On the (p, ηh )-convex function and some integral inequalities

1 Introduction
Let I and J be two intervals belongs to R.To encourage our work,let us recall the
definitions of some different classes of functions.

Definition 1 Let f : I → R be a function,it is called Godunova-Levin function or

that f belongs to the class Q(I) and f is non-negative,if

f (x) f (y)
f (αx + (1 − α)y) ≤ +
α 1−α

∀ (x, y) ∈ I and α ∈ (0, 1).

Definition 2 Let h : J → R be a non-negative function.we say that f : I → R is

an h-convex function and f is non-negative,if

f (αx + (1 − α)y) ≤ h(α)f (x) + h(1 − α)f (y)

∀ (x, y) ∈ I and α ∈ (0, 1).

Definition 3 A function f : I → R is called convex with respect to η (briefly

η-convex),if the following inequality holds

f (αx + (1 − α)y) ≤ f (y) + α η(f (x), f (y)),

∀ (x, y) ∈ I and α ∈ [0, 1].

Definition 4 Let f : I → R be a function and h be a non-negative function it

is said to be ηh ,if

f (αx + (1 − α)y) ≤ f (y) + h(α) η(f (x), f (y)),

∀ (x, y) ∈ I and α ∈ [0, 1] and η : A×A → B for appropriate A,B ⊆ R and h : j → R

is A non-negative function.

Definition 5 Let I be a convex set.A function f : I → R is said to be p con-

vex function or that f belongs to the class P C(I),if
1
f ([αxp + (1 − α)y p ] p ) ≤ αf (x) + (1 − α)f (y),

∀ (x, y) ∈ I and α ∈ [0, 1].

The convex functions h and p were introduced by Varsanec,Zhang and Wan,and

Jensen’s inequalities and a number of properties of the function were established(cf.[8]).
1
2

Definition 6 A function h : J → R is a non negative and non zero function.

f : I → R is called (p,h)-convex function or belongs to the class ghx(h, p, I),if f is
non-negative
1
f ([αxp + (1 − α)y p ] p ) ≤ h(α)f (x) + h(1 − α)f (y) (1)

∀ (x, y) ∈ I and α ∈ (0, 1)..Similarly,if inequality (1) is reversed,then f is called a

(p,h)-concave function or belong to the class ghv(h,p,I).
In this sequel of the paper,we introduce new definitions and properties for the
(p, ηh )-convex function.we will prove the theorem of composition and will give some
inequalities of Jansen’s type.

2 Definition and basic results

in this paper,we suppose that (0, 1) ⊆ J and functions f and h are real non negative
functions defined on J and I,respectively.We introduce definition of the new class
of convex function.

Definition 7 Let h : J → R be a non negative.we say that f : I → R is a

x
(p, ηh )-convex function or belongs to the class ηh,p (I),if f is non negative
1
f ([αxp + (1 − α)y p ] p ) ≤ f (y) + h(α) η(f (x), f (y))

∀ (x, y) ∈ I and α ∈ (0, 1).Similarly,if the above inequality is reversed,then f is

c
called (p, ηh )-concave function or that f belongs to the class ηh,p (I)

Remark 1 It can be obviously seen that,

(1) if h(α) = α, η(x−y) = x−y and p = 1 then it becomes classical convex function.
(2) if p=1 then it becomes ηh convex function.
(3) if h(α) = α1 and p = 1 then f belongs to the class of Q(I) which is called
Godunova-Levin function.
(4) if f is non-negative function and h(α) = α, η(x − y) = x − y and p = 1 then f
becomes a h-convex function.
x
(5) if h(α) = 1, η(x − y) = x − y and p = 1,then P (I) ⊆ ηh,p (I).

Example 1 Let hk (α) = αk where α is positive and k ≤ 1.f (x) = xp , x ≤ 0

is a function where p is odd number then we have
1
f ([αxp + (1 − α)y p ] p ) ≤ αf (x) + (1 − α)f (y) ≤ f (y) + hk (α) η (f (x), f (y)),

x
So,f∈ ηh,p (I).

Now,we discuss different properties of (p, ηh )-convex(concave) functions,which are

linearity,product,composition properties.

Property 1 Let f, g : I → R be two (p, ηh )-convex functions and λ ≥ 0 then,

 3

(a) f + g : I → R is (p, ηh )-convex.

