- Research
- Open access
- Published:
Refining and reversing the weighted arithmetic–geometric mean inequality involving convex functionals and application for the functional entropy
Journal of Inequalities and Applications volume 2020, Article number: 92 (2020)
Abstract
In this paper, we present some refinements and reverses for some inequalities involving the weighted arithmetic mean and the weighted geometric mean of two convex functionals. Inequalities involving the Heinz functional mean are also obtained. As applications, we give some refinements and reverses for the relative entropy and the Tsallis relative entropy involving operator or functional arguments.
1 Introduction
Let \(a,b>0\) and \(\lambda\in[0,1]\) be real numbers. The following expressions
are known in the literature as the λ-weighted arithmetic mean, the λ-weighted harmonic mean, and the λ-weighted geometric mean, respectively. For \(\lambda=1/2\), they are simply denoted by \(a\nabla b\), \(a!b\), and \(a\sharp b\), respectively. In the recent few years, these means have received extensive attention which led to several developments and interesting applications, see [1–3, 5, 7, 10, 13, 16–19].
The previous weighted means satisfy the following inequalities:
known as the weighted arithmetic–geometric–harmonic mean inequality. Some refinements and reverses of (1.2) have been discussed in the literature. In particular, the following result has been proved in [10, 13]:
where we set
Another weighted mean, known as the Heinz mean, is defined by
As it is well known, \(\operatorname{HZ}_{\lambda}(a,b)\) interpolates the arithmetic mean and the geometric mean in the sense that the inequality
holds for any \(a,b>0\).
A refinement and reverse of the right inequality in (1.6) have been recently obtained in the literature, see [12, 13], for instance,
Other reversed and refined versions of the same inequality can be found in [11, 12].
The extension of the previous means, from the case where the variables are positive real numbers to the case where the arguments are positive operators, has been investigated in the literature. Let H be a complex Hilbert space and let \({\mathcal {B}}(H)\) be the \(\mathbb {C}^{*}\)-algebra of bounded linear operators acting on H. We denote by \({\mathcal {B}}^{+*}(H)\) the open cone of all (self-adjoint) invertible positive operators in \({\mathcal {B}}(H)\). For \(A,B\in{\mathcal {B}}^{+*}(H)\), the following expressions:
are known in the literature as the λ-weighted arithmetic operator mean, the λ-weighted harmonic operator mean, and the λ-weighted geometric operator mean of A and B, respectively. For \(\lambda=1/2\), they are simply denoted by \(A\nabla B\), \(A!B\), and \(A\sharp B\), respectively. These operator means satisfy the following inequality:
which is the operator version of (1.2). Here, the notation ≤ stands for the Löwner partial order defined as follows: \(T\leq S\) if and only if T and S are self-adjoint and \(S-T\) is positive.
An operator version of (1.3) has been established in [10] and reads as follows:
where \(r_{\lambda}\) is defined by (1.4).
By analogy with the scalar case, the Heinz operator mean is defined by
We also have the following operator inequalities extending (1.6):
In fact, according to the Kubo–Ando theory [14], (1.9) and (1.11) can be immediately deduced from (1.3) and (1.6), respectively. For more inequalities related to the Heinz mean involving matrix/operator arguments, we refer the reader to [4, 8, 9, 11, 15, 24] and the related references cited therein.
2 Functional means
The previous operator means have been extended from the case where the variables are invertible positive operators to the case that the variables are convex functionals, see [21, 22]. To explain this, we need to recall some basic notions from convex analysis. Let \(f:H\to {\mathbb {R}}\cup\{+\infty\}\) be a given functional. The notation domf refers to the effective domain of f defined by
while \(f^{*}\) stands for the Fenchel conjugate of f defined through
The subdifferential \(\partial f(x)\) of f at \(x\in\operatorname{dom} f\) is the (possibly empty) subset of H defined by
and it is well known that
Henceforth, and by virtue of the definition of the subdifferential, whenever we write \(\partial f(x)\) it will be assumed that \(x\in \operatorname{dom} f\).
