Open Access

Nabla discrete fractional Grüss type inequality

  • A Feza Güvenilir1,
  • Billur Kaymakçalan2Email author,
  • Allan C Peterson3 and
  • Kenan Taş2
Journal of Inequalities and Applications20142014:86

https://doi.org/10.1186/1029-242X-2014-86

Received: 20 February 2013

Accepted: 24 January 2014

Published: 20 February 2014

Abstract

Properties of the discrete fractional calculus in the sense of a backward difference are introduced and developed. Here, we prove a more general version of the Grüss type inequality for the nabla fractional case. An example of our main result is given.

MSC: 39A12, 34A25, 26A33, 26D15, 26D20.

Keywords

nabla discrete fractional calculus nabla discrete Grüss inequality

1 Introduction

The Grüss inequality is of great interest in differential and difference equations as well as many other areas of mathematics [15]. The classical inequality was proved by Grüss in 1935 [3]: if f and g are continuous functions on [ a , b ] satisfying φ f ( t ) Φ and γ g ( t ) Γ for all t [ a , b ] , then
| 1 b a a b f ( t ) g ( t ) d t 1 ( b a ) 2 a b f ( t ) d t a b g ( t ) d t | 1 4 ( Φ φ ) ( Γ γ ) .
(1)

The literature on Grüss type inequalities is extensive, and many extensions of the classical inequality (1) have been intensively studied by many authors in the 21st century [612]. Here we are interested in Grüss type inequalities in the nabla fractional calculus case.

We begin with basic definitions and notation from the nabla calculus that are used in this paper. The delta calculus analog has been studied in detail, and a general overview is given in [9]. The domains used in this paper are denoted by N a where a R . This is a discrete time scale with a graininess of 1, so it is defined as
N a : = { a , a + 1 , a + 2 , } .
For N a , we use the terminology that a function’s domain, in the case studied here, is based at a. We use the ρ-function, or backward jump function, from the time scale as ρ : N a N a , given by ρ ( t ) : = max { a , t 1 } . We define the backward difference operator, or the nabla operator ( ) , for a function f : N a R by
( f ) ( t ) : = f ( t ) f ( ρ ( t ) ) = f ( t ) f ( t 1 ) .
In this paper, we use the convention that
f ( t ) = ( f ) ( t ) .
We then define higher order differences recursively by
n f ( t ) : = ( n 1 f ( t ) )

for t N a + n where n N . We take as convention that 0 is the identity operator.

Based on these preliminary definitions, we say F is an anti-nabla difference of f on N a if and only if F ( t ) = f ( t ) for t N a + 1 . We then define the definite nabla integral of f : N a R by
c d f ( t ) t : = { t = c + 1 d f ( t ) , if  c < d , 0 , if  c = d , t = d + 1 c f ( t ) , if  d < c ,

where c , d N a . We now give the fundamental theorem of nabla calculus.

Theorem 1.1 (Fundamental theorem of nabla calculus)

Let f : N a R and let F be an anti-nabla difference of f on N a , then for any c , d N a we have
c d f ( t ) t = F ( d ) F ( c ) .
The nabla product rule for two functions u , ν : N a R and t N a + 1 is given by
( u ( t ) ν ( t ) ) = u ( t ) v ( t ) + v ( ρ ( t ) ) u ( t ) .
This immediately leads to the summation by parts formula for the nabla calculus:
s = b + 1 c u ( t ) v ( t ) = u ( t ) v ( t ) | b c s = b + 1 c v ( ρ ( t ) ) u ( t ) .

Discrete fractional initial value problems have been intensively studied by many authors. Some of the very recent results are in [1315].

Now that we have established the basic definitions of the nabla calculus, we will move on to extending these definitions to the fractional case and establish definitions for the fractional sum and fractional difference which are analogues to the continuous fractional derivative.

In order to do this, we remind the reader of the rising factorial function. For n , t N , the rising factorial function is defined by
t n ¯ : = t ( t + 1 ) ( t + n 1 ) = ( t + n 1 ) ! ( t 1 ) ! .

This definition can be extended for fractional values using the gamma function as follows.

