Skip to main content

Optimality and duality results for a nondifferentiable multiobjective fractional programming problem

Abstract

The purpose of this paper is to study a class of nondifferentiable multiobjective fractional programming problems in which every component of objective functions contains a term involving the support function of a compact convex set. For a differentiable function, we introduce the definition of higher-order \((C,\alpha,\gamma,\rho,d)\)-convex function. A nontrivial example is also constructed which is in this class but not \((F,\alpha,\gamma,\rho,d)\)-convex. Based on the \((C,\alpha,\gamma,\rho,d)\)-convexity, sufficient optimality conditions for an efficient solution of the nondifferentiable multiobjective fractional programming problem are established. Further, a higher-order Mond-Weir type dual is formulated for this problem and appropriate duality results are proved under higher-order \((C,\alpha,\gamma,\rho,d)\)-assumptions.

Introduction

Higher-order duality is significant due to its computational importance as it provides higher bounds whenever an approximation is used. By introducing two different functions, \(h: R^{n}\times R^{n}\rightarrow R\) and \(k: R^{n}\times R^{n}\rightarrow R^{m}\), Mangasarian [1] formulated a higher-order dual for a nonlinear optimization problem, \(\{ \min f(x), \mbox{ subject to } g(x)\leq0\}\). Inspired by this concept, many researchers have worked in this direction. Chen [2] has formulated higher-order multiobjective symmetric dual programs and established duality relations under higher-order F-convexity assumptions. A higher-order vector optimization problem and its dual have been studied by Kassem [3].

In the last several years, various optimality and duality results have been obtained for multiobjective fractional programming problems. In Chen [2], multiobjective fractional problem and its duality theorems have been considered under higher-order \((F,\alpha,\rho ,d)\)-convexity. Later on, Suneja et al. [4] discussed higher-order Mond-Weir and Schaible type nondifferentiable dual programs and their duality theorems under higher-order \((F,\rho,\sigma )\)-type I-assumptions. Recently, Ying [5] has studied higher-order multiobjective symmetric fractional problem and formulated its Mond-Weir type dual. Further, duality results are obtained under higher-order \((F,\alpha,\rho,d)\)-convexity.

Yuan et al. [6] introduced a class of functions called \((C,\alpha,\rho,d)\)-convex functions and derived duality theorems for a nondifferentiable minimax fractional programming problem under \((C,\alpha,\rho,d)\)-convexity. Chinchuluun et al. [7] later studied nonsmooth multiobjective fractional programming problems in the framework of \((C,\alpha,\rho,d)\)-convexity.

In this paper, we first introduce the definition of higher-order \((C,\alpha,\gamma,\rho,d)\)-convex with respect to a differentiable function \(H:X\times R^{n}\longrightarrow R \) (\(X\subset R^{n}\)), \(p, s\in R^{n}\) and construct a nontrivial example which is higher-order \((C,\alpha ,\gamma,\rho,d)\)-convex but not a \((F,\alpha,\gamma,\rho,d)\)-convex function. We prove that the ratio of higher-order \((C,\alpha,\gamma,\rho ,d)\)-convex functions is also higher-order \((\bar{C},\bar{\alpha},\bar {\gamma},\bar{\rho},\bar{d})\)-convex. A sufficient optimality condition related to the efficient solution of a multiobjective nondifferentiable fractional problem has been established. Finally, we formulate a higher-order Mond-Weir type dual problem corresponding to the multiobjective nondifferentiable fractional programming problem and established usual duality relations under the aforesaid assumptions.

Preliminaries

Definition 2.1

A function \(C:X\times X\times R^{n}\longrightarrow R\) (\(X \subset R^{n}\)) is said to be convex on \(R^{n}\) iff, for any fixed \((x, u)\in X\times X\) and for any \(x_{1}, x_{2}\in R^{n}\),

$$C_{x,u}\bigl(\lambda x_{1}+(1-\lambda)x_{2}\bigr) \leq\lambda C_{x,u}(x_{1})+(1-\lambda)C_{x,u}(x_{2}), \quad \forall \lambda\in(0,1). $$

We now introduce the definition of higher-order \((C,\alpha,\gamma,\rho ,d)\)-convex function. Let C be a convex function with respect to the third variable such that \(C_{x,u}(0)=0\), \(\forall (x,u)\in X\times X\). Let \(H: X\times R^{n}\rightarrow R\), \(\phi: X\rightarrow R\) be differentiable functions on X. Assume that \(\alpha, \gamma: X\times X\rightarrow R_{+}\backslash\{0\}\), \(\rho\in R\), \(d : X\times X\rightarrow R_{+}\) satisfying \(d(x,x_{0})=0 \Leftrightarrow x=x_{0}\) and \(p,s\in R^{n}\).

