Skip to main content
Erschienen in: Journal of Inequalities and Applications 1/2014

Open Access 01.12.2014 | Research

On the approximation for generalized Szász-Durrmeyer type operators in the space L p [ 0 , )

verfasst von: Guofen Liu, Xiuzhong Yang

Erschienen in: Journal of Inequalities and Applications | Ausgabe 1/2014

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config
download
DOWNLOAD
print
DRUCKEN
insite
SUCHEN
loading …

Abstract

In this paper we give the direct approximation theorem, the inverse theorem, and the equivalence theorem for Szász-Durrmeyer-Bézier operators in the space L p [ 0 , ) ( 1 p ) with Ditzian-Totik modulus.
MSC:41A25, 41A27, 41A36.
Hinweise

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors conceived of the study, participated its design and coordination, drafted the manuscript, participated in the sequence alignment, and read and approved the final manuscript.

1 Introduction

In 1972 Bézier [1] introduced the Bézier basic function and Bézier-type operators are the generalized types of the original operators. The introduction of these operators should have some background. Some properties of the convergence and approximation for some Bézier-type operators have been studied (cf. [26]), but there are other aspects that have not yet been considered. For more information as regards the development of the study on this topic or related field, the interested readers can consult the monograph [7] and the paper [8]. In this paper we will consider the direct, inverse and equivalence theorems for the Szász-Durrmeyer-Bézier operator, which is defined by
D n , α ( f , x ) = k = 0 n 0 s n , k ( t ) f ( t ) d t [ J n , k α ( x ) J n , k + 1 α ( x ) ] ,
(1.1)
where α 1 , f L p [ 0 , ) , J n , k ( x ) = j = k s n , j ( x ) , s n , k = e n x ( n x ) k k ! . Obviously D n , α ( f , x ) is bounded and positive in the space L p [ 0 , ) . When α = 1 , D n , α ( f , x ) is the well-known Durrmeyer operator
D n , 1 ( f , x ) = k = 0 n 0 s n , k ( t ) f ( t ) d t s n , k ( x ) .
To describe our results, we give the definitions of the first order modulus of smoothness and the K-functional (cf. [9]). For f L p [ 0 , ) ( 1 p ), φ ( x ) = x ,
ω φ ( f , t ) p = sup 0 < h t { f ( x + h φ ( x ) 2 ) f ( x h φ ( x ) 2 ) p , x h φ ( x ) 2 0 } , K φ ( f , t ) p = inf g W p { f g p + t φ g p } , K ¯ φ ( f , t ) p = inf g W p { f g p + t φ g p + t 2 g p } ,
where W p = { f | f A . C . loc , φ f p < , f p < } .
It is well known that (cf. [9])
ω φ ( f , t ) p K φ ( f , t ) p K ¯ φ ( f , t ) p ,
(1.2)
here a b means that there exists c > 0 such that c 1 a b c a .
Now we state our equivalence theorem as follows.
Theorem For f L p [ 0 , ) ( 1 p ), φ ( x ) = x , 0 < β < 1 , we have
D n , α ( f , x ) f ( x ) p = O ( ( 1 n ) β ) ω φ ( f , t ) p = O ( t β ) .
(1.3)
Throughout this paper, C denotes a constant independent of n and x, but it is not necessarily the same in different cases.

2 Direct theorem

For convenience, we list some basic properties which will be used later and can be found in [9] and [5] or obtained by simple computation:
(1)
1 = J n , 0 ( x ) > J n , 1 ( x ) > > J n , k ( x ) > > 0 ;
(2.1)
 
(2)
0 < J n , k α ( x ) J n , k + 1 α ( x ) α s n , k ( x ) , α 1 ;
(2.2)
 
(3)
s n , k ( x ) = n φ 2 ( x ) ( k n x ) s n , k ( x ) ;
(2.3)
 
(4)
s n , k ( x ) = n s n , k 1 ( x ) n s n , k ( x ) , s n , 1 ( x ) = 0 ;
(2.4)
 
(5)
J n , 0 ( x ) = 0 , J n , k ( x ) = n s n , k 1 ( x ) ( k = 1 , 2 , ) ;
(2.5)
 