Definition 1 (Rising function)

For t , α R , the rising function is defined by
t α ¯ : = Γ ( t + α ) Γ ( t ) .

To motivate the definition of a fractional sum, we look at the definition of the integral sum derived from the repeated summation rule.

Definition 2 For any given positive real number α, the (nabla) left fractional sum of order α > 0 is defined by
a α f ( t ) = 1 Γ ( α ) s = a + 1 t ( t ρ ( s ) ) α 1 ¯ f ( s ) ,
where t { a + 1 , a + 2 , } , and
a α f ( t ) = 1 Γ ( α ) s = a t ( t ρ ( s ) ) α 1 ¯ f ( s ) ,

where t { a , a + 1 , a + 2 , } .

Theorem 1.2 ([16])

Let f be a real-valued function defined on N a and α , β > 0 . Then
a α a β f ( t ) = a ( α + β ) f ( t ) = a β a α f ( t ) .

Definition 3 (Fractional Caputo like nabla difference)

For μ > 0 , m 1 < μ < m , m = μ (where is the ceiling function), m N , v = m μ , we have the following:
a μ f ( t ) = a v ( m f ( t ) ) , t N a .

We also will use the following discrete Taylor formula.

Theorem 1.3 ([17])

Let f : Z R be a function and let a Z . Then, for all t Z with t N a + m we have the representation
f ( t ) = k = 0 m 1 ( t a ) k ¯ k ! k f ( a ) + 1 ( m 1 ) ! τ = a + 1 t ( t τ + 1 ) m 1 ¯ m f ( τ ) .
(2)

The following discrete backward fractional Taylor formula will be useful.

Theorem 1.4 ([18])

Let f : Z R be a function and let a Z . Here m 1 < μ < m , m = μ , μ > 0 . Then for all t N a + m we have the representation
f ( t ) = k = 0 m 1 ( t a ) k ¯ k ! k f ( a ) + 1 Γ ( μ ) τ = a + 1 t ( t τ + 1 ) μ 1 ¯ ( a + 1 ) μ f ( τ ) .
(3)

Corollary 1.1 (To Theorem 1.4)

Additionally assume that k f ( a ) = 0 , for k = 0 , 1 , 2 , , m 1 . Then
f ( t ) = 1 Γ ( μ ) τ = a + 1 t ( t τ + 1 ) μ 1 ¯ ( a + 1 ) μ f ( τ ) , t a + m .
(4)

Also, we have the following.

Theorem 1.5 ([18])

Let p N , ν > p , a N . Then
p ( a ν f ( t ) ) = a ( ν p ) f ( t )
(5)

for t N a .

Now, we give the following discrete backward fractional extended Taylor formula.

Theorem 1.6 ([18])

Let f : Z R be a function and let a Z + . Here m 1 < μ < m , m = μ , μ > 0 . Consider p N : μ > p . Then for all t a + m , t N , we have the representation
p f ( t ) = k = p m 1 ( t a ) k p ¯ ( k p ) ! k f ( a ) + 1 Γ ( μ p ) τ = a + 1 t ( t τ + 1 ) μ p 1 ¯ ( a + 1 ) μ f ( τ ) .
(6)

Corollary 1.2 (To Theorem 1.6)

Additionally assume that k f ( a ) = 0 , for k = p , , m 1 . Then
p f ( t ) = 1 Γ ( μ p ) τ = a + 1 t ( t τ + 1 ) μ p 1 ¯ ( a + 1 ) μ f ( τ ) , t a + m , t N .
(7)

Remark 1.1 Let f be defined on { a m + 1 , a m + 2 , , b } , where b a is an integer. Then (3) and (6) are valid only for t { a + m , , b } . Here we must assume that a + m < b .

2 Main results

We present the following discrete nabla Grüss type inequality.