(b) λf : I → R is (p, ηh )-convex.
c c
Similarly,if f, g ∈ ηh,p (I), λ ≥ 0 then we have f + g, λf ∈ ηh,p (I).

Proof(a) Given is,

1
f ([αxp + (1 − α)y p ] p ) ≤ f (y) + h(α) η(f (x), f (y)) (2.1)

∀ (x, y) ∈ I and α ∈ (0, 1).

1
g([αxp + (1 − α)y p ] p ) ≤ g(y) + h(α) η(g(x), g(y)) (2.2)

∀ (x, y) ∈ I and α ∈ (0, 1).

Consider,
1
(f +g)([αxp +(1−α)y p ] p ) ≤ [f (y)+h(α) η (f (x), f (y))]+[g(y)+h(α) η (g(x), g(y))]
= f (y) + g(y) + h(α)[η(f (x), f (y)) + η(g(x), g(y))]
= f (y) + g(y) + h(α)[η(f (x) + g(x), f (y) + g(y))]
≤ (f + g)(y) + h(α)[η(f + g(x), f + g(y))]

hence proved.

Proof(b) Consider,
1 1
(λf )([αxp + (1 − α)y p ] p ) = λf [(αxp + (1 − α)y p ] p ]
≤ λ[f (y) + h(α)η(f (x), f (y))]
= λf (y) + h(α)(λf (x), λf (y))
= (λf )(y) + h(α)η((λf )(x), (λf )(y))
So,λf is (p, ηh )-convex.

Definition 8 Let h : I → R be a function,it is said to be super-multiplicative

function if

h(xy) ≥ h(x)h(y)

∀ (x, y) ∈ J.

Property 2 Let hi : Ji → (0, ∞),where i= 1,2, are functions with h2 (J2 ) ⊆ J1

and h2 (α) + h2 (1 − α) ≤ 1, ∀ x, y ∈ (0, 1) and f : I1 → [0, ∞) and g : I2 → [0, ∞)
are functions with conditions g(I2 ) ⊆ I1 , and f (0) = 0

(1) If h1 is super multiplicative function,f is increasing f ∈ SX(h1 , I1 ) and g ∈

x
ηh,p (I2 ),then f og composite function belongs to ηhx1 oh2 (I)
c
Similarly, if f is decreasing,then g ∈ ηh,p (I2 ).

(2) If h1 is sub multiplicative function,f is increasing f ∈ SV (h1 , I1 ) and g ∈

x
ηh,p (I2 ),then f og composite function belongs to ηhx1 oh2 (I)
4

c
Similarly,if f is decreasing,then g ∈ ηh,p (I2 )

Proof: If g ∈ ηhx2 ,p (I2 ) and f is increasing function, then we have

1
(f og)([αxp + (1 − α)y p ] p ) ≤ f (g(y) + h2 (α)η(g(x), g(y)))

∀x, y ∈ l2 and α ∈ (0, 1)

using definition of (p, ηh )-convex function with p=1,we obtain

f (g(y) + h2 (α)η(g(x), g(y))) ≤ f (g(y)) + h1 (h2 (α)η(f (g(x)), f (g(y)))

≤ f o g(y) + h1 (h2 (α)η(f o g(x), f o g(y))

hence f o g ∈ ηhx1 oh2 (I2 ).

Property 3 Let we have two non-negative and non zero function h1 and h2 de-
fined on interval J with condition h2 ≤ h1 ∈ (0, 1).If f ∈ ηhx2 ,p (I),then f ∈
ηhx1 ,p (I).Similarly,if f ∈ ηhc 1 ,p (I),then f ∈ ηhc 2 ,p (I).

Proof if f ∈ ηhx2 ,p (I) then for any x, y ∈ I and α ∈ (0, 1),we have
 1

f [αxp + (1 − α)y p ] p ≤ f (y) + h2 (α) η (f (x), f (y))

≤ f (y) + h1 (α) η (f (x), f (y))

hence f ∈ ηhx1 ,p (I).

Property 4 Let f ∈ gηh x(ηh , p1 , I).