(6)
D n , 1 ( ( t x ) 2 , x ) n 1 δ n 2 ( x ) ,
(2.6)
 
where δ n ( x ) = φ ( x ) + 1 n .
Now we give the direct theorem.
Theorem 2.1 For f L p [ 0 , ) ( 1 p ), φ ( x ) = x , we have
D n , α ( f , x ) f ( x ) p C ω φ ( f , 1 n ) p .
(2.7)
Proof By the definition of K ¯ φ ( f , t ) p and the relation (1.2), for fixed n, we can choose g = g n such that
f g p + 1 n φ g p + 1 n g p C ω φ ( f , 1 n ) p .
Since
D n , α f f p D n , α ( f g ) p + f g p + D n , α g g p C f g p + D n , α g g p ,
(2.8)
we only need to estimate the second term in the above relation. By the Riesz-Thorin theorem (cf. [[10], Theorem 3.6]), we separate the proof of the assertions for p = and p = 1 .
I. p = . Noting that g ( t ) = g ( x ) + x t g ( u ) d u , we write
| D n , α ( g , x ) g ( x ) | = | D n , α ( x t g ( u ) d u , x ) | δ n g D n , α ( | x t δ n 1 ( u ) d u | , x ) .
Since
| x t φ 1 ( u ) d u | = | x t 1 u d u | 2 | t x | φ ( x ) ,
(2.9)
| x t ( 1 n ) 1 d u | = n | t x | ,
(2.10)
and min { φ 1 ( x ) , n } δ n 1 ( x ) , using the Hölder inequality, we have
| D n , α ( g , x ) g ( x ) | C δ n 1 ( x ) δ n g D n , α ( | t x | , x ) C δ n 1 ( x ) δ n g ( D n , α ( ( t x ) 2 , x ) ) 1 2 .
By (2.2) and (2.6) we have
( D n , α ( ( t x ) 2 , x ) ) 1 2 ( α D n , 1 ( ( t x ) 2 , x ) ) 1 2 C n 1 2 δ n ( x ) .
Then
| D n , α ( g , x ) g ( x ) | C n δ n g .
(2.11)
Then, by (2.8) and (2.11), we get
D n , α ( f , x ) f ( x ) C ( f g + 1 n δ n g ) C ( f g + 1 n φ g + 1 n g ) C ω φ ( f , 1 n ) .
(2.12)
II. p = 1 . By (2.2) and the Fubini theorem, we have
D n , α ( g , x ) g ( x ) 1 α 0 k = 0 s n , k ( x ) n 0 s n , k ( t ) d t | x t g ( u ) d u | d x = n α 0 | g ( u ) | { u 0 u + 0 u u } k = 0 s n , k ( t ) s n , k ( x ) d t d x d u = 2 α 0 | g ( u ) | ( k = 0 n u s n , k ( t ) d t 0 u s n , k ( x ) d x ) d u = : 2 α 0 | g ( u ) | H n ( u ) d u .
Now we estimate H n ( u ) , by using 0 s n , k ( t ) d t = 1 n and k = 0 s n , k ( u ) = 1 :
H n ( u ) = n k = 0 ( 0 s n , k ( t ) d t 0 u s n , k ( x ) d x 0 u s n , k ( t ) d t 0 u s n , k ( x ) d x ) = u n k = 0 0 u s n , k ( t ) d t 0 u s n , k ( x ) d x = u n k = 0 ( t s n , k ( t ) | 0 u 0 u t s n , k ( t ) d t ) 0 u s n , k ( x ) d x = u n k = 0 u s n , k ( u ) 0 u s n , k ( x ) d x + n k = 0 0 u t s n , k ( t ) d t 0 u s n , k ( x ) d x = : u I 1 + I 2 .
