1 Introduction
Let \(I\subseteq \mathbb{R}\) be an interval. Then a real-valued function \(\varPsi : I\mapsto \mathbb{R}\) is said to be convex on I if the inequality
holds for all \(a, b\in I\) and \(\lambda \in (0, 1)\). Ψ is said to be concave if inequality (1.1) is reversed.
$$ \varPsi \bigl[\lambda a+(1-\lambda )b\bigr]\leq \lambda \varPsi (a)+(1-\lambda ) \varPsi (b) $$
(1.1)
It is well known that the convexity theory has wide applications in special functions [1‐30], differential equations [31‐61] and bivariate means [62‐67]. Recently, the extensions, generalizations, refinements and variants for the convexity have attracted the attention of many researchers. For example, Schur convexity [68‐70], GA-convexity [71], GG-convexity [72], s-convexity [73, 74], preinvexity [75], strong convexity [76‐79] and others [80‐85].
Advertisement
Dragomir [86] defined the coordinate convex as follows.
Definition 1.1
(See [86])
Let \(I_{1}, I_{2}\subseteq \mathbb{R}\) be two interval, \(\varPsi : I_{1}\times I_{2}\mapsto \mathbb{R}\) be a real-valued function, and the partial mappings \(\varPsi _{y}: I_{1}\mapsto \mathbb{R}\) and \(\varPsi _{x}: I_{2}\mapsto \mathbb{R}\) be defined by
respectively. Then Ψ is said to be coordinate convex on \(I_{1}\times I_{2}\) if \(\varPsi _{y}\) is convex on \(I_{1}\) for all \(y\in I_{2}\) and \(\varPsi _{x}\) is convex on \(I_{2}\) for all \(x\in I_{1}\).
$$ \varPsi _{y}(u)=\varPsi (u, y), \qquad \varPsi _{x}(v)=\varPsi (x, v), $$
Remark 1.2
Dragomir [86] proved that every convex function is coordinate convex, but not vice versa.
Advertisement
Next, we recall the concept of η-convexity which can be found in the literature [87].
Definition 1.3
(See [87])
Let \(I\subseteq \mathbb{R}\) be an interval, and \(\varPsi : I\mapsto \mathbb{R}\) and \(\eta : \mathbb{R} \times \mathbb{R}\mapsto \mathbb{R}\) be two real-valued functions. Then Ψ is said to be η-convex if the inequality
holds for all \(x, y\in I\) and \(\mu \in [0, 1]\).
$$ \varPsi \bigl[\mu x+(1-\mu )y\bigr]\leq \varPsi (y)+\mu \eta \bigl[\varPsi (x), \varPsi (y) \bigr] $$
The main purpose of the article is to give a non-trivial example for a η-convex function defined on rectangle is not convex, prove that every η-convex function defined on rectangle is coordinate η-convex but not vice versa, define the coordinate \((\eta _{1}, \eta _{2})\)-convex function and establish its Hermite–Hadamard type inequality.
2 Main results
To begin this section, it is interesting to give the definition of η-convex function defined on rectangle, and give a non-trivial example for a η-convex function defined on rectangle is not convex.
Definition 2.1
Let \(I_{1}, I_{2}\subseteq \mathbb{R}\) be two intervals, and \(\varPsi : I_{1}\times I_{2}\mapsto \mathbb{R}\) and \(\eta : \mathbb{R}\times \mathbb{R}\mapsto \mathbb{R}\) be two real-valued functions. Then Ψ is said to be η-convex if the inequality
holds for all \((x, y), (z, w)\in I_{1}\times I_{2}\) and \(\mu \in [0, 1]\).
$$ \varPsi \bigl[\mu x+(1-\mu )z, \mu y+(1-\mu )w\bigr]\leq \varPsi (z, w)+\mu \eta \bigl[ \varPsi (x,y), \varPsi (z, w)\bigr] $$
Example 2.2
Let \(\varPsi : [1,5]\times [1,5]\mapsto \mathbb{R}\) and \(\eta : \mathbb{R}\times \mathbb{R}\mapsto \mathbb{R}\) be defined by
Then Ψ is η-convex on \([1,5]\times [1,5]\), but it is not convex.