Let us denote by \(\varGamma_{0}(H)\) the cone of all functionals \(f:H\to {\mathbb {R}}\cup\{+\infty\}\) which are convex lower semicontinuous and not identically equal to +∞. It is well known that \(f\in\varGamma _{0}(H)\) if and only if \(f^{**}:=(f^{*})^{*}=f\) and, if \(f\in\varGamma_{0}(H)\) then \(x^{*}\in\partial f(x)\) if and only if \(x\in\partial f^{*}(x^{*})\). If we denote by \(\operatorname{int}(\operatorname{dom} f)\) the topological interior of domf, we recall that if \(f\in\varGamma_{0} (H)\) and \(\operatorname {int}(\operatorname{dom} f)\) is nonempty then for all \(x\in \operatorname{int}(\operatorname{dom} f)\), f is continuous at x and \(\partial f(x)\) is nonempty.
The following example, which will be needed in the sequel, explains the previous notions in more detail.
Example 2.1
Let \(A\in{\mathcal {B}}(H)\) be a self-adjoint operator. Usually, we denote by \(f_{A}\) the quadratic form generated by the operator A, i.e.,
Such functional \(f_{A}\) enjoys the following properties:
- (i)
Let \(A,B\in{\mathcal {B}}(H)\) be self-adjoint. For any \(\alpha,\beta \in{\mathbb {R}}\) we have \(\alpha f_{A}+\beta f_{B}=f_{\alpha A+\beta B}\). Further, it is clear that \(f_{A}\leq f_{B}\) if and only if \(A\leq B\). If, moreover, A is positive then \(f_{A}\in\varGamma_{0}(H)\) and \(\partial f_{A}(x)=\{Ax\}\) for any \(x\in H\).
- (ii)
If \(A\in{\mathcal {B}}^{+*}(H)\) then we have
$$\forall x\in H,\quad f_{A}^{*}(x)=\frac{1}{2}\bigl\langle A^{-1}x,x\bigr\rangle , $$which can be written as \(f_{A}^{*}=f_{A^{-1}}\).
Throughout this paper, we use the following notation
Now, let \((f,g)\in{\mathcal {D}}(H)\) and \(\lambda\in(0,1)\). The following expressions:
are called (by analogy) the λ-weighted functional arithmetic mean, the λ-weighted functional harmonic mean, and the λ-weighted functional geometric mean of f and g, respectively. For \(\lambda=1/2\), they are simply denoted by \({\mathcal {A}}(f,g)\), \({\mathcal {H}}(f,g)\), and \({\mathcal {G}}(f,g)\), respectively. We also have the following double inequality:
known, by analogy, as the weighted arithmetic–geometric–harmonic functional mean inequality. Here, the symbol ≤ refers to the pointwise order defined as follows: \(f\leq g\) if and only if \(f(x)\leq g(x)\) for all \(x\in H\).
We can extend the previous functional means on the whole interval \([0,1]\) by setting
For all \((f,g)\in{\mathcal {D}}(H)\) and \(\lambda\in[0,1]\) we introduce the following expression [23]:
which is the Heinz functional mean of f and g. It is clear that \(\mathcal{HZ}_{\lambda}(f,g)\) is symmetric in f and g and \(\mathcal{HZ}_{\lambda}(f,g)=\mathcal{HZ}_{1-\lambda}(f,g)\). Further, \(\mathcal{HZ}_{0}(f,g)=\mathcal{HZ}_{1}(f,g)={\mathcal {A}}(f,g)\) and \(\mathcal{HZ}_{1/2}(f,g)={\mathcal {G}}(f,g)\). Otherwise, from (2.6) and (2.8) we can check that the inequality
holds for any \(\lambda\in[0,1]\) and \(x\in H\).
The following remark may be of interest for the reader.
Remark 2.1
We notice that the mentioned functionals can take the value +∞. For this, we observe the following:
- (i)
The equalities (2.7) cannot be deduced from (2.3), (2.4) and (2.5), respectively. This is because the involved functionals can take the value +∞, with the convention \(0\cdot(+\infty)=(+\infty)-(+\infty)=(+\infty)+(-\infty):=+\infty\), as is usual in convex analysis.
- (ii)
The functional inequality (2.9) can be easily proved, but with some precautions. These precautions should be taken in consideration in the proof of any functional equality or inequality. This latter point is explained in the following items.