(a) For (0, 1] ⊇ I,if f is strictly(monotone) increasing and strictly decreasing func-
tion having condition p2 ≥ p1 ≥ 0 or p2 ≤ p1 ≤ 0 and (p1 ≥ p2 ≥ 0 or p1 ≤ p2 ≤ 0 )
then f ∈ gηh x(ηh , p2 , I).

(b) For [1, ∞) ⊇ I,if f is strictly(monotone) increasing(strictly decreasing) function

having condition p1 ≥ p2 ≥ 0 or p1 ≤ p2 ≤ 0 and (p2 ≥ p1 ≥ 0 or p2 ≤ p1 ≤ 0)
then f ∈ gηh x(ηh , p2 , I).

Let f ∈ gηh v(ηh , p, I).

(c) For (0, 1] ⊇ I,if f is strictly(monotone) increasing(strictly decreasing) function

having condition p1 ≥ p2 ≥ 0 or p1 ≤ p2 ≤ 0 and (p2 ≥ p1 ≥ 0 or p2 ≤ p1 ≤ 0 )
then f ∈ gηh v(ηh , p2 , I).

(d) For [1, ∞) ⊇ I,if f is strictly(monotone) increasing(strictly decreasing) function

having condition p2 ≥ p1 ≥ 0 or p2 ≤ p1 ≤ 0 and (p1 ≥ p2 ≥ 0 or p1 ≤ p2 ≤ 0 )
then f ∈ gηh v(ηh , p2 , I).
 5

1
Proof(a) Take g(p) = (αxp + (1 − α)y p ) p then we have

1 1
−1
g 0 (p) = (αxp + (1 − α)y p ) p (αxp ln(x) + (1 − α)y p ln(y))
p

when x, y ∈ (0, 1] and p ≥ 0 then we have g 0 (p) ≤ 0 and so g(p2 ) ≤ g(p2 ).then we
obtain

f (g(p2 )) ≤ f (g(p2 )) ≤ f (y) + h(α) η (f (x), f (y))

since f ∈ gηh x(ηh , p1 , I) and that f is monotone increasing.Hence f ∈ gηh x(ηh , p2 , I)

Theorem
Let fj : I → R, j ∈ J is non empty collection of (p, ηh )-convex functions such that
(a) there exist α ∈ [0, ∞] and β ∈ [−1, ∞] such that η(x, y) = αx + βy ∀x, y ∈ R
(b) for each x ∈ I, sup fj (x) exists in R
j∈J
then the function f : I → R defined by f (x) = sup fJ (x) for each x ∈ I is (p, ηh )-
j∈J
convex.
Proof: For any x, y ∈ I and λ ∈ [0, 1],we have
1 1
f [αxp + (1 − α)y p ] p = sup fj [αxp + (1 − α)y p ] p
j∈J

≤ sup {fj (y) + h(α)η(fj (x), fj (y))}

j∈J

= sup {fj (y) + h(α)(αfj (x), βfj (y))}

j∈J

= sup {(1 + βh(α))fj (y) + αh(α)fj (x)}

j∈J

≤ (1 + βh(α) sup fj (y) + αh(α) sup fj (x)

j∈J j∈J

= (1 + βh(α)f (y) + αh(α)f (x)

= f (y) + h(α)(αf (x) + βf (y))
= f (y) + h(α)η(f (x), f (y)).
which is required.
3 Jensen’s-type inequality
In this section,we introduce some Jansen’s-type inequality for (p, ηh )-convex func-
tions.

Statement Let w1 , w2 , ..., wn ∈ R+ with n ≥ 2.Let f : I → R (p, ηh ) be-convex

function and h be non-negative super multiplicative function and η be non decreas-
ing,non negatively sub linear variable
Then we have following inequality

n n
1 X 1 X Wi
f ([ wi xpi ] p ) ≤ f (xn ) + h( )ηf (xi + xi+1 , ...xn ) (3.1)
Wn i=1 i=1
W n
6

n
P
where Wn = wi . and ηf (xi + xi+1 , ...xn ) = η(ηf (xi + xi+1 , ...xn−1 ), f (xn )) and
i=1
ηf (x) = f (x) ∀x ∈ I

c
If we have h is sub-multiplicative and f ∈ ηh,p then the inequality sign in (3.1)
is reversed.