$$ \varPsi (x, y)=x^{2}y^{2}, \qquad \eta (x, y)=104x+103y. $$
Proof
Let \(\mu \in [0, 1]\). Then for any \((x, y), (z, w)\in [1,5]\) we have
Note that
It follows from (2.1) and (2.2) that
which shows that Ψ is η-convex on \([1,5]\times [1,5]\). It is easily to verify that Ψ is not convex on \([1,5]\times [1,5]\), for details see [79]. □
$$\begin{aligned}& \varPsi \bigl[\mu x+(1-\mu )z,\mu y+(1-\mu )w\bigr] \\& \quad =\bigl[\mu x+(1-\mu )z\bigr]^{2}\bigl[\mu y+(1-\mu )w \bigr]^{2} \\& \quad =\bigl[z^{2}+\mu \bigl(\mu x^{2}+\mu z^{2}-2z^{2}\bigr)+2\mu (1-\mu )xz\bigr] \\& \qquad {}\times \bigl[w^{2}+\mu \bigl(\mu y^{2}+\mu w^{2}-2w^{2}\bigr)+2\mu (1-\mu )yw\bigr] \\& \quad \leq \bigl[z^{2}+\mu x^{2}+2\mu (1-\mu )xz\bigr] \bigl[w^{2}+\mu y^{2}+2\mu (1-\mu )yw\bigr] \\& \quad \leq \bigl[z^{2}+\mu x^{2}+\mu (1-\mu ) \bigl(x^{2}+z^{2}\bigr)\bigr] \bigl[w^{2}+\mu y^{2}+ \mu (1-\mu ) \bigl(y^{2}+w^{2}\bigr) \bigr] \\& \quad \leq \bigl[z^{2}+\mu \bigl(x^{2}+x^{2}+z^{2} \bigr)\bigr] \bigl[w^{2}+\mu \bigl(y^{2}+y^{2}+w^{2} \bigr)\bigr] \\& \quad =z^{2}w^{2}+\mu \bigl[2y^{2}z^{2}+z^{2}w^{2}+2x^{2}w^{2}+w^{2}z^{2} \bigr]+\mu ^{2}\bigl[4x^{2}y^{2}+2x^{2}w^{2}+2y^{2}z^{2}+z^{2}w^{2} \bigr] \\& \quad \leq \varPsi (z,w)+\mu \bigl[2y^{2}z^{2}+z^{2}w^{2}+2x^{2}w^{2}+w^{2}z^{2} \bigr]+ \mu \bigl[4x^{2}y^{2}+2x^{2}w^{2}+2y^{2}z^{2}+z^{2}w^{2} \bigr] \\& \quad =\varPsi (z,w)+\mu \bigl[4x^{2}y^{2}+3z^{2}w^{2}+4 \bigl(z^{2}y^{2}+x^{2}w^{2}\bigr) \bigr]. \end{aligned}$$
(2.1)
$$ z^{2}\leq 25x^{2}, \qquad x^{2}\leq 25z^{2}. $$
(2.2)
$$\begin{aligned}& \varPsi \bigl[\mu x+(1-\mu )z,\mu y+(1-\mu )w\bigr] \\& \quad \leq \varPsi (z,w)+\mu \bigl[104x^{2}y^{2}+103z^{2}w^{2} \bigr] \\& \quad =\varPsi (z,w)+\mu \eta \bigl[\varPsi (x,y),\varPsi (z,w)\bigr], \end{aligned}$$
Next, we introduce the definition of coordinate \((\eta _{1}, \eta _{2})\)-convexity.
Definition 2.3
Let \(I_{1}, I_{2}\subseteq \mathbb{R}\) be two intervals, \(\varPsi : I_{1}\times I_{2}\mapsto \mathbb{R}\), \(\eta _{1}, \eta _{2}: \mathbb{R}\times \mathbb{R}\mapsto \mathbb{R}\) be three real-valued functions, and the partial mappings \(\varPsi _{y}: I_{1} \mapsto \mathbb{R}\) and \(\varPsi _{x}: I_{2}\mapsto \mathbb{R}\) be defined by
Then Ψ is said to be coordinate \((\eta _{1}, \eta _{2})\)-convex on \(I_{1}\times I_{2}\) if \(\varPsi _{y}\) is \(\eta _{1}\)-convex on \(I_{1}\) and \(\varPsi _{x}\) is \(\eta _{2}\)-convex on \(I_{2}\). In particular, if \(\eta _{1}=\eta _{2}=\eta \), then Ψ is said to be coordinate η-convex.
$$ \varPsi _{y}(u)=\varPsi (u, y), \qquad \varPsi _{x}(v)=\varPsi (x, v). $$
Example 2.4
Let \(\varPsi : [0, \infty )\times [0, \infty ) \mapsto \mathbb{R}\) be defined by \(\varPsi (x, y)=-|x|-y^{2}\), \(\eta _{1}(x, y)=-x-y\) and \(\eta _{2}(x, y)=-x-2y\). Then Ψ is coordinate \((\eta _{1}, \eta _{2})\)-convex on \([0, \infty )\times [0, \infty )\).