Since
0 u s n , k ( x ) d x = 0 u e n x ( n x ) k k ! d x = 1 n s n , k ( u ) + 0 u s n , k 1 ( x ) d x = 1 n ( s n , k ( u ) + + s n , 1 ( u ) ) + 0 u e n x d x = 1 n ( s n , k ( u ) + + s n , 0 ( u ) ) + 1 n ,
we have
I 1 = n k = 0 u s n , k ( u ) [ 1 n j = 0 k s n , j ( u ) + 1 n ] = k = 0 u s n , k ( u ) k = 0 u s n , k ( u ) j = 0 k s n , j ( u ) = u k = 0 u s n , k ( u ) j = 0 k s n , j ( u ) .
Using the equation k + 1 n s n , k + 1 ( u ) = u s n , k ( u ) and (2.4), we have
I 2 = n k = 0 0 u t n ( s n , k 1 ( t ) s n , k ( t ) ) d t 0 u s n , k ( x ) d x = n k = 0 0 u t s n , k ( t ) d t 0 u n ( s n , k + 1 ( x ) s n , k ( x ) ) d x = n k = 0 0 u t s n , k ( t ) d t 0 u ( s n , k + 1 ( x ) ) d x = n k = 0 0 u k + 1 n s n , k + 1 ( t ) d t s n , k + 1 ( u ) = n k = 0 0 u s n , k + 1 ( t ) d t u s n , k ( u ) = n k = 0 [ 1 n ( s n , k + 1 ( u ) + + s n , 0 ( u ) ) + 1 n ] u s n , k ( u ) = k = 0 u s n , k ( u ) j = 0 k + 1 s n , j ( u ) k = 0 u s n , k ( u ) = k = 0 u s n , k ( u ) j = 0 k s n , j ( u ) + u k = 0 s n , k ( u ) s n , k + 1 ( u ) u .
So
H n ( u ) = u u + 2 u k = 0 s n , k ( u ) j = 0 k s n , j ( u ) + u k = 0 s n , k ( u ) s n , k + 1 ( u ) u .
Since
k = 0 s n , k ( u ) j = 0 k s n , j ( u ) = k = 0 s n , k ( u ) j = k s n , j ( u ) = k = 0 s n , k ( u ) j = k + 1 s n , j ( u ) + k = 0 s n , k ( u ) s n , k ( u ) ,
we can write
H n ( u ) = u ( k = 0 s n , k ( u ) j = 0 k s n , j ( u ) + k = 0 s n , k ( u ) j = k + 1 s n , j ( u ) ) + u k = 0 s n , k ( u ) ( s n , k ( u ) + s n , k + 1 ( u ) ) u = u k = 0 s n , k ( u ) ( s n , k ( u ) + s n , k + 1 ( u ) ) .
Using the result of [[5], Lemma 3]
s n , k ( u ) 1 n u ,
we get
H n ( u ) 2 u n .
Consequently
D n , α ( g , x ) g ( x ) 1 4 α 0 | g ( u ) | φ ( u ) n d u = 4 α n φ g 1 .
(2.13)
By (2.8) and (2.13) we have
D n , α ( f , x ) f ( x ) 1 C ω φ ( f , 1 n ) 1 .
(2.14)
From (2.12) and (2.14), (2.7) is obtained. □
Remark 1 In [11] we show that the second order modulus cannot be used for the Baskakov-Bézier operators. Similarly in (2.7) ω φ 2 ( f , x ) p cannot be used instead of ω φ ( f , x ) p .