Theorem 2.1 Let m 1 < μ < m , m = μ non-integer, μ > 0 ; p , a Z + with μ > p . Consider b N such that a + m < b . Let f and g two real-valued functions defined on { a m + 1 , a m + 2 , , b } . Here j { a + m , , b } . Assume that k f ( a ) = k g ( a ) = 0 , for k = p + 1 , , m 1 and
m 1 ( a + 1 ) μ f ( s ) M 1 , m 2 ( a + 1 ) μ g ( s ) M 2
for s { a + 1 , , b } , then
1 b a m j = a + m + 1 b [ p f ( j ) p g ( j ) ] 1 ( b a m ) 2 ( j = a + m + 1 b p f ( j ) ) ( j = a + m + 1 b p g ( j ) ) < M 1 M 2 C 1 m 1 m 2 C 2 ( b a m ) 2 [ Γ ( μ p + 1 ) ] 2 ,
where m 1 , m 2 , M 1 , and M 2 are positive constants, and
C 1 : = [ Q ( b a m ) ( b a m + 4 ) 1 2 + ( b a m ) 2 [ 2 μ p ¯ ] 2 ] , C 2 : = [ ( b a + 1 ) μ p + 1 ¯ ( m ) μ p + 1 ¯ μ p + 1 ( b a m ) Γ ( μ p + 2 ) ] 2 , Q : = ( s = m + 3 b a + 2 [ ( s ) μ p ¯ ] 4 ) 1 2 .
Proof By Theorem 1.6, we have
p f ( j ) = p f ( a ) + 1 Γ ( μ p ) [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ] , p g ( j ) = p g ( a ) + 1 Γ ( μ p ) [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ]

for j { a + m + 1 , a + m + 2 , , b } .