Proof
Let \(x_{1}, y_{1}\in [0, \infty )\) and \(\mu \in [0, 1]\). Then for any \((x, y)\in [0, \infty )\) we clearly see that
It follows from (2.3)–(2.6) that
$$\begin{aligned}& \varPsi _{y}\bigl(\mu x_{1}+(1-\mu )x_{2}\bigr)=- \bigl\vert \mu x_{1}+(1-\mu )x_{2} \bigr\vert -y^{2}, \end{aligned}$$
(2.3)
$$\begin{aligned}& \varPsi _{y}(x_{2})+\mu \eta _{1}\bigl(\varPsi _{y}(x_{1}), \varPsi _{y}(x_{2}) \bigr) \\& \quad =- \vert x_{2} \vert -y^{2}+\mu \eta _{1}\bigl(- \vert x_{1} \vert -y^{2}, - \vert x_{2} \vert -y^{2}\bigr) \\& \quad =- \vert x_{2} \vert -y^{2}+\mu \bigl( \vert x_{1} \vert + \vert x_{2} \vert +2y^{2} \bigr), \end{aligned}$$
(2.4)
$$\begin{aligned}& \varPsi _{x}\bigl(\mu y_{1}+(1-\mu )y_{2} \bigr)=- \vert x \vert -\bigl(\mu y_{1}+(1-\mu )y_{2} \bigr)^{2}, \end{aligned}$$
(2.5)
$$\begin{aligned}& \varPsi _{x}(y_{2})+\mu \eta _{2}\bigl(\varPsi _{x}(y_{1}), \varPsi _{x}(y_{2}) \bigr) \\& \quad =- \vert x \vert -y_{2}^{2}+\mu \eta _{2}\bigl(- \vert x \vert -y_{1}^{2}, - \vert x \vert -y_{2}^{2}\bigr) \\& \quad=- \vert x \vert -y_{2}^{2}+\mu \bigl(y_{1}^{2}+2y_{2}^{2}+3 \vert x \vert \bigr). \end{aligned}$$
(2.6)
$$\begin{aligned}& \varPsi _{y}(x_{2})+\mu \eta _{1}\bigl(\varPsi _{y}(x_{1}), \varPsi _{y}(x_{2}) \bigr)-\varPsi _{y}\bigl(\mu x_{1}+(1-\mu )x_{2} \bigr) \\& \quad =\mu \bigl( \vert x_{1} \vert + \vert x_{2} \vert +2y^{2}\bigr)+ \bigl\vert \mu x_{1}+(1-\mu )x_{2} \bigr\vert - \vert x_{2} \vert \\& \quad \geq 2\mu y^{2}+\mu \vert x_{1} \vert +\mu \vert x_{2} \vert +(1-\mu ) \vert x_{2} \vert -\mu \vert x_{1} \vert - \vert x _{2} \vert \\& \quad =2\mu y^{2}\geq 0, \end{aligned}$$
(2.7)
$$\begin{aligned}& \varPsi _{x}(y_{2})+\mu \eta _{2}\bigl(\varPsi _{x}(y_{1}), \varPsi _{x}(y_{2}) \bigr)-\varPsi _{x}\bigl(\mu y_{1}+(1-\mu )y_{2} \bigr) \\& \quad =3\mu \vert x \vert +2\mu (1-\mu )y_{1}y_{2}+ \mu (1+\mu )y_{1}^{2}+\mu ^{2}y_{2} ^{2}\geq 0. \end{aligned}$$
(2.8)
Theorem 2.5
Let
\(I_{1}, I_{2}\subseteq \mathbb{R}\)be two interval and
\(\eta : \mathbb{R}\times \mathbb{R}\mapsto \mathbb{R}\)be a real-valued function. ThenΨis coordinateη-convex on
\(I_{1}\times I_{2}\)ifΨisη-convex on
\(I_{1}\times I _{2}\).