- (iii)
For instance, the equality \(f-f=0\) is not always true for any \(f:H\to{\mathbb {R}}\cup\{+\infty\}\) by virtue of the same reason as in (i). Precisely, we have \(f-f=\varPsi_{\operatorname{dom} f}\) for any \(f:H\to{\mathbb {R}}\cup\{+\infty\}\), where the notation \(\varPsi_{C}\) refers to the indicator function of the subset C of H defined as \(\varPsi_{C}(z)=0\) if \(z\in C\) and \(\varPsi_{C}(z)=+\infty\) if \(z\notin C\).
- (iv)
Similarly, the equality \(f-g=-(g-f)\) is not always true, unless f or g is everywhere finite. For the same reason, the functional inequality \(f\leq g\) is equivalent to \(g-f\geq0\) but it is not equivalent to \(f-g\leq0\).
Clearly, \({\mathcal {A}}_{\lambda}(f,f)=f\) for any \(f:H\to{\mathbb {R}}\cup\{ +\infty\}\). Indeed, when \(f\in\varGamma_{0}(H)\) then \(f^{**}=f\) and so the previous functional means satisfy the following relations:
Further, all the previous functional means are respectively extensions of their related operator means in the following sense:
where the notation \(f_{A}\) (resp. \(f_{B}\)) was defined in Example 2.1. With this, (1.8) is an immediate consequence of (2.6); see [21, 22], for instance.
The main goal of this paper is to give some refined and reversed inequalities between the weighted arithmetic mean and the weighted geometric mean involving convex functionals. We also obtain some refinements and reverses of (1.11) when the operator variables A and B are replaced by convex functionals. As applications, we obtain some refinements and reverses for some inequalities involving the relative entropy and the Tsallis entropy with operator or functional arguments.
3 Some needed results
Before stating our main results, we need to recall the following lemmas.
Lemma 3.1
([21])
Let\((f,g)\in{\mathcal {D}}(H)\). For each\(t,s\in(0,1)\), the following inequalities:
hold for any\(x\in H\), where we set
Lemma 3.2
([23])
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). For all\(x\in H\), one has
In particular, for all\(x\in H\)we have
Lemma 3.3
([23])
Let\((f,g)\in{\mathcal {D}}(H)\). Then the map
is pointwise integrable on\((0,1)\). That is, for any\(x\in H\), the integral
exists in\({\mathbb {R}}\cup\{+\infty\}\).
Let \(f:H\to{\mathbb {R}}\cup\{+\infty\}\) and \(x,x^{*}\in H\). We set
Following the so-called Fenchel inequality, we have
We also define
with the usual convention \(\sup_{\emptyset}(\cdot)=-\infty\). With this, it is not hard to check that the following formula:
holds for any \(x\in H\). Using (3.7) or (3.8), we can see that \((f\diamond g )(x)\leq g(x)\) holds for any \(x\in H\).
Below, we present an example which explains the previous notions in more detail. Such an example will be needed throughout this paper.
Example 3.1
Take \(f(x)=(1/2)\langle Ax,x\rangle\) and \(g(x)=(1/2)\langle Bx,x\rangle \), where \(A,B\in{\mathcal {B}}^{+*}(H)\). We write \(f=f_{A}\) and \(g=f_{B}\), for the sake of simplicity. Since \(\partial f(x)=\{Ax\}\) and \(g^{*}=f_{B^{-1}}\), (3.7) yields
where \(A\diamond B\) is given by
We notice that \(f\diamond g\) is not always a convex functional, and so \(A\diamond B\) is not always a positive operator. However, from (3.7) we can immediately deduce that \(f\diamond g\leq g\), which implies that \(B-A\diamond B\) is always positive.
Finally, let \(a>0\), \(b\geq0\) and \(\lambda,s\in(0,1)\). For the sake of simplicity, we set
By simple computation of integrals, it is easy to see that
We can then check that
We have the following result.
Proposition 3.4
For all\(s\in(0,1)\), we have
Proof
It is a simple exercise of real analysis. We first use the change of variables by setting \(u=\sqrt{t/(1-t)}\) and then we proceed by an appropriate integration by parts. The details are straightforward and therefore omitted. □
4 The main results
We preserve the same notations as in the previous sections. We start this section by stating the following result.
Theorem 4.1
Let\((f,g)\in{\mathcal {D}}(H)\)and\(t\in(0,1)\). Let\(x\in H\)be such that\(\partial f(x)\neq\emptyset\)and\(\partial g(x)\neq\emptyset\). Then the following inequality:
holds for any\(x^{*}\in\partial f(x)\)and\(z^{*}\in\partial g(x)\).