3 Inverse theorem

To prove the inverse theorem, we need the following lemmas.
Lemma 3.1 For f L p [ 0 , ) ( 1 p ), φ ( x ) = x , δ n ( x ) = φ ( x ) + 1 n , we have
δ n D n , α ( f ) p C n f p .
(3.1)
Proof We will show (3.1) for the two cases of p = and p = 1 . Since
D n , α ( f , x ) = k = 0 α [ J n , k α 1 ( x ) J n , k ( x ) J n , k + 1 α 1 ( x ) J n , k + 1 ( x ) ] n 0 s n , k ( t ) f ( t ) d t = α k = 0 [ ( J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ) J n , k + 1 ( x ) + J n , k α 1 ( x ) s n , k ( x ) ] n 0 s n , k ( t ) f ( t ) d t
using 0 s n , k ( t ) d t = 1 n , we have
| D n , α ( f , x ) | α f ( k = 0 [ J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ] J n , k + 1 ( x ) + k = 0 J n , k α 1 ( x ) | s n , k ( x ) | ) = : α f ( I 1 + I 2 ) .
(3.2)
For x E n = ( 1 n , ) , δ n ( x ) φ ( x ) , by (2.1) and (2.3) we get
δ n ( x ) I 2 δ n ( x ) k = 0 | s n , k ( x ) | n δ n ( x ) φ 2 ( x ) k = 0 | k n x | s n , k ( x ) n δ n ( x ) φ 2 ( x ) ( k = 0 ( k n x ) 2 s n , k ( x ) ) 1 2 = n δ n ( x ) φ 2 ( x ) φ ( x ) n 2 n ,
(3.3)
here we used (cf. [[9], p.128, Lemma 9.4.3])
k = 0 ( k n x ) 2 s n , k ( x ) = φ 2 ( x ) n .
For x E n c = [ 0 , 1 n ] , δ n ( x ) 1 n , by (2.4) we have
δ n ( x ) I 2 δ n ( x ) k = 0 n ( s n , k 1 ( x ) + s n , k ( x ) ) 4 n .
(3.4)
By (3.3) and (3.4), we get
δ n ( x ) I 2 6 n .
(3.5)
Noting J n , 0 ( x ) = 0 , we have
I 1 = k = 0 ( J n , k α 1 ( x ) ( J n , k ( x ) s n , k ( x ) ) J n , k + 1 α 1 ( x ) J n , k + 1 ( x ) ) k = 0 J n , k α 1 ( x ) J n , k ( x ) k = 0 J n , k + 1 α 1 ( x ) J n , k + 1 ( x ) + k = 0 J n , k α 1 ( x ) | s n , k ( x ) | = I 2 ,
then
δ n ( x ) I 1 6 n .
(3.6)
Combining (3.2), (3.5), and (3.6) we get for p =
δ n D n , α ( f ) C n f .
(3.7)
For p = 1 , we have
| D n , α ( f ) | α k = 0 n 0 s n , k ( t ) | f ( t ) | d t [ ( J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ) J n , k + 1 ( x ) + J n , k α 1 ( x ) | s n , k ( x ) | ] = : α ( J ˜ 1 + J ˜ 2 ) .
(3.8)
Hence we can write
0 | δ n ( x ) D n , α ( f , x ) | d x α 0 δ n ( x ) ( J ˜ 1 + J ˜ 2 ) d x = α ( E n c + E n ) δ n ( x ) ( J ˜ 1 + J ˜ 2 ) d x .
(3.9)
Now we estimate the last part of (3.9) in four phases:
E n c δ n ( x ) J ˜ 2 d x E n c δ n ( x ) k = 1 n 0 s n , k ( t ) | f ( t ) | d t n ( s n , k 1 ( x ) + s n , k ( x ) ) d x + E n c δ n ( x ) n 0 s n , 0 ( t ) | f ( t ) | d t n s n , 0 ( x ) d x .
For x E n c , δ n ( x ) 2 n , s n , 0 ( t ) 1 , noting 0 s n , k ( x ) d x = 1 n , we have
E n c δ n ( x ) J ˜ 2 d x 4 n n f 1 + 2 n n f 1 6 n f 1 .
(3.10)
Since J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) 1 and J n , k + 1 ( x ) = n s n , k ( x ) , we have
E n c δ n ( x ) J ˜ 1 d x E n c δ n ( x ) k = 0 n 0 s n , k ( t ) | f ( t ) | d t n s n , k ( x ) d x 2 n f 1 .
(3.11)
To estimate E n δ n ( x ) J ˜ 2 d x , we will need the relation [[9], p.129, (9.4.15)]
E n ( k n x ) 2 φ 2 ( x ) s n , k ( x ) d x C n 2 .
By the Hölder inequality and (2.3), we get