We get
( p f ( j ) ) ( p g ( j ) ) = 1 [ Γ ( μ p ) ] 2 [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( a + 1 ) μ f ( τ ) ] [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( a + 1 ) μ g ( τ ) ] + p f ( a ) Γ ( μ p ) [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( a + 1 ) μ g ( τ ) ] + p g ( a ) Γ ( μ p ) [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( a + 1 ) μ f ( τ ) ] + ( p f ( a ) ) ( p g ( a ) ) .
If we take the sum from a + m + 1 to b we get
j = a + m + 1 b [ ( p f ( j ) ) ( p g ( j ) ) ] = ( p f ( a ) ) ( p g ( a ) ) + p f ( a ) Γ ( μ p ) { j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ] } + p g ( a ) Γ ( μ p ) { j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ] } + 1 [ Γ ( μ p ) ] 2 j = a + m + 1 b { [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ] [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ] } .
Then
1 b a m j = a + m + 1 b [ ( p f ( j ) ) ( p g ( j ) ) ] = 1 b a m ( p f ( a ) ) ( p g ( a ) ) + 1 ( b a m ) [ Γ ( μ p ) ] 2 j = a + m + 1 b { [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ] [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ] } + p f ( a ) ( b a m ) Γ ( μ p ) { j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ] } + p g ( a ) ( b a m ) Γ ( μ p ) { j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ] } .
(8)
On the other hand,
1 b a m [ j = a + m + 1 b ( p f ( j ) ) ] = p f ( a ) + 1 ( b a m ) Γ ( μ p ) j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) )
and
1 b a m [ j = a + m + 1 b ( p g ( j ) ) ] = p g ( a ) + 1 ( b a m ) Γ ( μ p ) j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ) .
Multiplying the above two terms, we get
1 ( b a m ) 2 [ j = a + m + 1 b ( p f ( j ) ) ] [ j = a + m + 1 b ( p g ( j ) ) ] = ( p f ( a ) ) ( p g ( a ) ) + 1 ( b a m ) 2 [ Γ ( μ p ) ] 2 [ j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ) ] [ j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ) ] + ( p f ( a ) ) ( b a m ) Γ ( μ p ) [ j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ g ( τ ) ) ) ] + ( p g ( a ) ) ( b a m ) Γ ( μ p ) [ j = a + m + 1 b ( τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ( ( a + 1 ) μ f ( τ ) ) ) ] .
(9)
Using (8) and (9), we obtain
1 ( b a m ) j = a + m + 1 b [ ( p f ( j ) ) ( p g ( j ) ) ] 1 ( b a m ) 2 [ j = a + m + 1 b ( p f ( j ) ) ] [ j = a + m + 1 b ( p g ( j ) ) ] M 1 M 2 ( b a m ) [ Γ ( μ p ) ] 2 j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] m 1 m 2 ( b a m ) 2 [ Γ ( μ p ) ] 2 [ j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] ] [ j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] ] = M 1 M 2 ( b a m ) [ Γ ( μ p ) ] 2 j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] 2 m 1 m 2 ( b a m ) 2 [ Γ ( μ p ) ] 2 [ j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] ] 2 .
Next, in order to calculate the last two sums, we first observe that
τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ = a j ( j τ + 1 ) μ p 1 ¯ τ = 1 μ p [ Γ ( j a + 2 + μ p ) Γ ( j a + 2 ) Γ ( μ p + 2 ) ] .
Therefore,
j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] = 1 μ p j = a + m + 1 b [ Γ ( j a + 2 + μ p ) Γ ( j a + 2 ) Γ ( μ p + 2 ) ] = 1 μ p [ Γ ( m + 3 + μ p ) Γ ( m + 3 ) + + Γ ( b a + 2 + μ p ) Γ ( b a + 2 ) ] b a m μ p Γ ( μ p + 2 ) = 1 μ p [ ( m + 3 ) μ p ¯ + ( m + 4 ) μ p ¯ + + ( b a + 2 ) μ p ¯ ] b a m μ p Γ ( μ p + 2 ) = 1 μ p s = m + 3 b a + 2 ( s ) μ p ¯ b a m μ p Γ ( μ p + 2 ) = 1 μ p m + 2 b a + 2 τ μ p ¯ τ b a m μ p Γ ( μ p + 2 ) = 1 μ p [ ( b a + 1 ) μ p + 1 ¯ ( m + 1 ) μ p + 1 ¯ μ p + 1 ] b a m μ p Γ ( μ p + 2 ) .
Similarly, we get
j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] 2 1 ( μ p ) 2 [ j = a + m + 1 b ( j τ + 2 ) μ p ¯ ] 2 + 1 ( μ p ) 2 [ 2 μ p ¯ ] 2 ( b a m ) ,
which by use of Hölder’s inequality transforms into
j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] 2 ( b a m + 4 ) 1 / 2 ( μ p ) 2 ( s = m + 3 b a + 2 [ ( s ) μ p ¯ ] 4 ) 1 / 2 + b a m ( μ p ) 2 [ 2 μ p ¯ ] 2 .
By hypothesis,
j = a + m + 1 b [ τ = a + 1 j ( j τ + 1 ) μ p 1 ¯ ] 2 Q ( b a m + 4 ) 1 / 2 ( μ p ) 2 + ( b a m ) [ 2 μ p ¯ ] 2 ( μ p ) 2 .
Consequently, we get
1 b a m j = a + m + 1 b [ ( p f ( j ) ) ( p g ( j ) ) ] 1 ( b a m ) 2 [ j = a + m + 1 b p f ( j ) ] [ j = a + m + 1 b p g ( j ) ] M 1 M 2 C 1 m 1 m 2 C 2 ( b a m ) 2 [ Γ ( μ p + 1 ) ] 2 .

 □

Example Let μ = 3.5 , m = 4 , p = 2 , b = 7 , a = 1 in Theorem 2.1. Define
f ( t ) = g ( t ) = ( t 1 ) 4 ¯ ,
where t { 2 , 1 , 0 , 1 , , 7 } . Here j { 5 , 6 , 7 } . Using Theorem 2.1, we obtain
1 2 t = 6 7 ( 2 ( t 1 ) 4 ¯ ) ( 2 ( t 1 ) 4 ¯ ) 1 4 [ t = 6 7 ( 2 ( t 1 ) 4 ¯ ) ] [ t = 6 7 ( 2 ( t 1 ) 4 ¯ ) ] M 1 M 2 C 1 m 1 m 2 C 2 4 [ Γ ( 2.5 ) ] 2 ,

where M 1 = 36 , M 2 = 37 , m 1 = 15 , m 2 = 20 and C 1 = 3 , 362 , C 2 = 6 , 572 , and Q = 676.92 .

