Volume 2013, Issue 1 732069
Research Article
Open Access

Degree of Approximation by Hybrid Operators

Naokant Deo

Naokant Deo

Department of Applied Mathematics, Delhi Technological University (Formerly Delhi College of Engineering), Bawana Road, Delhi 110042, India dce.edu

Search for more papers by this author
Hee Sun Jung

Corresponding Author

Hee Sun Jung

Department of Mathematics Education, Sungkyunkwan University, Seoul 110-745, Republic of Korea skku.ac.kr

Search for more papers by this author
Ryozi Sakai

Ryozi Sakai

Department of Mathematics, Meijo University, Nagoya 468-8502, Japan meijo-u.ac.jp

Search for more papers by this author
First published: 20 August 2013
Citations: 1
Academic Editor: Irena Rachůnková

Abstract

We consider hybrid (Szász-beta) operators, which are a general sequence of integral type operators including beta function, and we give the degree of approximation by these Szász-beta-Durrmeyer operators.

1. Introduction

The Lupaş-Durrmeyer operators were introduced by Sahai and Prasad [1] who studied the asymptotic formula for simultaneous approximation, and many mathematicians have given different results for the Durrmeyer operators (see [26]). Now we consider here a sequence of linear positive operators, which was introduced by Gupta et al. [7] as follows. Let n and β be positive integers. For fC[0, ) satisfying ,
()
where β is a positive integer,
()
Let 0 < p. For a function f on [0, ), we define the norm by
()
Recently Jung and Sakai [8] investigated the Lupaş-Durrmeyer operators and studied the circumstances of convergence. Motivated with the idea of Jung and Sakai [8], we give the degree of approximation by Szász-Beta-Durrmeyer operators in this paper.

2. Basic Results

Lemma 1 (cf. [7]). Let α, m, n, and r be integers with m ≥ 0, r ≥ 1, and n + α > m:

()
Then one has
  • (i)

    Rn,0,r(α; x) = 1 and Rn,1,r(α; x) = ((−α + 1)x + r + 1)/(n + α − 1),

  • (ii)

    for m⩾1

    ()

  • (iii)

    ()

where gn,m,r(α; x) is a polynomial of degree ⩽m such that the coefficients of gn,m,r(α; x) are bounded independently of n.

Proof. Let Rn,m,r(x): = Rn,m,r(α; x). Then (i)

()
Using
()
we see that
()
(ii) Using , we obtain
()
Since we know that
()
we have
()
Then substituting (12) into (10), we consider the following:
()
Then since we see
()
we have
()
Here the last equation follows from integration by parts. Furthermore, we easily see
()
Therefore, we conclude
()
Consequently, (ii) is proved.

(iii) For m = 1, (6) holds. Let us assume (6) for m⩾1. We note

()
So, we have, by the assumption of induction,
()
Here, if m is even, then
()
and if m is odd, then
()
Hence we have
()
and here we see that gn,m+1,r(x) is a polynomial of degree ⩽m + 1 such that the coefficients of gn,m+1,r(x) are bounded independently of n.

Lemma 2 (cf. [7]). Let n, β, and r be integers with r ≥ 0. Let fC(r)[0, ) satisfy for a positive integer δ

()
Then one has, for n + βr > δ,
()
where
()

Proof. Using

()
we have
()

3. Main Results

Theorem 3. Let 0 < p, and let δ and r be nonnegative integers. Let n and β be integers with n + βr > δ. Let fC(r+1)[0, ) satisfy

()
Then one has uniformly, for f and n,
()

Proof. Let |tx| < ε and x < ξ < t. By the second inequality of (28),

()
Let ε = nν,   0 < ν < 1. Then using Lemma 2 and
()
we have
()
From (30) and Lemma 1, we have
()
Next, we estimate E2. By the use of the first inequality in (28), we have
()
Now using and the notation
()
we have
()
Then, with ε = nν,
()
Here for i ≥ 1, we get
()
because
()
Finally we get
()
From (32),
()
If we put ν = 1/3, then we get
()

In the following, we let ϕ(t) = 1/(1 + t),  t ∈ [0, ).

Theorem 4. Let r and γ be nonnegative integers. Let n and β be integers with n + βr > 2γ + 1. Let fC(r+2)[0, ) satisfy

()
Then one has uniformly, for f and n,
()

Proof. For fC(r+2)[0, ), we have

()
()
From (45), (46), and Lemma 2, we get
()
()
Using (1 + t) 2γC((tx) 2γ + (1 + x) 2γ), we obtain
()
Therefore, we have
()
For x ∈ [0, ), we have |gn,1,r(βr; x)|ϕ(x) ≤ C, (1 + x) 2γ|gn,2,r(βr; x)|ϕ2(γ+1)(x) ≤ C, and |gn,2(γ+1),r(βr; x)|ϕ2(γ+1)(x) ≤ C. Hence
()

Let us define the weighted modulus of smoothness by
()
where
()
()

Theorem 5. Let r and γ be nonnegative integers. Let n and β be integers with n + βr > 2γ + 2. Then one has, for fCr([0, )),

()

To prove Theorem 5, we need the following theorem.

Theorem 6. Let r and γ be nonnegative integers. Let n and β be integers with n + βr > 2γ. Let fC(r)([0, )) satisfy

()
Then one has uniformly, for n, f, and x ∈ [0, ),
()

Proof. Using (1 + y) 2γC((1 + x) 2γ + (yx) 2γ), we have

()
Therefore, by Lemma 1 (6), we have
()
Since |gn,2γ,r(βr; x)ϕ2γ(x)| is uniformly bounded on [0, ), we have with Lemma 2 and (59)
()
Therefore, we have the result.

The Steklov function [f] h(x) for fC([0, )) is defined as follows:
()
Then for the Steklov function [f] h(x) with respect to fC([0, )), we have the following properties.

Lemma 7 (see [8], Lemma 2.4.)Let f(x) ∈ C([0, )), and let η(x) be a positive and nonincreasing function on [0, ). Then

  • (i)

    [f] h(x) ∈ C2([0, )),

  • (ii)

    ()

  • (iii)

    ()

  • (iv)

    ()

Now, we prove Theorem 5.

Proof of Theorem 5. We know that, for f(x) ∈ Cr([0, )),

()
Then first, we split it as follows:
()
Then for the first term, we have, using Theorem 6, (62), and (65),
()
Here, we suppose 0 < h ⩽ 1, and then we know that
()
For the second term, from Theorem 4, (65), (63), and (64) of Lemma 7,
()
Therefore, we have
()
If we let , then
()
because .

Acknowledgment

The authors thank the referees for many kind suggestions and comments.

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.