Proof
By definition, we have
from which we deduce that the inequality
holds for any \(x,x^{*}\in H\). It follows that
or equivalently,
If \(x^{*}\in\partial f(x)\) then \(f(x)+f^{*}(x^{*})=\Re e\langle x^{*},x\rangle\) and so
Similarly, if \(z^{*}\in\partial g(x)\), we have
It follows that if \(x^{*}\in\partial f(x)\) and \(z^{*}\in\partial g(x)\) then
and
By adding (4.2) and (4.3), with a simple manipulation, we obtain (4.1). □
Corollary 4.2
For any\(x\in H\)such that\(\partial f(x)\cap\operatorname{dom} g^{*}\neq \emptyset\)and\(\partial g(x)\cap\operatorname{dom} f^{*}\neq\emptyset\), the integral (3.4) is finite, i.e., \({\mathcal {J}}(f,g)(x)<+\infty \). Further, the inequality
holds for any\(x^{*}\in\partial f(x)\cap\operatorname{dom} g^{*}\)and\(z^{*}\in\partial g(x)\cap\operatorname{dom} f^{*}\).
Proof
First, by Lemma 3.3, the integral in (3.4) exists in \({\mathbb {R}}\cup\{+\infty\}\). This, with (4.1) and (3.5), yields the desired result. The details are simple and therefore omitted here. □
The following result will be needed in the sequel.
Proposition 4.3
For all\(x\in H\), we have
Proof
By (3.2), we can write
If we replace λ by \(1-\lambda\) in this latter formula, we obtain
Adding (4.5) and (4.6), we obtain (4.4). □
We now state our second main result which reads as follows.
Theorem 4.4
With the same hypotheses as in Theorem4.1, the following inequalities:
hold for any\(x^{*}\in\partial f(x)\)and\(z^{*}\in\partial g(x)\).
Proof
By (4.5) and (4.1), we can write
If B and Γ denote the standard beta and gamma special functions, respectively, then we have
Thus inequality (4.7) follows.
If we replace λ by \(1-\lambda\) in (4.7), we obtain
Adding (4.7) and (4.9) and remarking that for \(x\in H\) such that \(f(x)<+\infty\) we have
we obtain (4.8). The proof is completed. □
We can write the previous theorem in another form, which is symmetric in f and g. Precisely, the following result may be stated.
Theorem 4.5
With the same hypotheses as in Theorem4.1, the following inequalities:
hold, where\(f\diamond g\)is defined by (3.7).
Proof
According to (4.7), the functional
is lower-bounded, with
Following (3.5) and (3.7), we have
Similarly, one has
Summarizing, we have proved that
Substituting this into (4.7), we obtain (4.10).
Similarly, (4.11) can be deduced from (4.8). The proof is complete. □
The operator versions of Theorems 4.1 and 4.4 (or Theorem 4.5) are given in the following.
Corollary 4.6
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda\in[0,1]\). Then we have the following operator inequalities:
Now, let us observe the following remark which explains the interest of our functional approach and its application for scalar/operator means.
Remark 4.1
The inequalities of the preceding corollary are important due to the fact that they refine some existing operator inequalities. By virtue of the Kubo–Ando theory for monotone operator means, it is enough to discuss the scalar case when justifying our claim. We restrict ourselves to inequality (4.13), since the others can be investigated in a similar manner.
Indeed, the scalar version of (4.13) is reduced to
We would like to compare (4.14) with the right inequality in (1.3). In fact, by the techniques of real analysis, it is not hard to show that there exist \(\lambda_{0}\) and \(\lambda_{1}\in(0,1)\) such that if \(0\leq\lambda_{0}\leq\lambda\) or \(\lambda_{1}\leq\lambda\leq1\) then (4.14) is better than the right inequality of (1.3).
Now we will present some results giving simultaneously refinements and reverses of the inequalities \({\mathcal {G}}_{\lambda}(f,g)\leq{\mathcal {A}}_{\lambda}(f,g)\) and \(\mathcal{HZ}_{\lambda}(f,g)\leq{\mathcal {A}}(f,g)\).
Theorem 4.7
Let\((f,g)\in{\mathcal {D}}(H)\). For each\(\lambda, s\in(0,1)\), the following inequalities:
hold for any\(x\in H\).