Definition 2.2

The function ϕ is said to be higher-order (strictly) \((C,\alpha,\gamma,\rho,d)\)-convex at u with respect to H, p and s if for each \(x\in X\),

$$\begin{aligned} \frac{1}{\alpha(x,u)}\bigl[\phi(x)-\phi(u)\bigr] \geq& (>) C_{x,u} \bigl( \nabla\phi(u)+\nabla_{p} H(u,p) \bigr)\\ &{}+\frac{1}{\gamma(x,u)} \bigl[H(u,s)-s^{T} \nabla_{s}H(u,s)\bigr] +\frac{\rho d(x,u)}{\alpha(x,u)}. \end{aligned}$$

Remark 2.1

A differentiable function \(f = (f_{1},f_{2},\ldots,f_{k}): X\rightarrow R^{k}\) is \((C,\alpha,\gamma,\rho,d)\)-convex if for all \(i=1,2,\ldots,k\), \(f_{i}\) is \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex.

Remark 2.2

  1. (i)

    If \(H(u,\cdot)=0\), then Definition 2.2 becomes that of a \((C,\alpha,\rho,d)\)-convex function as given in [6, 8]. Further if \(\alpha(x,u)=1\), \(\rho=0\), and \(C_{x,u}(a)=\eta^{T}(x,u)a\), for \(\eta:X\times X\rightarrow R^{n}\), then \((C,\alpha,\gamma,\rho ,d)\)-convexity reduces to invexity (see Hanson [9]).

  2. (ii)

    If \(H(u,\cdot)= \frac{1}{2}(\cdot)^{T}\nabla^{2} f(u)(\cdot)\), \(\alpha(x,u)=\gamma(x,u)\), and \(p=s\), then Definition 2.2 reduces to the first expression of a second-order \((C,\alpha,\rho ,d)\)-type I-convex function given in Gupta et al. [10].

  3. (iii)

    If \(\alpha(x,u)=\gamma(x,u)=1\), \(\rho =0\), \(H(u,\cdot)= \frac{1}{2}(\cdot)^{T}\nabla^{2} f(u)(\cdot)\), \(p=s\), and \(C_{x,u}(a)=\eta^{T}(x,u)a\), where \(\eta: X\times X\rightarrow R^{n}\), the above definition becomes that of η-bonvexity introduced by Pandey [11].

  4. (iv)

    If C is sublinear with respect to third variable and \(p=s\), then the above definition reduces to higher-order \((F,\alpha ,\gamma,\rho,d)\)-convexity as given in Gulati and Saini [12]. Furthermore, if \(\gamma(x,u)=\alpha(x,u)\), then Definition 2.2 reduces to higher-order \((F,\alpha,\rho,d)\)-convexity as in Ying [5]. Moreover, if \(H(u,\cdot)=0\), then Definition 2.2 becomes that of a \((F,\alpha,\rho ,d)\)-convex function as introduced by Liang et al. [13].

Remark 2.3

Every \((F,\alpha,\gamma,\rho,d)\)-convex function is \((C,\alpha,\gamma,\rho,d)\)-convex. However, the converse need not be true. This is illustrated by the following example.

Example 2.1

Let \(X = \{x : x\geq1\}\subset R\), \(f:X\rightarrow R\), \(C:X\times X\times R\rightarrow R\), \(H: X\times R\rightarrow R\), and \(d: X\times X\rightarrow R_{+}\) be defined as

$$f(x) = \frac{x^{2}+x+2}{x+1},\qquad C_{x,u}(a)= a^{2}(x-u)^{2}, \qquad H(u,\cdot)=\frac{(\cdot)}{u+1},\qquad d(x,u)=(x-u)^{2}. $$

Clearly, the function C (defined above) is convex with respect to the third variable satisfying \(C_{x,u}(0)=0\), \(\forall x, u\in X\). Also, \(d(x,u)=0 \Leftrightarrow x=u\).