Proof
n n−1
1 X 1 Wn−1 X wi p wn p p1
f ([ wi xpi ] p ) = f ([ x x ] )
Wn i=1 Wn i=1 Wn−1 i Wn n

n−1
Wn−1 X wi p 1
≤ f (xn ) + h( )η(f ([ xi ] p ), f (xn ))
Wn i=1
W n−1

n−2
Wn−1 Wn−2 X wi p wn−1 p 1
= f (xn ) + h( )η(f [ xi + xn−1 ] p , f (xn ))
Wn Wn−1 i=1 Wn−2 Wn−1

since η be non negative sub linear in first variable so,

n−2
Wn−1 Wn−2 X wi p 1
≤ f (xn ) + h( )η(f (xn−1 ) + h( )η(f [ xi ] p , f (xn−1 )), f (xn ))
Wn Wn−1 i=1
W n−2

and h is super multiplicative

n−2
Wn−1 Wn−2 X wi p 1
≤ f (xn )+h( )η(f (x−n − 1), f (xn ))+h( )η(η(f [ xi ] p , f (xn−2 )), f (xn ))
Wn Wn i=1
W n−2

≤ ...

Wn−1 Wn−2 W1
≤ f (xn )+h( )η(f (x−n − 1), f (xn ))+h( )η(η(f (xn−2 ), f (xn )), f (xn ))+...+h( )η(η(...η(f (x1 ), f (x2 )
Wn Wn Wn

Wn−1 wn−2 W1
= f (xn )+h( )ηf (xn−1 , xn )+h( )ηf (xn−2 , xn−1 , xn )+...+h( )ηf (x1 , x2 , ..., xn−1 , xn )
Wn Wn Wn

n−1
X Wi
= f (xn ) + h( )ηf (xi , xi+1 , ..., xn )
i=1
Wn

Schur-type inequalities
In this section, we established Schur-type inequalities of (p, ηh )-convex functions.
 7

Statement Let h : J → R be a non-negative super multiplicative function and

x
let f be a function defined on interval I such that f ∈ ηh,p (I).then ∀x1 , x2 , x3 ∈ I
p p p p p p
such that x1 < x2 < x3 and x3 − x1 , x3 − x2 , x2 − x1 ∈ J the following inequality
holds:

f (x3 )h(xp3 − xp1 ) − f (x2 )h(xp3 − xp1 ) + h(xp3 − xp2 )η((f (x1 ), f (x3 )) ≥ 0
.
c
If the function is sub multiplicative and f ∈ ηh,p (I),then the above inequality is
reversed.

x
Proof. Let f ∈ ηh,p (I) and let x1 , x2 , x3 ∈ I be given.Then we can easily see
that

xp3 − xp2 xp2 − xp1

, ∈ (0, 1) ⊆ J
xp3 − xp1 xp3 − xp1
and
xp3 − xp2 xp2 − xp1
p p + p =1
x3 − x1 x3 − xp1
We also have

xp3 − xp2 p xp3 − xp2

h(xp3 − xp2 ) = h( p
p
p (x3 − x1 )) ≥ h( p )h(xp3 − xp1 )
x3 − x1 x3 − xp1
and

xp2 − xp1
h(xp2 − xp1 ) ≥ h( )h(xp3 − xp1 )
xp3 − xp1
xp −xp
Setting α = xp3 −xp2 , x = x1 and y = x3 in definition of (p, ηh )-convex function.We
3 1
have xp2 = αxp + (1 − α)y p and

xp3 − xp2
f (x2 ) ≤ f (x3 ) + h( )η((f (x1 ), f (x3 ))
xp3 − xp1

h(xp3 − xp2 )
f (x2 ) ≤ f (x3 ) + η((f (x1 ), f (x3 ))
h(xp3 − xp1 )
Assuming h(xp3 − xp1 ) > 0 and multiplying both sides of the inequality above by
h(xp3 − xp1 ) we obtain,

f (x2 )h(xp3 − xp1 ) ≤ f (x3 )h(xp3 − x1 ) + h(xp3 − xp2 )η(f (x1 ), f (x3 ))

f (x3 )h(xp3 − xp1 ) − f (x2 )h(xp3 − xp1 ) + h(xp3 − xp2 )η(f (x1 ), f (x3 )) ≥ 0

hence proved.
Remark 2 Let η(x, y) = x − y in above inequality then we obtain the schur-type
inequality for (p, h)-convex function.
8