Proof
First, we show the right inequality of (4.15). By virtue of (2.6), we can write, for all \(x\in H\),
It is not hard to check that
The right inequality of (4.15) follows by Lemma 3.3 and (3.2).
We now show the left inequality of (4.15). For the right hand-side of (3.2), we use the left inequality in (3.1). For any fixed \(s\in(0,1)\), it is easy to see that
If in this latter integral, we use a simple change of variables, we deduce the desired inequality with the help of (4.7) and a simple manipulation. □
The operator version of Theorem 4.7 reads as follows.
Corollary 4.8
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda,s\in(0,1)\). Then we have
where we set
with
Proof
If in (4.15) we take \(f=f_{A}\) and \(g=f_{B}\), we obtain the left inequality of (4.16) by (2.10). To obtain the right inequality of (4.16), we can check that, see [23],
where \(J(A,B)\) is given by (4.17). The details are simple and therefore omitted here. □
We need to introduce the following notation:
The following inequalities:
hold for any \(x\in H\). For the left inequality, see [23]. For the other inequalities, they are immediate from (2.6), (2.8), and (2.9).
As an application of the previous theorem, we may state the following corollary which concerns a reverse and refinement of (2.9).
Corollary 4.9
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda, s\in(0,1)\). Then the following inequalities:
hold for any\(x\in H\).
Proof
First, if \(x\in H\) is such that \(f(x)=+\infty\) then the inequalities in (4.19) are reduced to an equality, since all sides of (4.19) are equal to +∞ by virtue of the usual convention \(+\infty +c=+\infty\) for any \(c\in[-\infty,+\infty]\). Assume that below \(f(x)<+\infty\). If in (4.15) we replace λ by \(1-\lambda\) and s by \(1-s\) and add the results, we obtain the desired inequalities with the help of (2.8), (4.18), and the fact that
is valid for any \(s\in(0,1)\) and \(x\in\operatorname{dom} f\). The proof is completed. □
The following result is of interest.
Corollary 4.10
Let\((f,g)\in{\mathcal {D}}(H)\). Then the following inequalities:
hold for any\(x\in H\).
Proof
Taking \(\lambda=s=1/2\) in (4.19) and using Proposition 3.4, we immediately obtain the desired inequalities after a simple computation and reduction. The details are straightforward and therefore omitted here. □
The right inequality of (4.20) gives a lower bound of \({\mathcal {J}}(f,g)\), and the left one yields a reverse of \({\mathcal {A}}(f,g)-{\mathcal {G}}(f,g)\leq{\mathcal {A}}(f,g)-{\mathcal {H}}(f,g)\). Note that Corollary 4.2 contains an upper bound of \({\mathcal {J}}(f,g)\).
Also, we have the following result.
Theorem 4.11
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). Then, for all\(x\in H\), we have the following inequalities:
Proof
From (1.3) we immediately deduce that, for all \(a,b>0\),
or equivalently,
First, observe that \(1-2r_{\lambda}=|2\lambda-1|\). If we take \(a=1-t\) and \(b=t\) in the latter inequalities, we get, after simple manipulations,
Substituting the result into (4.4), we obtain
with
and
It is not hard to see that
which, with (3.3), immediately yields the left inequality of (4.21).
Similarly, we show the right inequality of (4.21). □
Theorem 4.12
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). Let\(x\in H\)be such that\(\partial f(x)\neq\emptyset\)and\(\partial g(x)\neq\emptyset \). Then the following inequalities:
hold for any\(x^{*}\in\partial f(x)\)and\(z^{*}\in\partial g(x)\).
Proof
If in (1.7) we take \(a=1-t\) and \(b=t\) then we obtain
Substituting the result into (4.4), we get
where
and
According to (4.1), with the fact that \((1-2t)\log ((1-t)/t )\geq0\) for any \(t\in(0,1)\), it is easy to see that
for any \(x^{*}\in\partial f(x)\) and \(z^{*}\in\partial g(x)\), where we set
A simple integration by parts leads to \(\alpha=1\), and so the left inequality of (4.22) is obtained.
Otherwise, it is not hard to see that
This, with (3.3) and the fact that \(1-4\lambda(1-\lambda)=(2\lambda -1)^{2}\), immediately yields the right inequality of (4.22). The proof of the theorem is finished. □
Theorem 4.12 implies an interesting result which may be recited in the following.