Let \(\rho=-1\) and \(\alpha=\gamma= \frac{3}{4}\). Then at \(u=1\), for all \(x\in X\), we have

$$\begin{aligned} \frac{1}{\alpha(x,u)}\bigl[f(x)-f(u)\bigr] =& \frac{4}{3} \biggl( \frac {x^{2}-x}{x+1} \biggr)\geq C_{x,u}\bigl(\nabla_{x}f(u)+ \nabla_{p} H(u,p)\bigr)\\ &{}+\frac {1}{\gamma(x,u)}\bigl[H(u,s)-s^{T} \nabla_{s}H(u,s)\bigr]+\frac{\rho d(x,u)}{\alpha (x,u)}=- \frac{1}{3}(x-1)^{2}. \end{aligned}$$

Hence, f is a higher-order \((C,\alpha,\gamma,\rho,d)\)-convex function with respect to H, p, and s. But considering the same C, f is not higher-order \((F,\alpha,\gamma,\rho,d)\)-convex because C is not a sublinear functional with respect to the third variable.

Definition 2.3

[14]

Let C be a compact convex set in \(R^{n}\). The support function of C is defined by

$$S(x|C) = \max\bigl\{ x^{T}y: y\in C\bigr\} . $$

Problem formulation and optimality conditions

Consider the following nondifferentiable multiobjective programming problem:

$$\textstyle\begin{array}{@{}l@{\quad}l@{\quad}l} (\mathbf{MFP}) & \mbox{Minimize} &\displaystyle F(x)= \biggl\{ \frac {f_{1}(x)+S(x|C_{1})}{g_{1}(x)-S(x|D_{1})}, \frac {f_{2}(x)+S(x|C_{2})}{g_{2}(x)-S(x|D_{2})},\ldots,\frac {f_{k}(x)+S(x|C_{k})}{g_{k}(x)-S(x|D_{k})} \biggr\} \\ & \mbox{subject to} & x\in X^{0} =\bigl\{ x\in X\subset R^{n}: h_{j}(x)+S(x|E_{j})\leq0, j=1,2,\ldots,m\bigr\} , \end{array} $$

where \(f_{i}, g_{i}: X\rightarrow R \) (\(i=1,2,\ldots,k\)) and \(h_{j}: X\rightarrow R \) (\(j=1,2,\ldots,m\)) are continuously differentiable functions. Assume that \(f_{i}(\cdot)+ S(\cdot|C_{i})\geq0\) and \(g_{i}(\cdot)- S(\cdot|D_{i})> 0\); \(C_{i}\), \(D_{i}\), and \(E_{j}\) are compact convex sets in \(R^{n}\) and \(S(x|C_{i})\), \(S(x|D_{i})\), and \(S(x|E_{j})\) denote the support functions of compact convex sets, \(C_{i}\), \(D_{i}\), and \(E_{j}\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\), respectively.

Definition 3.1

[15]

A point \(x^{0}\in X^{0}\) is a weakly efficient solution of (MFP), if there exists no \(x\in X^{0}\) such that for every \(i=1,2,\ldots,k\), \(\frac {f_{i}(x)+S(x|C_{i})}{g_{i}(x)-S(x|D_{i})}< \frac {f_{i}(x^{0})+S(x^{0}|C_{i})}{g_{i}(x^{0})-S(x^{0}|D_{i})}\).

Definition 3.2

[16]

A point \(x^{0}\in X^{0}\) is an efficient solution (or a Pareto optimal solution) of (MFP), if there exists no \(x\in X^{0}\) such that for every \(i =1,2,\ldots,k\), \(\frac{f_{i}(x)+S(x|C_{i})}{g_{i}(x)-S(x|D_{i})}\leq \frac {f_{i}(x^{0})+S(x^{0}|C_{i})}{g_{i}(x^{0})-S(x^{0}|D_{i})}\) and for some \(r = 1,2,\ldots,k\), \(\frac{f_{r}(x)+S(x|C_{r})}{g_{r}(x)-S(x|D_{r})}< \frac{f_{r}(x^{0})+S(x^{0}|C_{r})}{g_{r}(x^{0})-S(x^{0}|D_{r})}\).

Lemma 3.1

[17]

If for a given \(\bar{\lambda}_{i} > 0\) (\(\bar{\lambda}_{i}\geq0\) and \(\bar{\lambda} \neq 0\)), \(i=1,2,\ldots,k\), \(\bar{x}\in X\) is an optimal solution for the following single-objective problem:

$$\textstyle\begin{array}{@{}l@{\quad}l@{\quad}l} (\mathbf{FP}_{\lambda}) & \textit{minimize} & \displaystyle G(x) = \sum_{i=1}^{k}\bar{\lambda}_{i} \biggl(\frac {f_{i}(x)+S(x|C_{i})}{g_{i}(x)-S(x|D_{i})} \biggr)\\ &\textit{subject to} & x\in X \subset R^{n}, \end{array} $$

then is an efficient solution (a weakly efficient solution) for (MFP).