Proof
Let \((x,y)\in I_{1}\times I_{2}\), \(u, v\in I_{2}\) and \(z, w\in I_{1}\). Then it follows from the η-convexity of the function Ψ on \(I_{1}\times I_{2}\) that
and
$$\begin{aligned} \varPsi _{x}\bigl(\mu v+(1-\mu )u\bigr) =&\varPsi \bigl(x, \mu v+(1-\mu )u\bigr) \\ =&\varPsi \bigl(\mu x+(1-\mu )x, \mu v+(1-\mu )u\bigr) \\ \leq& \varPsi (x, u)+\mu \eta \bigl(\varPsi (x, v), \varPsi (x, u)\bigr) \\ =&\varPsi _{x}(u)+\mu \eta \bigl(\varPsi _{x}(v), \varPsi _{x}(u)\bigr) \end{aligned}$$
(2.9)
$$\begin{aligned} \varPsi _{y}\bigl(\mu z+(1-\mu )w\bigr) =&\varPsi \bigl(\mu z+(1-\mu )w, y \bigr) \\ =&\varPsi \bigl(\mu z+(1-\mu )w, \mu y+(1-\mu )y\bigr) \\ \leq &\varPsi (w, y)+\mu \eta \bigl(\varPsi (z, y), \varPsi (w, y)\bigr) \\ =&\varPsi _{y}(w)+\mu \eta \bigl(\varPsi _{y}(z), \varPsi _{y}(w)\bigr). \end{aligned}$$
(2.10)
Example 2.6
Let \(I_{1}=I_{2}=[0, \infty )\), \(\varPsi , \eta : I_{1}\times I_{2}\mapsto [0, \infty )\) be defined by
Then Ψ is coordinate η-convex on \(I_{1}\times I_{2}\) but it is not η-convex on \(I_{1}\times I_{2}\).
$$ \varPsi (x, y)=xy, \qquad \eta (x, y)=x+y. $$
(2.11)
Proof
Let \(x, y, u, v, z, w\in [0, \infty )\) and \(\mu \in [0, 1]\). Then it follows from (2.11) that
Inequalities (2.12)–(2.15) imply that
and
Note that
and
Therefore, Ψ is coordinate η-convex on \(I_{1}\times I_{2}\) follows from (2.16)–(2.19).
$$\begin{aligned}& \varPsi _{x}\bigl(\mu u+(1-\mu )v\bigr)=\varPsi \bigl(x, \mu u+(1-\mu )v \bigr) \\& \hphantom{\varPsi _{x}\bigl(\mu u+(1-\mu )v\bigr)} =x\bigl(\mu u+(1-\mu )v\bigr)=-\mu xv+x(\mu u+v), \end{aligned}$$
(2.12)
$$\begin{aligned}& \varPsi (x, v)+\mu \eta \bigl(\varPsi (x, u), \varPsi (x, v)\bigr)=xv+\mu \eta (xu, xv) \\& \hphantom{\varPsi (x, v)+\mu \eta \bigl(\varPsi (x, u), \varPsi (x, v)\bigr)} =xv+\mu (xu+xv)=\mu xv+x(\mu u+v), \end{aligned}$$
(2.13)
$$\begin{aligned}& \varPsi _{y}\bigl(\mu z+(1-\mu )w\bigr)=\varPsi \bigl(\mu z+(1-\mu )w, y \bigr) \\& \hphantom{\varPsi _{y}\bigl(\mu z+(1-\mu )w\bigr)} =y\bigl(\mu z+(1-\mu )w\bigr)=-\mu yw+y(\mu z+w), \end{aligned}$$
(2.14)
$$\begin{aligned}& \varPsi (w, y)+\mu \eta \bigl(\varPsi (z, y), \varPsi (w, y)\bigr)=wy+\mu \eta (zy, wy) \\& \hphantom{\varPsi (w, y)+\mu \eta \bigl(\varPsi (z, y), \varPsi (w, y)\bigr)} =wy+\mu (zy+wy)=\mu yw+y(\mu z+w). \end{aligned}$$
(2.15)
$$ \varPsi _{x}\bigl(\mu u+(1-\mu )v\bigr)\leq \varPsi (x, v)+\mu \eta \bigl(\varPsi (x, u), \varPsi (x, v)\bigr) $$
(2.16)
$$ \varPsi _{y}\bigl(\mu z+(1-\mu )w\bigr)\leq \varPsi (w, y)+\mu \eta \bigl(\varPsi (z, y), \varPsi (w, y)\bigr). $$
(2.17)
$$ \varPsi _{x}\bigl(\mu u+(1-\mu )v\bigr)=\varPsi \bigl(\mu x+(1-\mu )x, \mu u+(1-\mu )v\bigr) $$
(2.18)
$$ \varPsi _{y}\bigl(\mu z+(1-\mu )w\bigr)=\varPsi \bigl(\mu z+(1-\mu )w, \mu y+(1-\mu )y\bigr). $$
(2.19)
Next, we prove that Ψ is not η-convex on \(I_{1}\times I _{2}\).