E n δ n ( x ) J ˜ 2 d x 2 k = 0 n 0 s n , k ( t ) | f ( t ) | d t E n φ ( x ) n φ 2 ( x ) | k n x | s n , k ( x ) d x 2 k = 0 n 2 0 s n , k ( t ) | f ( t ) | d t ( E n ( k n x ) 2 φ 2 ( x ) s n , k ( x ) d x ) 1 2 ( E n s n , k ( x ) d x ) 1 2 C n k = 0 0 s n , k ( t ) | f ( t ) | d t = C n f 1 .
(3.12)
In order to estimate E n δ n ( x ) J ˜ 1 d x , we consider the two cases of α 2 and 1 < α < 2 (when α = 1 , J ˜ 1 = 0 ).
For α 2 , J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ( α 1 ) s n , k ( x ) . Using integration by parts, we can deduce
E n δ n ( x ) J ˜ 1 d x C k = 0 n 0 s n , k ( t ) | f ( t ) | d t E n φ ( x ) s n , k ( x ) J n , k + 1 ( x ) d x = C k = 0 n 0 s n , k ( t ) | f ( t ) | d t × ( φ ( x ) s n , k ( x ) J n , k + 1 ( x ) | 1 n 1 n J n , k + 1 ( x ) d ( φ ( x ) s n , k ( x ) ) ) = C k = 0 n 0 s n , k ( t ) | f ( t ) | d t φ ( x ) s n , k ( x ) J n , k + 1 ( x ) | 1 n C k = 0 n 0 s n , k ( t ) | f ( t ) | d t × ( 1 n J n , k + 1 ( x ) 1 2 x s n , k ( x ) d x + 1 n φ ( x ) s n , k ( x ) J n , k + 1 ( x ) d x ) .
Noting that φ ( x ) s n , k ( x ) J n , k + 1 ( x ) | 1 n < 0 and 1 n J n , k + 1 ( x ) 1 2 x s n , k ( x ) d x > 0 , and from (2.3), we have
E n δ n ( x ) J ˜ 1 d x C k = 0 n 0 s n , k ( t ) | f ( t ) | d t 1 n | φ ( x ) s n , k ( x ) J n , k + 1 ( x ) | d x C k = 0 n 0 s n , k ( t ) | f ( t ) | d t n ( E n ( k n x ) 2 φ 2 ( x ) s n , k ( x ) d x ) 1 2 ( E n s n , k ( x ) d x ) 1 2 C n f 1 .
(3.13)
For 1 < α < 2 , using the mean value theorem, we know
J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) = ( α 1 ) ( ξ k ( x ) ) α 2 s n , k ( x ) ,
where J n , k + 1 ( x ) < ξ k ( x ) < J n , k ( x ) and α 2 < 0 , then
J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ( α 1 ) J n , k + 1 α 2 ( x ) s n , k ( x ) .
For 1 < α < 2 , we get from the procedure of (3.13)
E n δ n ( x ) J ˜ 1 d x k = 0 n 0 s n , k ( t ) | f ( t ) | d t E n ( α 1 ) φ ( x ) s n , k ( x ) J n , k + 1 α 2 ( x ) J n , k + 1 ( x ) d x = k = 0 n 0 s n , k ( t ) | f ( t ) | d t E n φ ( x ) s n , k ( x ) d J n , k + 1 α 1 ( x ) k = 0 n 0 s n , k ( t ) | f ( t ) | d t E n φ ( x ) | s n , k ( x ) | d x C n f 1 .
(3.14)
Combining (3.13) and (3.14), we get for α 1
E n δ n ( x ) J ˜ 1 d x C n f 1 .
(3.15)
From (3.8)-(3.12) and (3.15), we obtain
δ n D n , α ( f ) 1 C n f 1 .
(3.16)
By (3.7) and (3.16), Lemma 3.1 holds. □
Lemma 3.2 For f W p , φ ( x ) = x , δ n ( x ) = φ ( x ) + 1 n , we have
δ n D n , α ( f ) p C δ n f p .
(3.17)
Proof By the Riesz-Thorin theorem, we shall prove Lemma 3.2 for p = and p = 1 . For f W p and noting that D n , α ( 1 , x ) = 0 , we have
k = 0 n 0 f ( x ) s n , k ( t ) d t ( J n , k α ( x ) J n , k + 1 α ( x ) ) = 0 .
Then
| D n , α ( f , x ) | = | k = 0 n 0 ( f ( t ) f ( x ) ) s n , k ( t ) d t ( J n , k α ( x ) J n , k + 1 α ( x ) ) | = | k = 0 n 0 x t f ( u ) d u s n , k ( t ) d t ( J n , k α ( x ) J n , k + 1 α ( x ) ) | α k = 0 n 0 s n , k ( t ) | x t f ( u ) d u | d t × { [ J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ] J n , k + 1 ( x ) + J n , k α 1 ( x ) | s n , k ( x ) | } .