Corollary 4.13
Let\((f,g)\in{\mathcal {D}}(H)\)and let\(x\in H\)be as in Theorem4.12. Then we have
Proof
It is immediate from (4.22). The details are simple and therefore omitted. □
Another interesting result that gives a reverse of (4.20) is given in what follows.
Corollary 4.14
Let\((f,g)\in{\mathcal {D}}(H)\)and let\(x\in H\)be as in Theorem4.12. Then the following inequalities hold:
Proof
According to (4.12), the left inequality of (4.22) yields (4.23) after a simple manipulation. If we take \(\lambda=1/2\) in (4.23), we immediately obtain (4.24). □
We end this section by stating the following remark which may be of interest.
Remark 4.2
-
(i)
We mention that if \(\lambda=1/2\) then the inequalities in (4.21) are reduced to a trivial equality. Otherwise, if we take \(\lambda =0\) or \(\lambda=1\) in (4.21) and use Corollary 4.13, we obtain (4.20).
However, (4.15), (4.19), and (4.22) are reduced to a trivial equality when \(\lambda=0\) and \(\lambda=1\).
-
(ii)
We leave to the reader the routine task of formulating the operator versions of Theorems 4.11 and 4.12, as well as Corollary 4.14, in a similar manner as previously.
5 Applications for functional entropies
In this section we will apply the previous results for obtaining some refined upper bounds for the relative entropy and Tsallis entropy of two (convex) functionals. The operator versions will be immediately obtained. We first need to recall some basic notions.
For \(A,B\in{\mathcal {B}}^{+*}(H)\) and \(\lambda\in(0,1)\), the relative operator entropy \(S(A|B)\) and the Tsallis relative operator entropy \(T_{\lambda}(A|B)\) are defined by (see [6, 7], for instance)
It is well known that \(T_{\lambda}(A|B)\) extends \(S(A|B)\) in the sense that the following equality:
holds for all \(A,B\in{\mathcal {B}}^{+*}(H)\), where the limit is taken in the strong operator topology.
The operators \(S(A|B)\) and \(T_{\lambda}(A|B)\) have been extended from the case where the variables are invertible positive operators to the case where the variables are (convex) functionals (see [20], for instance) as
The following relationship:
has been proved in [20]. Here, the limit is taken for the pointwise topology.
Some inequalities giving lower and upper bounds of \({\mathcal {E}}(f|g)\) and \({\mathcal {R}}_{\lambda}(f|g)\) can be found in [20]. In particular, the following inequalities:
hold for any \(x\in\operatorname{dom} f\).
In what follows, applying some results of the previous section, we will give some reverses and refinements of (5.3). The related operator versions will be immediately deduced without any additional tools.
Let us first state the following result which is useful and will be needed in the sequel.
Proposition 5.1
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). Let\(x\in \operatorname{dom} f\cap\operatorname{dom} g\). Then we have
In particular, the following equality:
holds for any\(x\in\operatorname{dom} f\cap\operatorname{dom} g\).
Proof
If \(x\in\operatorname{dom} f\) then (5.1) implies that
We infer that if \(x\in\operatorname{dom} g\) then
If \(x\in\operatorname{dom} f\cap\operatorname{dom} g\), we can add (5.6) and (5.7). We then obtain (5.4). Equality (5.5) follows from (5.4) when combined with (5.2). The proof is finished. □
Combining Corollary 4.13 and Proposition 5.1, we immediately deduce the following result which contains a relationship between the relative functional entropy and the functional \({\mathcal {J}}(f,g)\) previously defined.
Corollary 5.2
The following equality:
holds for any\(x\in\operatorname{dom} f\cap\operatorname{dom} g\).
By using (4.17), it is easy to see that the operator version of (5.8) is given by
Now, we are in the position to state the following result which gives some reverses of (5.3).
Theorem 5.3
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). Let\(x\in H\)be as in Theorem4.1. Then the following inequalities:
hold for any\(x^{*}\in\partial f(x)\)and\(z^{*}\in\partial g(x)\).
Proof
First, for any \(x\in\operatorname{dom} f\), it is not hard to see that
Substituting this into (4.7) and then dividing by \(\lambda>0\), we obtain (5.9).