Let \(\mu \in (0, 1)\), \(x=w=1\) and \(y=z=0\). Then (2.11) leads to
From (2.20) and (2.21) we clearly see that Ψ is not η-convex on \(I_{1}\times I_{2}\). □
$$\begin{aligned}& \varPsi \bigl(\mu x+(1-\mu )z, \mu y+(1-\mu )w\bigr) \\& \quad =\varPsi (\mu , 1-\mu )=\mu (1-\mu )>0, \end{aligned}$$
(2.20)
$$\begin{aligned}& \varPsi (z, w)+\mu \eta \bigl(\varPsi (x, y), \varPsi (z, w)\bigr) \\& \quad =\varPsi (0, 1)+\mu \eta \bigl(\varPsi (1, 0), \varPsi (0, 1)\bigr)=0. \end{aligned}$$
(2.21)
Next, we establish a Hermite–Hadamard type inequality for the coordinate \((\eta _{1}, \eta _{2})\)-convex function.
Theorem 2.7
Let
\(a, b, c, d\in \mathbb{R}\)with
\(a< b\)and
\(c< d\), \(\varPsi : [a, b]\times [c, d]\mapsto \mathbb{R}\), \(\eta _{1}, \eta _{2}: \mathbb{R}\times \mathbb{R}\mapsto \mathbb{R}\)be three real-valued functions such thatΨis coordinate
\((\eta _{1}, \eta _{2})\)-convex on
\([a, b]\times [c, d]\)and
for all
\(x, y\in \mathbb{R}\), where
\(M_{\eta _{1}}\)and
\(M_{\eta _{2}}\)are two positive constants. Then
$$ \eta _{1}(x, y)\leq M_{\eta _{1}}, \qquad \eta _{2}(x, y)\leq M_{\eta _{2}} $$
$$\begin{aligned}& \begin{gathered}[b] \varPsi \biggl(\frac{a+b}{2}, \frac{c+d}{2} \biggr)- \frac{M_{\eta _{1}}+M _{\eta _{2}}}{2} \\ \quad \leq \frac{1}{2} \biggl[\frac{1}{b-a} \int _{a}^{b}\varPsi \biggl(x, \frac{c+d}{2} \biggr)\,dx +\frac{1}{d-c} \int _{c}^{d}\varPsi \biggl(\frac{a+b}{2}, y \biggr)\,dy \biggr]-\frac{M _{\eta _{1}}+M_{\eta _{2}}}{4} \\ \quad \leq \frac{1}{(b-a)(d-c)} \int _{c}^{d} \int _{a}^{b}\varPsi (x, y)\,dx\,dy \\ \quad \leq \frac{1}{4} \biggl[\frac{1}{b-a} \int _{a}^{b} \bigl(\varPsi (x, c)+ \varPsi (x, d) \bigr)\,dx+ \frac{1}{d-c} \int _{c}^{d} \bigl(\varPsi (a, y)+ \varPsi (b, y) \bigr)\,dy \biggr]\\ \qquad {}+\frac{M_{\eta _{1}}+M_{\eta _{2}}}{4} \\ \quad \leq \frac{1}{4} \bigl[\varPsi (a, c)+\varPsi (b, c)+\varPsi (a, d)+\varPsi (b, d) \bigr]+ \frac{5}{4} [M_{\eta _{1}}+M_{\eta _{2}} ]. \end{gathered} \end{aligned}$$
(2.22)
Proof
For any fixed \(x\in [a, b]\), \(\varPsi _{x}(y)=\varPsi (x, y)\) is \(\eta _{2}\)-convex on \([c, d]\) due to Ψ is coordinate \((\eta _{1}, \eta _{2})\)-convex on \([a, b]\times [c, d]\). It follows from [77, Theorem 5] that
Integrating each side of inequality (2.23) with respect to the variable x on \([a, b]\) leads to
By similar arguments we have
Adding (2.24) and (2.25) we get the second and third inequalities of (2.22).