Declarations

Authors’ Affiliations

(1)
Department of Mathematics, Faculty of Science, Ankara University
(2)
Department of Mathematics and Computer Science, Çankaya University
(3)
Department of Mathematics, University of Nebraska-Lincoln

References

  1. Boros G, Moll V: Irresistible Integrals: Symbols, Analysis and Experiments in the Evaluation of Integrals. Cambridge University Press, Cambridge; 2004.View ArticleGoogle Scholar
  2. Cerone P, Dragomir SS: A refinement of the Grüss inequality and applications. Tamkang J. Math. 2007, 38: 37-49.MathSciNetMATHGoogle Scholar
  3. Grüss G:Über das Maximum des absoluten Betrages von 1 b a a b f ( x ) g ( x ) d x 1 ( b a ) 2 a b f ( x ) d x a b g ( x ) d x . Math. Z. 1935, 39: 215-226. 10.1007/BF01201355MathSciNetView ArticleGoogle Scholar
  4. Mercer AMD: An improvement of the Grüss inequality. JIPAM. J. Inequal. Pure Appl. Math. 2005.,6(4): Article ID 93. (electronic)Google Scholar
  5. Mitrinović DS, Pec̆arić JE, Fink AM: Classical and New Inequalities in Analysis. Kluwer Acadademic, Dordrecht; 1993.View ArticleMATHGoogle Scholar
  6. Anastassiou GA: Multivariate Fink type identity and multivariate Ostrowski, comparison of means and Grüss type inequalities. Math. Comput. Model. 2007, 46: 351-374. 10.1016/j.mcm.2006.11.008MathSciNetView ArticleMATHGoogle Scholar
  7. Bohner M, Matthews T: The Grüss inequality on time scales. Commun. Math. Anal. 2007, 3: 1-8. (electronic)MathSciNetMATHGoogle Scholar
  8. Dragomir SS: A Grüss type discrete inequality in inner product spaces and applications. J. Math. Anal. Appl. 2000, 250: 494-511. 10.1006/jmaa.2000.6977MathSciNetView ArticleMATHGoogle Scholar
  9. Liu Z: Notes on a Grüss type inequality and its applications. Vietnam J. Math. 2007, 35: 121-127.MathSciNetMATHGoogle Scholar
  10. Perić I, Rajić R: Grüss inequality for completely bounded maps. Linear Algebra Appl. 2004, 390: 292-298.Google Scholar
  11. Pec̆arić JE, Tepes̆ B: On the Grüss type inequalities of Dragomir and Fedotov. J. Inequal. Pure Appl. Math. 2003.,4(5): Article ID 91. (electronic)Google Scholar
  12. Pec̆arić JE, Tepes̆ B: A note on Grüss type inequality in terms of Δ-seminorms. Pril. - Maked. Akad. Nauk. Umet., Odd. Mat.-Teh. Nauki 2004, 23/24: 29-35.Google Scholar
  13. Goodrich CS: Continuity of solutions to discrete fractional initial value problems. Comput. Math. Appl. 2010, 59: 3489-3499. 10.1016/j.camwa.2010.03.040MathSciNetView ArticleMATHGoogle Scholar
  14. Goodrich CS: Existence and uniqueness of solutions to a fractional difference equation with nonlocal conditions. Comput. Math. Appl. 2011, 61: 191-202. 10.1016/j.camwa.2010.10.041MathSciNetView ArticleMATHGoogle Scholar
  15. Goodrich CS: On discrete sequential fractional boundary value problems. J. Math. Anal. Appl. 2012, 385: 111-124. 10.1016/j.jmaa.2011.06.022MathSciNetView ArticleMATHGoogle Scholar
  16. Abdeljawad T, Atici FM: On the definitions of nabla fractional operators. Abstr. Appl. Anal. 2012., 2012: Article ID 406757. Special IssueGoogle Scholar
  17. Anderson DR: Taylor polynomials for nabla dynamic equations on time scales. Panam. Math. J. 2002, 12: 17-27.MATHGoogle Scholar
  18. Anastassiou GA: Nabla discrete fractional calculus and nabla inequalities. Math. Comput. Model. 2010, 51: 562-571. 10.1016/j.mcm.2009.11.006MathSciNetView ArticleMATHGoogle Scholar

Copyright

© Güvenilir et al.; licensee Springer. 2014

This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.