(3.18)
By (2.9) and (2.10) we have
| x t δ n ( u ) d u | C δ n 1 ( x ) | t x | ,
hence
| δ n ( x ) D n , α ( f , x ) | C δ n f k = 0 n 0 s n , k ( t ) | t x | d t × { [ J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ] J n , k + 1 ( x ) + J n , k α 1 ( x ) | s n , k ( x ) | } = : C δ n f ( J 1 + J 2 ) .
(3.19)
For x E n c , δ n ( x ) 1 n and by (2.1) and (2.4) we have
J 2 = k = 0 n 0 s n , k ( t ) | t x | d t J n , k α 1 ( x ) | s n , k ( x ) | n k = 0 n 0 s n , k ( t ) | t x | d t ( s n , k 1 ( x ) + s n , k ( x ) ) = n k = 1 n 0 s n , k ( t ) | t x | d t s n , k 1 ( x ) + n k = 0 n 0 s n , k ( t ) | t x | d t s n , k ( x ) = : K 1 + K 2 .
By (2.6) we get
K 2 = n D n , 1 ( | t x | , x ) n ( D n , 1 ( ( t x ) 2 , x ) ) 1 2 n δ n ( x ) 2 .
For K 1 , we write
K 1 = n k = 2 n 0 s n , k ( t ) | t x | d t s n , k 1 ( x ) + n 2 0 s n , 1 ( t ) | t x | d t s n , 0 ( x ) .
First, using (2.6), 0 s n , k ( t ) d t = 1 n , and the Hölder inequality, we have
n k = 2 n 0 s n , k ( t ) | t x | d t s n , k 1 ( x ) = n k = 2 n 0 s n , k ( t ) | t x | d t s n , k ( x ) k n x n ( k = 2 n 2 ( 0 s n , k ( t ) | t x | d t ) 2 s n , k ( x ) ) 1 2 ( k = 2 s n , k ( x ) k 2 ( n x ) 2 ) 1 2 n ( k = 2 n 2 0 s n , k ( t ) d t 0 s n , k ( t ) ( t x ) 2 d t s n , k ( x ) ) 1 2 ( k = 2 s n , k ( x ) k 2 ( n x ) 2 ) 1 2 C n ( k = 2 n 0 s n , k ( t ) ( t x ) 2 d t s n , k ( x ) ) 1 2 ( k = 2 s n , k 2 ( x ) ) 1 2 C n δ n ( x ) n C .
Next, for x E n c , e n x 1 , and 0 s n , k ( t ) d t = 1 n , we have
n 2 0 s n , 1 ( t ) | t x | d t s n , 0 ( x ) n 2 0 s n , 1 ( t ) ( t + x ) d t e n x n 2 ( 2 n 0 s n , 2 ( t ) d t + x 0 s n , 1 ( t ) d t ) = n 2 ( 2 n 2 + x n ) 3 .
Then we get
J 2 C , x E n c .
(3.20)
Noting that J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) 1 , by (2.5) we have
J 1 K 2 2 .
(3.21)
For x E n , δ n ( x ) φ ( x ) , using (2.3), 0 s n , k ( t ) d t = 1 n and the Hölder inequality, we have
J 2 k = 0 n 0 s n , k ( t ) | t x | d t n φ 2 ( x ) | k n x | s n , k ( x ) ( k = 0 n 0 s n , k ( t ) ( t x ) 2 d t s n , k ( x ) ) 1 2 ( k = 0 ( k n x ) 2 s n , k ( x ) ) 1 2 n φ 2 ( x ) δ n ( x ) n φ ( x ) n n φ 2 ( x ) 2 .
Noting that J n , 0 ( x ) = 0 , one has
J 1 = k = 0 n 0 s n , k ( t ) | t x | d t [ J n , k α 1 ( x ) J n , k + 1 α 1 ( x ) ] J n , k + 1 ( x ) k = 1 n 0 s n , k ( t ) | t x | d t J n , k α 1 ( x ) J n , k ( x ) k = 0 n 0 s n , k ( t ) | t x | d t J n , k + 1 α 1 ( x ) J n , k + 1 ( x ) + k = 0 n 0 s n , k ( t ) | t x | d t J n , k α 1 ( x ) | s n , k ( x ) | .
The third term of the above is J 2 , J 3 denotes the difference of the front two terms, and we need only to consider J 3 . By (2.1), (2.4), (2.5), and integration by parts, we have
J 3 k = 1 n 0 ( s n , k ( t ) s n , k 1 ( t ) ) | t x | d t J n , k α 1 ( x ) J n , k ( x ) k = 1 n | 0 1 n s n , k ( t ) | t x | d t | n s n , k 1 ( x ) k = 1 ( | 0 x s n , k ( t ) ( x t ) d t | + | x s n , k ( t ) ( t x ) d t | ) n s n , k 1 ( x ) = k = 1 ( | 0 x ( x t ) d s n , k ( t ) | + | x ( t x ) d s n , k ( t ) | ) n s n , k 1 ( x ) k = 1 0 s n , k ( t ) d t n s n , k 1 ( x ) = 1 .