Letting \(\lambda\rightarrow0\) in (5.9) and using (5.2), we immediately get (5.10). □
The operator version of Theorem 5.3 is recited in the following.
Corollary 5.4
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda\in(0,1)\). Then we have
We now state another result which gives a refinement and reverse of (5.3).
Theorem 5.5
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda, s\in(0,1)\). The following inequalities:
and
hold for any\(x\in\operatorname{dom} f\).
Proof
Substituting (5.11) into the middle part of (4.15) and then dividing by \(\lambda>0\), we obtain the desired inequalities (5.12).
We now prove (5.13). If in all sides of (5.12) we let \(\lambda \downarrow0\) and use (3.10) and (5.2), with the fact that
we then get the desired inequalities (5.13). □
The operator version of Theorem 5.5 is recited in the following result.
Corollary 5.6
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda,s\in(0,1)\). Then we have
and
where\(J(A,B)\)is given by (4.17).
Now we will see other applications for the relative functional entropies. The first result reads as follows.
Theorem 5.7
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda, s\in(0,1)\). The following inequalities:
hold for any\(x\in\operatorname{dom} f\cap\operatorname{dom} g\).
Proof
From (4.19) we can write, for any \(x\in\operatorname{dom} f\cap \operatorname{dom} g\),
Substituting (5.4) in the middle part of these inequalities and then dividing by \(\lambda/2\), we get (5.14). □
Corollary 5.8
Let\((f,g)\in{\mathcal {D}}(H)\)and\(s\in(0,1)\). The following inequality:
holds for any\(x\in\operatorname{dom} f\cap\operatorname{dom} g\).
Proof
Letting \(\lambda\rightarrow0\) in (5.14) with the help of (5.2) and (3.10), we obtain (5.16). □
The operator versions of Theorem 5.7 and Corollary 5.8 are given in the following result.
Corollary 5.9
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda,s\in(0,1)\). Then we have
and
where we set
Another main result is given in the following.
Theorem 5.10
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). The following inequalities:
hold for any\(x\in\operatorname{dom} f\cap\operatorname{dom} g\).
Proof
Here, we use (4.21). The details are similar to those of the proof of the preceding results. □
The operator version of Theorem 5.10 is given below.
Corollary 5.11
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda\in(0,1)\). Then we have
Finally, our last result of applications is stated in the following.
Theorem 5.12
Let\((f,g)\in{\mathcal {D}}(H)\)and\(\lambda\in(0,1)\). Let\(x\in H\)be as in Theorem4.1. Then the following inequalities:
hold for any\(x^{*}\in\partial f(x)\)and\(z^{*}\in\partial g(x)\).
Proof
It is also similar to the previous proofs, by using (4.22). We leave to the reader the routine task of providing the details of this proof. □
The operator version of Theorem 5.12 can be immediately deduced and reads as follows.
Corollary 5.13
Let\(A,B\in{\mathcal {B}}^{+*}(H)\)and\(\lambda\in(0,1)\). Then we have
References
Aldaz, J.M.: A refinement of the inequality between arithmetic and geometric means. J. Math. Inequal. 2(4), 473–477 (2008). https://doi.org/10.7153/jmi-02-42
Aldaz, J.M.: Self-improvement of the inequality between arithmetic and geometric means. J. Math. Inequal. 3(2), 213–216 (2009). https://doi.org/10.7153/jmi-03-21
Aldaz, J.M.: Comparison of difference between arithmetic and geometric means. Tamkang J. Math. 42(4), 453–462 (2011). https://doi.org/10.5556/j.tkjm.42.2011.453-462
Dinh, T.H., Dumitru, R., Franco, J.A.: The matrix power means and interpolations. Adv. Oper. Theory 3(3), 647–654 (2018). https://doi.org/10.15352/aot.1801-1288
Dragomir, S.S.: Bounds for the normalized Jensen functional. Bull. Aust. Math. Soc. 74(3), 471–478 (2006). https://doi.org/10.1017/s000497270004051x
Fujii, J.I., Kamei, E.: Relative operator entropy in noncommutative information theory. Math. Jpn. 34, 341–348 (1989)