Lemma suppose,f : I → R be a differential function on I o .If f 0 ∈ L0 [a, b],where

a, b ∈ I having condition a < b then the following inequality holds:

Zb Z1
a+b 1 b−a a+b a+b
|f ( )− f (x)dx| ≤ [|f 0 ( t+(1−t)a)|+|f 0 ( t+(1−t)b)|]dt
2 b−a 4 2 2
0 0
(4.1)
Theorem Suppose f : [a, b] → R,[a, b] ⊂ [0, ∞) be a differentiable function on (a,b)
such that f 0 ∈ L0 [a, b].If f is (p, ηh )-convex function,then we obtain the inequality;

Z1
b−a 0
[|f (a)| + |f 0 (b)| + 2η(|f 0 (a + b)|, |f 0 (a + b)|)( th(t)dt)
8
0

Proof Here we using lemma4.1,Since f 0 is (p, ηh )-convex function on [a,b] for any
t ∈ [0, 1] we obtain

a+b a+b
|f 0 ( t + (1 − t)a)| ≤ |f 0 (a)| + h(t)η(|f 0 ( )|, |f 0 (a)|) (4.2)
2 2
a+b a+b
|f 0 ( t + (1 − t)b)| ≤ |f 0 (b)| + h(t)η(|f 0 ( )|, |f 0 (b)|) (4.3)
2 2

Now (4.1),(4.2) and (4.3) imply

Zb Z1
a+b 1 b−a a+b a+b
|f ( )− f (x)dx| ≤ t[(|f 0 (a)|+h(t)η(|f 0 ( )|, |f 0 (a)|)+(|f 0 (b)|+h(t)η(|f 0 ( )|, |f 0 (b)|))]
2 b−a 4 2 2
0 0

Z1
b−a 0
= [|f (a)|+|f 0 (b)|+2η(|f 0 (a+b)|, |f 0 (a+b)|)( th(t)dt)
8
0

hence proved.

x
Hadamard type inequality Let f ∈ ηh,p (I) ∩ L1 ([a, b]) for a, b ∈ I with con-
dition a < b, then we obtain the inequality

Z1 Zb
ap + bp p1 1 p p 1
p p p 1
f ([ ] )−h( ) η(f ([ta +(1−t)b ] )), f ([(1−t)a +tb ] )))dt ≤ p
p p xp−1 f (x)dx ≤ f (b)+η(f (a), f (
2 2 b − ap
0 a
p p p p p p
Proof Take x = ta + (1 − t)b , y = (1 − t)a + tb

xp + y p ap + bp
=
2 2

So,
ap + bp p1 xp + y p p1
f ([ ] ) = f ([ ] )
2 2
 9

by defintion of (p, ηh )-convex functions,we have

ap + bp p1 1 1 1 1
f ([ ] ) ≤ f ([1−t)ap +tbp ] p )+h( )η(f ([tap +(1−t)bp ] p ), f ([1−t)ap +tbp ] p ))
2 2
integrating above inequality w.r.t ’t’ on [0,1],we get

Z1 Z1
ap + bp p1 p 1 p 1 1 1
f ([ ] )≤ f ([(1−t)a +tb ] )dt+h( ) p η(f ([tap +(1−t)bp ] p ), f ([(1−t)ap +tbp ] p ))dt (5.2)
2 2
0 0

integration by substitution
Z1 Zb
p p p 1
f ([(1 − t)a + tb ] )dt = p p y p−1 f (y)dy
b − ap
0 a

put in (5.2)

Z1 Zb
ap + bp p1 1 p p 1 p p p 1
f ([ ] )−h( ) η(f ([ta +(1−t)b ] ), f ([(1−t)a +tb ] ))dt ≤ p
p p y p−1 f (y)dy (
2 2 b − ap
0 a

Now,
1
f ([(1−t)ap +tbp ] p ) ≤ f (b)+h(1−t)η(f (a), f (b)) (5.3)

As,
Zb Z1
p p−1 1
y f (y)dy = f ([(1 − t)ap + tbp ] p )
bp − ap
a 0

using(5.3)
Z1
≤ f (b)+η(f (a), f (b)) h(1−t)dt (ii)
0

Combining (i) and (ii),we obtain the result

Z1 Zb
ap + bp p1 1 p p 1 p p p 1
f ([ ] )−h( ) (f ([ta +(1−t)b ] ), f ([(1−t)a +tb ] ))dt ≤ p
p p xp−1 f (x)dx ≤ f (b)+η(f (a), f (b))
2 2 b − ap
0 a

Remark 3 In (5.1),if we choose h(1-t)=t,η(x, y) = x−y and p=1 then(5.1) reduces

to classical Hermite Hadamard type inequality for function.