Thus
J 1 J 2 + J 3 3 .
(3.22)
So we get
δ n D n , α ( f ) C δ n f .
(3.23)
For p = 1 , using (2.1), (2.4), (2.5), and integration by parts, we have
D n , α ( f , x ) = α k = 1 n 0 f ( t ) s n , k ( t ) d t J n , k α 1 ( x ) J n , k ( x ) α k = 0 n 0 f ( t ) s n , k ( t ) d t J n , k + 1 α 1 ( x ) J n , k + 1 ( x ) = α k = 1 n 0 f ( t ) ( s n , k ( t ) s n , k 1 ( t ) ) d t J n , k α 1 ( x ) J n , k ( x ) = α k = 1 0 f ( t ) s n , k ( t ) d t J n , k α 1 ( x ) J n , k ( x ) = α k = 1 0 f ( t ) s n , k ( t ) d t J n , k α 1 ( x ) J n , k ( x ) α k = 1 0 f ( t ) s n , k ( t ) d t n s n , k 1 ( x ) .
Let
δ n D n , α ( f ) 1 = δ n D n , α ( f ) 1 E n c + δ n D n , α ( f ) 1 E n = : K 1 ˜ + K 2 ˜ .
(3.24)
For x E n c , noting that n δ n ( t ) 1 and n δ n ( x ) 2 , we have
K 1 ˜ α E n c δ n ( x ) k = 1 0 | f ( t ) | s n , k ( t ) d t n s n , k 1 ( x ) d x α E n c n δ n ( x ) k = 1 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t n s n , k 1 ( x ) d x 2 α k = 1 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t n E n c s n , k 1 ( x ) d x 2 α δ n f 1 .
(3.25)
For x E n , we estimate K 2 ˜
K 2 ˜ 2 α E n φ ( x ) k = 1 0 | δ n ( t ) f ( t ) | φ ( t ) s n , k ( t ) d t n s n , k 1 ( x ) d x = 2 n α k = 1 0 | δ n ( t ) f ( t ) | φ ( t ) s n , k ( t ) d t E n φ ( x ) s n , k 1 ( x ) d x .
Using the Hölder inequality and 2 a b a + b ( a , b > 0 ), we have
0 | δ n ( t ) f ( t ) | φ ( t ) s n , k ( t ) d t ( 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t 0 | δ n ( t ) f ( t ) | φ 2 ( t ) s n , k ( t ) d t ) 1 2 = ( 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t 0 | δ n ( t ) f ( t ) | n k s n , k 1 ( t ) d t ) 1 2 ( n k ) 1 2 ( 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t + 0 | δ n ( t ) f ( t ) | s n , k 1 ( t ) d t )
and
E n φ ( x ) s n , k 1 ( x ) d x 1 n ( E n φ 2 ( x ) s n , k 1 ( x ) d x ) 1 2 = 1 n ( k n ) 1 2 ( E n s n , k ( x ) d x ) 1 2 1 n ( k n ) 1 2 .
Therefore we have
K 2 ˜ 2 α k = 1 ( 0 | δ n ( t ) f ( t ) | s n , k ( t ) d t + 0 | δ n ( t ) f ( t ) | s n , k 1 ( t ) d t ) 2 α δ n f 1 .
(3.26)
By (3.24)-(3.26) we have
δ n D n , α ( f ) 1 C δ n f 1 .
(3.27)
By (3.23) and (3.27), Lemma 3.2 holds. □
Using Lemmas 3.1 and 3.2, we can prove the inverse theorem.
Theorem 3.3 For f L p [ 0 , ) ( 1 p ), φ ( x ) = x , 0 < β < 1 ,
D n , α ( f , x ) f ( x ) p = O ( n β 2 )
implies ω φ ( f , t ) p = O ( t β ) .
Proof Using Lemmas 3.1 and 3.2, for a suitable function g, we have
K φ ( f , t ) p f D n , α ( f ) p + t δ n D n , α ( f ) p C n β 2 + t ( δ n D n , α ( f g ) p + δ n D n , α ( g ) p ) C n β 2 + C t ( n f g p + δ n g p ) C n β 2 + C t n ( f g p + 1 n φ g p + 1 n g p ) C ( n β 2 + t n 1 2 K ¯ φ ( f , n 1 2 ) p ) C ( n β 2 + t n 1 2 K φ ( f , n 1 2 ) p )
which by the Berens-Lorentz lemma (cf. [[9], Lemma 9.3.4]) implies that
K φ ( f , t ) p = O ( t β ) .
(3.28)
From (1.2) and (3.28), we see that the proof of Theorem 3.3 is completed. □