Theorem 3.1

For some t, assume \(f_{t}(\cdot)+(\cdot)^{T}z_{t}\) and \(-(g_{t}(\cdot)-(\cdot)^{T}v_{t})\) are \((C,\alpha_{t},\gamma_{t},\rho_{t},d_{t})\)-convex at \(u\in X\) with respect to \(H_{t}\), p, and s. Then \((\frac{f_{t}(\cdot)+ (\cdot)^{T}z_{t}}{g_{t}(\cdot)- (\cdot)^{T}v_{t}} )\) is also higher-order \((\bar{C},\bar{\alpha}_{t},\bar{\gamma}_{t},\bar{\rho}_{t},\bar{d}_{t})\)-convex at \(u\in X\) with respect to \(\bar{H}_{t}\), p, and s, where

$$\begin{aligned}& \bar {\alpha}_{t}(x,u)= \biggl(\frac {g_{t}(u)-u^{T}v_{t}}{g_{t}(x)-x^{T}v_{t}} \biggr) \alpha_{t}(x,u), \qquad\bar{\gamma} _{t}(x,u)= \gamma_{t}(x,u),\\& \bar{\rho}_{t}=\rho_{t} \biggl(1+\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr),\qquad \bar{H}_{t}(u,\cdot)= \biggl(\frac{1}{(g_{t}(u)-u^{T}v_{t})}+\frac {f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)H_{t}(u,\cdot),\\& \bar {d}_{t}(x,u)= \biggl(\frac{d_{t}(x,u)}{g_{t}(x)-x^{T}v_{t}} \biggr) , \end{aligned}$$

and

$$\begin{aligned}& \bar{C}_{x,u}(a)= \biggl(\frac{f_{t}(u)+u^{T}z_{t} +g_{t}(u)-u^{T} v_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)C_{x,u} \biggl( \frac{(g_{t}(u)-u^{T} v_{t} )^{2} a}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggr),\\& a= \nabla \biggl( \frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr)+\nabla_{p} \bar {H}_{t}(u,p). \end{aligned}$$

Proof

For any \(x,u\in X\),

$$\begin{aligned} \biggl(\frac{f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) =& \biggl(\frac {f_{t}(x)+x^{T}z_{t}-f_{t}(u)-u^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}} \biggr)\\ &{}-\bigl[f_{t}(u)+u^{T}z_{t}\bigr] \biggl( \frac {g_{t}(x)-x^{T}v_{t}-g_{t}(u)+u^{T}v_{t}}{(g_{t}(x)-x^{T}v_{t})(g_{t}(u)-u^{T}v_{t})} \biggr). \end{aligned}$$

Since \(f_{t}(\cdot)+(\cdot)^{T}z_{t}\) and \(-(g_{t}(\cdot)-(\cdot)^{T}v_{t})\) are \((C,\alpha_{t},\gamma _{t},\rho_{t},d_{t})\)-convex at \(u\in X\) with respect to \(H_{t}\), p, and s, we have

$$\begin{aligned} &\frac{1}{\alpha_{t}(x,u)} \biggl(\frac {f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\quad\geq \frac{1}{(g_{t}(x)-x^{T}v_{t})} \biggl\{ C_{x,u} \bigl(\nabla \bigl(f_{t}(u)+u^{T}z_{t} \bigr)+\nabla_{p} H_{t}(u,p) \bigr) \\ &\qquad{}+\frac{1}{\gamma _{t}(x,u)} \bigl[H_{t}(u,s)-s^{T}\nabla_{s} H_{t}(u,s)\bigr]+\frac {\rho_{t} d_{t}(x,u)}{\alpha_{t}(x,u)} \biggr\} \\ &\qquad{}+\frac {f_{t}(u)+u^{T}z_{t}}{(g_{t}(x)-x^{T}v_{t})(g_{t}(u)-u^{T}v_{t})} \biggl\{ C_{x,u} \bigl(-\nabla\bigl(g_{t}(u)-u^{T}v_{t}\bigr)+ \nabla_{p}H_{t}(u,p) \bigr) \\ &\qquad{}+\frac {1}{\gamma_{t}(x,u)}\bigl[H_{t}(u,s)-s^{T} \nabla_{s} H_{t}(u,s)\bigr]+\frac{\rho_{t} d_{t}(x,u)}{\alpha_{t}(x,u)} \biggr\} , \end{aligned}$$

