sequence q = (qn ) satisfies (3.2). Then for any function f ∈ +
([n]q + β )2
C[0, ν] ⊂ C[0, ∞), ν > 0, we have 2
q (1 + q) + 2qα + α 2
+
st − lim Dn(α,β ) ( f ; q; ·) − f = 0, (3.4) ([n]q + β )2
n
where C[0, ν] denotes the space of all real bounded functions f where μ2 = max{ν 2 , ν, 1} = ν 2 .
which are continuous in [0, ν]. Now, if we choose
Proof. Let fi = t i , where i = 0, 1, 2. Using Dn(α,β ) (1; qn ; x) = 1, [n]q (q[n]q + 1)
αn = − 1 ,
it is clear that q([n]q + β )2
st − lim Dn(α,β ) (1; qn ; x) − 1 = 0. (1 + q(q + 2))[n]q + 2α[n]q
βn = ,
n
([n]q + β )2
Now by Lemma (2) (ii), we have 2
q (1 + q) + 2qα + α 2
γn = ,
[n]q x + q + α ([n]q + β )2
lim Dn(α,β ) (t; qn ; x) − x =
[n] + β − x
n→∞ q now using (3.2), we can write
(q + α) β
≤ + x. st − lim αn = 0 = st − lim βn = st − lim γn . (3.6)
[n]q + β ([n]q + β ) n→∞ n→∞ n→∞
For given > 0, we define the following sets: Now for given > 0, we define the following four sets
L = {k : Dn(α,β ) (t; qk ; x) − x ≥ }, U = k : Dn(α,β ) (t 2 ; qk ; x) − x2 ≥ ,
and
U 1 = k : αk ≥ 2 ,
μ
(q + α) β
L = k: + x . (3.5)
[k]q + β [k]q + β U 2 = k : βk ≥ 2 ,
μ
It is obvious that L ⊂ L , it can be written as
U3 = k : γk ≥ 2 .
δ k ≤ n : Dn(α,β ) (t; qk ; x) − x ≥ μ
Please cite this article as: V.N. Mishra et al., On statistical approximation properties of q-Baskakov–Szász–Stancu operators,
Journal of the Egyptian Mathematical Society (2015), http://dx.doi.org/10.1016/j.joems.2015.07.005
JID: JOEMS
ARTICLE IN PRESS [m;September 28, 2015;12:33]
4 V.N. Mishra et al.
n→∞
(1 + q(q + 2))[n]q + 2α[n]q x
− lim Dn(α,β ) (t 2 ; qn ; x) − x2 + sup
n→∞ ([n]q + β ) 2
x∈[0,∞) 1 + x
2
Please cite this article as: V.N. Mishra et al., On statistical approximation properties of q-Baskakov–Szász–Stancu operators,
Journal of the Egyptian Mathematical Society (2015), http://dx.doi.org/10.1016/j.joems.2015.07.005
JID: JOEMS
ARTICLE IN PRESS [m;September 28, 2015;12:33]
On statistical approximation properties of q-Baskakov–Szász–Stancu operators 5
is satisfied.
Ṽα,E = f : sup(1 + x)α fα (x, y)
Let CB [0, ∞), the space of all bounded and continuous func-
tions on [0, ∞) and x ≥ 0. Then, for δ > 0, the modulus of 1
≤M ; x ≥ 0 and y ∈ E ,
continuity of f denoted by ω(f; δ) is defined to be (1 + y)α
ω( f ; δ) = sup | f (t) − f (x)|, t ∈ [0, ∞). where f is bounded and continuous function on [0, ∞), M is a
|t−x|≤δ positive constant and 0 < α ≤ 1.
Also, let d(x, E) be the distance between x and E, that is,
It is known that lim ω( f ; δ) = 0 for f ∈ CB [0, ∞) and also, for
δ→0
any δ > 0 and each t, x ≥ 0, we have d(x, E ) = inf{|x − y|; y ∈ E}.
|t − x| Theorem 5. If Dn(α,β ) be defined by (1.4), then for all f ∈ Ṽα,E
| f (t) − f (x)| ≤ ω( f ; δ) 1 + . (5.2)
δ α
|Dn(α,β ) ( f ; qn ; x) − f (x)| ≤ M(δn2 + d α (x, E )), (5.3)
Theorem 4. Let (qn ) be a sequence satisfying (3.2). For every
non-decreasing f ∈ CB [0, ∞), x ≥ 0 and n ∈ N, we have where
[n]q (q[n]q + 1) 2[n]q
|Dn(α,β ) ( f ; qn ; x) − f (x)| ≤ 2ω( f ; δn (x)), δn (x) = + 1 − x2
q([n]q + β )2 [n]q + β
where [n]q + q2 [n]q − 2αβ − 2qβ
+ x
[n]q (q[n]q + 1) 2[n]q ([n]q + β )2
δn (x) = + 1 − x2
q([n]q + β )2 [n]q + β q2 (1 + q) + 2qα + α 2
+ . (5.4)
[n]q + q2 [n]q − 2αβ − 2qβ ([n]q + β )2
+ x
([n]q + β )2
Proof. Let x0 ∈ Ē, where Ē denote the closure of the set E such
q2 (1 + q) + 2qα + α 2 that |x − x0 | = d (x, E ), where x ∈ [0, ∞). Then we have
+ .
([n]q + β )2
| f (t) − f (x)| ≤ | f (t) − f (x0 )| + | f (x0 ) − f (x)|.
Please cite this article as: V.N. Mishra et al., On statistical approximation properties of q-Baskakov–Szász–Stancu operators,
Journal of the Egyptian Mathematical Society (2015), http://dx.doi.org/10.1016/j.joems.2015.07.005
JID: JOEMS
ARTICLE IN PRESS [m;September 28, 2015;12:33]
6 V.N. Mishra et al.
Since Dn(α,β ) is a positive and linear operators, f ∈ Ṽα,E and us- ports of the manuscript timely. The second and third authors
ing the above inequality P. Sharma and L.N. Mishra acknowledge MHRD, New Delhi,
India for supporting this research article. The authors declare
|Dn(α,β ) ( f ; qn ; x) − f (x)| ≤ Dn(α,β ) (| f (t) − f (x0 )|; qn ; x) that there are no conflict of interests regarding the publication
+ (| f (x0 ) − f (x)|)Dn(α,β ) (1; qn ; x) of this research article.
≤M Dn(α,β ) (|t − x0 |α ; qn ; x) + |x − x0 |α Dn(α,β ) (1; qn ; x) .
(5.5) References
Please cite this article as: V.N. Mishra et al., On statistical approximation properties of q-Baskakov–Szász–Stancu operators,
Journal of the Egyptian Mathematical Society (2015), http://dx.doi.org/10.1016/j.joems.2015.07.005