Furuichi, S.: Inequalities for Tsallis relative entropy and generalized skew information. Linear Multilinear Algebra 59(10), 1143–1158 (2011). https://doi.org/10.1080/03081087.2011.574624
Kapil, Y., Conde, C., Moslehian, M.S., Singh, M., Sababheh, M.: Norm inequalities related to the Heron and Heinz means. Mediterr. J. Math. 14(5), Article ID 213 (2017). https://doi.org/10.1007/s00009-017-1015-6
Kittaneh, F., Krnić, M., Lovričević, N., Pečarić, J.: Improved arithmetic–geometric and Heinz means inequalities for Hilbert space operators. Publ. Math. (Debr.) 80(3–4), 465–478 (2012). https://doi.org/10.5486/pmd.2012.5193
Kittaneh, F., Manasrah, Y.: Improved Young and Heinz inequalities for matrices. J. Math. Anal. Appl. 36, 262–269 (2010). https://doi.org/10.1016/j.jmaa.2009.08.059
Kittaneh, F., Manasrah, Y.: Reverse Young and Heinz inequalities for matrices. Linear Multilinear Algebra 59, 1031–1037 (2011). https://doi.org/10.1080/03081087.2010.551661
Kittaneh, F., Moslehian, M.S., Sababheh, M.: Quadratic interpolation of the Heinz mean. Math. Inequal. Appl. 21(3), 739–757 (2018). https://doi.org/10.7153/mia-2018-21-53
Krnić, M.: More accurate Young, Heinz, and Hölder inequalities for matrices. Period. Math. Hung. 71, 78–91 (2015). https://doi.org/10.1007/s10998-015-0086-z
Kubo, F., Ando, T.: Means of positive linear operators. Math. Ann. 246, 205–224 (1980). https://doi.org/10.1007/bf01371042
Liang, J., Shi, G.: Refinements of the Heinz operator inequalities. Linear Multilinear Algebra 63, 1337–1344 (2015). https://doi.org/10.1080/03081087.2014.936434
Qi, F., Guo, B.N.: The reciprocal of the weighted geometric mean is a Stieltjes function. Bol. Soc. Mat. Mexicana (3) 24(1), 181–202 (2018). https://doi.org/10.1007/s40590-016-0151-5
Qi, F., Zhang, X.J., Li, W.H.: An integral representation for the weighted geometric mean and its applications. Acta Math. Sin. Engl. Ser. 30(1), 61–68 (2014). https://doi.org/10.1007/s10114-013-2547-8
Qi, F., Zhang, X.J., Li, W.H.: Lévy–Khintchine representations of the weighted geometric mean and the logarithmic mean. Mediterr. J. Math. 11(2), 315–327 (2014). https://doi.org/10.1007/s00009-013-0311-z
Qi, F., Zhang, X.J., Li, W.H.: The harmonic and geometric means are Bernstein functions. Bol. Soc. Mat. Mexicana (3) 23(2), 713–736 (2017). https://doi.org/10.1007/s40590-016-0085-y
Raïssouli, M.: Functional versions for some operator entropy inequalities. Ann. Funct. Anal. 6(2), 204–211 (2015). https://doi.org/10.15352/afa/06-2-17
Raïssouli, M.: Functional versions of some refined and reversed operator mean-inequalities. Bull. Aust. Math. Soc. 96(3), 496–503 (2017). https://doi.org/10.1017/s0004972717000594
Raïssouli, M., Bouziane, H.: Arithmetico–geometrico–harmonic functional mean in convex analysis. Ann. Sci. Math. Qué. 30(1), 79–107 (2006)
Raïssouli, M., Furuichi, S.: Some inequalities involving Heron and Heinz means of two convex functionals (2018). arXiv:1809.01820v2
Zou, L.: Inequalities related to Heinz and Heron means. J. Math. Inequal. 7(3), 389–397 (2013). https://doi.org/10.7153/jmi-07-34
Acknowledgements
We would like to express our sincere thanks to the three anonymous referees for their useful comments and suggestions which substantially helped improve the quality of the present manuscript.
Availability of data and materials
Not applicable.
Funding
Not applicable.
Author information
Authors and Affiliations
Contributions
All authors have contributed to, checked, read, and approved the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, 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 licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Raïssouli, M., Almozini, M. Refining and reversing the weighted arithmetic–geometric mean inequality involving convex functionals and application for the functional entropy. J Inequal Appl 2020, 92 (2020). https://doi.org/10.1186/s13660-020-02355-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13660-020-02355-3