$$ \varPsi \biggl(x, \frac{c+d}{2} \biggr)- \frac{M_{\eta _{2}}}{2}\leq \frac{1}{d-c} \int _{c}^{d}\varPsi (x, y)\,dy \leq \frac{\varPsi (x, c)+\varPsi (x, d)}{2}+\frac{M_{\eta _{2}}}{2}. $$
(2.23)
$$\begin{aligned}& \frac{1}{b-a} \int _{a}^{b}\varPsi \biggl(x, \frac{c+d}{2} \biggr)\,dx-\frac{M _{\eta _{2}}}{2} \\& \quad \leq \frac{1}{(b-a)(d-c)} \int _{c}^{d} \int _{a}^{b}\varPsi (x, y)\,dx\,dy \\& \quad \leq \frac{1}{2(b-a)} \int _{a}^{b} \bigl[\varPsi (x, c)+\varPsi (x, d) \bigr]\,dx+\frac{M _{\eta _{2}}}{2}. \end{aligned}$$
(2.24)
$$\begin{aligned}& \frac{1}{d-c} \int _{c}^{d}\varPsi \biggl(\frac{a+b}{2}, y \biggr)\,dy-\frac{M _{\eta _{1}}}{2} \\& \quad \leq \frac{1}{(b-a)(d-c)} \int _{c}^{d} \int _{a}^{b}\varPsi (x, y)\,dx\,dy \\& \quad \leq \frac{1}{2(d-c)} \int _{c}^{d} \bigl[\varPsi (a, y)+\varPsi (b, y) \bigr]\,dy+\frac{M _{\eta _{1}}}{2}. \end{aligned}$$
(2.25)
Making use of the \((\eta _{1}, \eta _{2})\)-convexity of the function Ψ on \([a, b]\times [c, d]\) and [88, Theorem 5] again we get
$$\begin{aligned}& \varPsi \biggl(\frac{a+b}{2}, \frac{c+d}{2} \biggr)- \frac{M_{\eta _{2}}}{2}\leq \frac{1}{b-a} \int _{a}^{b}\varPsi \biggl(x, \frac{c+d}{2} \biggr)\,dx, \end{aligned}$$
(2.26)
$$\begin{aligned}& \varPsi \biggl(\frac{a+b}{2}, \frac{c+d}{2} \biggr)- \frac{M_{\eta _{1}}}{2}\leq \frac{1}{d-c} \int _{c}^{d}\varPsi \biggl(\frac{a+b}{2}, y \biggr)\,dy, \end{aligned}$$
(2.27)
$$\begin{aligned}& \frac{1}{b-a} \int _{a}^{b}\varPsi (x, c)\,dx\leq \frac{\varPsi (a, c)+\varPsi (b, c)}{2}+\frac{M_{\eta _{2}}}{2}, \end{aligned}$$
(2.28)
$$\begin{aligned}& \frac{1}{b-a} \int _{a}^{b}\varPsi (x, d)\,dx\leq \frac{\varPsi (a, d)+\varPsi (b, d)}{2}+\frac{M_{\eta _{2}}}{2}, \end{aligned}$$
(2.29)
$$\begin{aligned}& \frac{1}{d-c} \int _{c}^{d}\varPsi (a, y)\,dy\leq \frac{\varPsi (a, c)+\varPsi (a, d)}{2}+\frac{M_{\eta _{1}}}{2}, \end{aligned}$$
(2.30)
$$\begin{aligned}& \frac{1}{d-c} \int _{c}^{d}\varPsi (b, y)\,dy\leq \frac{\varPsi (b, c)+\varPsi (b, d)}{2}+\frac{M_{\eta _{1}}}{2}. \end{aligned}$$
(2.31)
Therefore, the first inequality of (2.22) follows from (2.26) and (2.27) with adding \(-\frac{1}{2}M_{\eta _{2}}\) and \(-\frac{1}{2}M_{\eta _{1}}\) respectively, and the last inequality in (2.22) can be derived from (2.28)–(2.31) immediately, with adding \(\frac{1}{4} [M_{\eta _{1}}+M_{\eta _{2}} ]\). □
3 Results and discussion
In the article, we establish a non-trivial example for a η-convex function defined on rectangle is not convex, prove that every η-convex function defined on rectangle is coordinate η-convex and its converse is not true in general. Furthermore, we define a new class of function which is coordinate \((\eta _{1}, \eta _{2})\)-convex function and prove its well-known Hermite–Hadamard type inequality.
4 Conclusion
We find an example for η-convex function defined on rectangle is not convex. The authors define a coordinate \((\eta _{1}, \eta _{2})\)-convex function and prove its results. Our approach may have further applications in the theory of η-convexity.
Availability of data and materials
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided 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.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.