Acknowledgements

Xiuzhong Yang was supported by the National Natural Science Foundation of China (grant No. 11371119) and both authors were supported by the Natural Science Foundation of Education Department of Hebei Province (grant No. Z2014031). The authors express their thanks to the referees for their constructive suggestions in improving the quality of the paper.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://​creativecommons.​org/​licenses/​by/​4.​0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors conceived of the study, participated its design and coordination, drafted the manuscript, participated in the sequence alignment, and read and approved the final manuscript.
Literatur
1.
Zurück zum Zitat Bézier P: Numerical Control: Mathematics and Applications. Wiley, London; 1972.MATH Bézier P: Numerical Control: Mathematics and Applications. Wiley, London; 1972.MATH
2.
Zurück zum Zitat Chang G: Generalized Bernstein-Bézier polynomials. J. Comput. Math. 1983, 4: 322–327.MATH Chang G: Generalized Bernstein-Bézier polynomials. J. Comput. Math. 1983, 4: 322–327.MATH
3.
Zurück zum Zitat Liu Z: Approximation of continuous functions by the generalized Bernstein-Bézier polynomials. Approx. Theory Appl. 1986,4(2):105–130.MATH Liu Z: Approximation of continuous functions by the generalized Bernstein-Bézier polynomials. Approx. Theory Appl. 1986,4(2):105–130.MATH
4.
Zurück zum Zitat Zeng X, Piriou A: On the rate of convergence of two Bernstein-Bézier type operators for bounded variation functions. J. Approx. Theory 1998, 95: 369–387. 10.1006/jath.1997.3227MathSciNetCrossRefMATH Zeng X, Piriou A: On the rate of convergence of two Bernstein-Bézier type operators for bounded variation functions. J. Approx. Theory 1998, 95: 369–387. 10.1006/jath.1997.3227MathSciNetCrossRefMATH
5.
Zurück zum Zitat Zeng X: On the rate of the convergence of the generalized Szász type operators for functions of bounded variation. J. Math. Anal. Appl. 1998, 226: 309–325. 10.1006/jmaa.1998.6063MathSciNetCrossRefMATH Zeng X: On the rate of the convergence of the generalized Szász type operators for functions of bounded variation. J. Math. Anal. Appl. 1998, 226: 309–325. 10.1006/jmaa.1998.6063MathSciNetCrossRefMATH
6.
Zurück zum Zitat Zeng X, Gupta V: Rate of convergence of Baskakov-Bézier type operators for locally bounded functions. Comput. Math. Appl. 2002, 44: 1445–1453. 10.1016/S0898-1221(02)00269-9MathSciNetCrossRefMATH Zeng X, Gupta V: Rate of convergence of Baskakov-Bézier type operators for locally bounded functions. Comput. Math. Appl. 2002, 44: 1445–1453. 10.1016/S0898-1221(02)00269-9MathSciNetCrossRefMATH
7.
Zurück zum Zitat Gupta V, Agarwal RP: Convergence Estimates in Approximation Theory. Springer, Cham; 2014. xiv+361 pp.CrossRefMATH Gupta V, Agarwal RP: Convergence Estimates in Approximation Theory. Springer, Cham; 2014. xiv+361 pp.CrossRefMATH
8.
Zurück zum Zitat Verma DK, Gupta V, Agrawal PN: Some approximation properties of Baskakov-Durrmeyer-Stancu operators. Appl. Math. Comput. 2012,218(11):6549–6556. 10.1016/j.amc.2011.12.031MathSciNetCrossRefMATH Verma DK, Gupta V, Agrawal PN: Some approximation properties of Baskakov-Durrmeyer-Stancu operators. Appl. Math. Comput. 2012,218(11):6549–6556. 10.1016/j.amc.2011.12.031MathSciNetCrossRefMATH
10.
Zurück zum Zitat Bennett C, Sharpley R Pure and Applied Mathematics 129. In Interpolation of Operator. Academic Press, New York; 1988. Bennett C, Sharpley R Pure and Applied Mathematics 129. In Interpolation of Operator. Academic Press, New York; 1988.
11.
Zurück zum Zitat Guo S, Qi Q, Liu G: The central approximation theorems for Baskakov-Bézier operators. J. Approx. Theory 2007, 147: 112–124. 10.1016/j.jat.2005.02.010MathSciNetCrossRefMATH Guo S, Qi Q, Liu G: The central approximation theorems for Baskakov-Bézier operators. J. Approx. Theory 2007, 147: 112–124. 10.1016/j.jat.2005.02.010MathSciNetCrossRefMATH
Metadaten
Titel
On the approximation for generalized Szász-Durrmeyer type operators in the space
verfasst von
Guofen Liu
Xiuzhong Yang
Publikationsdatum
01.12.2014
Verlag
Springer International Publishing
Erschienen in
Journal of Inequalities and Applications / Ausgabe 1/2014
Elektronische ISSN: 1029-242X
DOI
https://doi.org/10.1186/1029-242X-2014-447

Weitere Artikel der Ausgabe 1/2014

Journal of Inequalities and Applications 1/2014 Zur Ausgabe

Premium Partner