Theorem Suppose f and g are (p, ηh )-convex functions,f g ∈ L1 ([0, 1]) and
h1 h2 ∈ ([0, 1]) with a, b ∈ I and a < b.Then we have,
10

Theorem Suppose f and g are (p, ηh )-convex functions,f g ∈ L1 ([0, 1]) and h1 h2 ∈
([0, 1]) with a, b ∈ I and a < b.Then we have,

p
Rb R1 R1 R1
bp −ap xp−1 f (x)g(x) ≤ f (b)g(b)+L(a, b) h1 (t)dt+M (a, b) h2 (t)dt+N (a, b) h1 (t)h2 (t)dt
a 0 0 0

Where L(a, b) = g(b)η(f (a), f (b)), M (a, b) = f (b)η(g(a), g(b)) and N (a, b) = η(f (a), f (b))η(g(a), g(b))

Proof Since f ∈ ηhx1 ,p (I) and g ∈ ηhx2 ,p (I),we have

1
f ([tap + (1 − t)bp ] p ) ≤ f (b) + h1 (t)η(f (a), f (b))

1
g([tap + (1 − t)bp ] p ) ≤ g(b) + h2 (t)η(g(a), g(b))

∀t ∈ [0, 1].Here f and g are non-negative function,we get the inequality

1 1
f ([tap + (1 − t)bp ] p )g([tap + (1 − t)bp ] p ) ≤ f (b)g(b) + h1 (t)g(b)η(f (a), f (b)) +
h2 (t)f (b)η(g(a), g(b)) + h1 (t)h2 (t)η(f (a), f (b))η(g(a), g(b))

Integrating both sides of the above inequality over (0,1),we get inequality

R1 1 1 R1
f ([tap +(1−t)bp ] p )g([tap +(1−t)bp ] p )dt ≤ f (b)g(b)+g(b)η(f (a), f (b)) h1 (t)dt+
0 0
R1 R1
f (b)η(g(a), g(b)) h2 (t)dt + η(f (a), f (b))η(g(a), g(b)) h1 (t)h2 (t)dt
0 0
1
Setting x = [tap + (1 − t)bp ] p ,we obtain
p
Rb p−1 R1 R1
bp −ap x f (x)g(x) ≤ f (b)g(b)+g(b)η(f (a), f (b)) h1 (t)dt+f (b)η(g(a), g(b)) h2 (t)dt+
a 0 0
R1
η(f (a), f (b))η(g(a), g(b)) h1 (t)h2 (t)dt
0
p
Rb R1 R1 R1
bp −ap xp−1 f (x)g(x) ≤ f (b)g(b)+L(a, b) h1 (t)dt+M (a, b) h2 (t)dt+N (a, b) h1 (t)h2 (t)dt.
a 0 0 0

Theorem Let f ∈ ηhx1 ,p (I), g ∈ ηhx2 ,p (I) be functions such that f g ∈ L1 ([a, b])
and h1 h2 ∈ L1 ([0, 1]),and let a, b ∈ I with a < b.Then we have

1 p 1
+bp p 1
ap +bp p p
Rb R1 R1
2h1 ( 12 )h2 ( 12 )
f(a 2 ) g( 2 ) − bp +ap xp−1 f (x)g(x)dx ≤ L(a, b) h1 (t)dt+M (a, b) h2 (t)dt+
a 0 0
R1
N (a, b) h1 (t)h2 (t)dt
0

Proof Since
ap + bp tap + (1 − t)bp (1 − t)ap + tbp
= +
2 2 2
,we have
p p 1
ap +bp
f ([ a +b
2 ] )g([ 2 ]
p