1. Introduction
Convex functions have played a fundamental and transversal role in many branches of mathematics, and their importance has grown significantly in the
and
centuries. From convex optimization to Measure and Probability Theory, through various fields such as: Real and Functional Analysis, Economics and Game Theory, Machine Learning and Data Science, Information Theory and Entropy, among others, it has been acquiring marked relevance, which has translated into more and more researchers becoming interested in this concept, which has brought an increase in research and results, causing various generalizations and extensions of the classical concept (interested readers can consult [
16]).
Definition 1.
A set is said to be convex function if
for each and
Definition 2.
The function , is said to be convex function if the following inequality holds:
for all and .
The above inequality holds in opposite direction for concave function. Convex functions are at the heart of several classical integral inequalities, which are key tools in real analysis, probability, measure theory and other areas: Jensen’s inequality, Dragomir-Agarwal inequality, Prekopa-Leindler inequality and, by far the best known, the Hermite-Hadamard inequality.
These inequalities are of central importance. Below is a statement of this double inequality:
Suppose that
is a convex function on the closed real interval
where
. Therefore (see[
1,
8,
9,
13,
15])
For more recent developments related to the Hermite–Hadamard inequality, the reader may consult references [
2,
5,
6,
7,
11,
12,
18].
Several important inequalities have been established using different types of convexity. One such type is the modified –convexity.
This class was defined in [
3,
4] as follows:
Definition 3.
Let and . If inequality
is fulfilled and , where and . The function ψ is then referred to as a modified –convex function of the first type on .
Definition 4.
Let and . If inequality
is fulfilled and , where and .The function ψ is then referred to as a modified –convex function of the second type on .
Remark 1. Those interested can check that Definitions 3 and 4 encompass many of the known notions of convexity: classic convex, s-convex, -convex, -convex, -convex, -convex and others. So we have
Example 1: Power Function Let , with . This function is convex in and, under certain conditions on h, m, and s, it can satisfy the above inequalities.
Example 2: Exponential Function Let with . This function is convex and, as in the previous case, can satisfy the conditions of -modified convexity for appropriate values of the parameters.
Example 3: Logarithmic Function Let . This function is concave in , but considering the definitions of -modified concavity, it can be a valid example.
It can also be extended to the generalized logarithmic family, for example:
, with Or even to functions like: , as long as m are chosen small enough and are chosen to smooth the inequality sufficiently.
In the following we present some basic concepts of the Fourier transform of a function (see page 75 of [
17] and page 580 of [
10]).
Definition 5.
If a function is piecewise continuous in each finite interval and is absolutely integrable in , then the Fourier transform of denoted by is given by the integral
The inverse Fourier transform is given by
The properties of the Fourier Transform and its inverse, defined above, can be consulted in the texts cited above.
In this paper, we present various new forms of the Hermite–Hadamard inequality, within the context of -convex modified functions of second type, using Fourier Transform.
Main outcomes
The following result is a version of the classic Hermite-Hadamard inequality.
Theorem 1.
Let be a convex function, I a real interval, with and . Then, the following inequalities:
for Fourier integral transform are fulfilled, where is the Fourier Transform of function , with .
Proof. From
-convexity of
g we have (with
and
we have
Let
, let’s multiply the previous inequality by
and integrate the result between 0 and 1 with respect to
t, so we have
Thus, we have the first part of the inequality (
7).
For the second part, let’s use the
-convexity of
g, so we have
Let’s multiply both inequalities by
and integrating witgh respect to
t over
leads us to
From here, using the definitions of
and
we have
After changing variables in the integrals of both left members, multiplying the first inequality by
, the second by
and adding the results obtained, we get
After multiplying both sides by , we obtain the second part of the desired inequality.
This completes the proof. □
Remark 2. For convex functions, this result becomes Theorem 3.1 of [14].
To prove our main Theorems, we need the following equality:
Lemma 1.
Let g be a real-valued function defined on a closed real interval , differentiable on , and let . If , then the following equality holds:
Proof. Let
as before and let us denote for simplicity
Integrating by parts in
, we get
result obtained after making the change of variable
in the integral and using the definition of Fourier Transform and the shift property.
Analogously, making
Subtracting (
11) from (
10) gives the desired result. □
Hereafter, we use the following notation:
Theorem 2.
Let be differentiable function on such that If is a –convex modified function of the second type on for some fixed and , then the following inequality holds:
Proof. Applying modulus in Lemma 1, we have
From the definition of
and
we easily have
and
Using the
convexity of
we derive
and
Using these last two inequalities we obtain the desired result. □
Corollary 1. Under the assumptions of Theorem 2,
-
1.
-
If we choose , then we derive the following inequality
and are as before.
-
2.
-
Next we refine equation
12 by imposing additional conditions on
.
Theorem 3.
Let be differentiable function on such that If , with , is a –convex modified function of the second type on for some fixed and , then the following inequality holds:
where , and .
Proof. By using Hölder inequality in view of the fact that
is a
–convex modified of the second type, for
and
, we get
and
Thus, from
we have (
14). The proof is complete. □
Corollary 2. Under the assumptions of Theorem 3,
-
1.
Choosing , then we obtain the following inequality
-
2.
Theorem 4.
Let be differentiable function on such that If , with , is a –convex modified function of the second type on for some fixed and , then the following inequality holds:
where and as in the Theorem 2.
Proof. By using power–mean inequality in view of the fact that
is a
–convex modified of the second type, for
and
, we get
Or, taking into account the accepted notations, we can write:
Similarly, for
we get
Thus, for
we have
This inequality completes the proof. □
Theorem 5.
Let be differentiable function on such that If , with , is a –convex modified function of the second type on for some fixed and , then the following inequality holds:
where and , and as in the Theorem 3.
Proof. By using Young inequality view and of the fact that
is a
–convex modified of the second type, for
and
as before, we get
Thus, Adding these last two inequalities together, we obtain the desired result. This completes the proof of the Theorem. □
Remark 3. The reader will have no difficulty in stating the Corollaries corresponding to these last results, for -convex and h-convex functions.