which implies

$$\begin{aligned} &\frac{1}{\alpha_{t}(x,u)} \biggl(\frac {f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\quad\geq \biggl(\frac {f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}}{(g_{t}(x)-x^{T}v_{t})(g_{t}(u)-u^{T}v_{t})} \biggr) \biggl(\frac{g_{t}(u)-u^{T}v_{t}}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\qquad{}\times\biggl\{ C_{x,u} \bigl(\nabla\bigl(f_{t}(u)+u^{T}z_{t} \bigr)+\nabla _{p}H_{t}(u,p) \bigr)\\ &\qquad{}+\frac{1}{\gamma_{t}(x,u)} \bigl[H_{t}(u,s)-s^{T}\nabla_{s} H_{t}(u,s)\bigr]+\frac{\rho_{t} d_{t}(x,u)}{\alpha_{t}(x,u)} \biggr\} \\ &\qquad{}+ \biggl(\frac{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}}{(g_{t}(x)-x^{T}v_{t})(g_{t}(u)-u^{T}v_{t})} \biggr) \biggl(\frac{f_{t}(u)+u^{T}z_{t}}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\qquad{}\times \biggl\{ C_{x,u} \bigl(-\nabla\bigl(g_{t}(u)-u^{T}v_{t} \bigr)+\nabla_{p}H_{t}(u,p) \bigr)\\ &\qquad{}+\frac{1}{\gamma _{t}(x,u)} \bigl[H_{t}(u,s)-s^{T}\nabla_{s} H_{t}(u,s)\bigr]+\frac{\rho_{t} d_{t}(x,u)}{\alpha _{t}(x,u)} \biggr\} . \end{aligned}$$

This further yields

$$\begin{aligned} &\frac{1}{\alpha_{t}(x,u)} \biggl(\frac {f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr)\\ &\quad\geq \biggl(\frac {f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}}{(g_{t}(x)-x^{T}v_{t})(g_{t}(u)-u^{T}v_{t})} \biggr) \\ &\qquad{}\times\biggl\{ C_{x,u} \biggl( \frac {(g_{t}(u)-u^{T}v_{t})^{2}}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggl(\nabla \biggl(\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\qquad{}+ \biggl(\frac{1}{g_{t}(u)-u^{T}v_{t}}+\frac{f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)\nabla_{p} H_{t}(u,p) \biggr) \biggr) \biggr\} +\frac{1}{\gamma_{t}(x,u)(g_{t}(x)-x^{T}v_{t})} \\ &\qquad{}\times \biggl[ \biggl(1+\frac {f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})} \biggr) \bigl(H_{t}(u,s)-s^{T} \nabla _{s}H_{t}(u,s)\bigr) \biggr]\\ &\qquad{}+\frac{\rho_{t} d_{t}(x,u)}{\alpha_{t}(x,u)(g_{t}(x)-x^{T}v_{t})} \biggl(1+\frac{f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})} \biggr). \end{aligned}$$

Multiplying by \((\frac {g_{t}(x)-x^{T}v_{t}}{g_{t}(u)-u^{T}v_{t}} )\), the above inequality gives

$$\begin{aligned} &\frac{(g_{t}(x)-x^{T}v_{t})}{\alpha_{t}(x,u)(g_{t}(u)-u^{T}v_{t})} \biggl(\frac{f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\quad\geq \biggl(\frac {f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr) \\ &\qquad{}\times\biggl\{ C_{x,u} \biggl( \frac{(g_{t}(u)-u^{T}v_{t})^{2}}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggl(\nabla \biggl(\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr)\\ &\qquad{}+ \biggl( \frac {1}{g_{t}(u)-u^{T}v_{t}}+\frac{f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)\nabla_{p} H_{t}(u,p) \biggr) \biggr) \biggr\} \\ &\qquad{}+\frac{1}{\gamma_{t}(x,u)(g_{t}(u)-u^{T}v_{t})} \biggl[ \biggl(1+\frac{f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})} \biggr) \bigl(H_{t}(u,s)-s^{T}\nabla _{s}H_{t}(u,s) \bigr) \biggr]\\ &\qquad{}+\frac{\rho_{t} d_{t}(x,u)}{\alpha_{t}(x,u)(g_{t}(u)-u^{T}v_{t})} \biggl(1+\frac{f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})} \biggr). \end{aligned}$$

Setting

$$\begin{aligned}& \bar{\alpha}_{t}(x,u)= \biggl(\frac {g_{t}(u)-u^{T}v_{t}}{g_{t}(x)-x^{T}v_{t}} \biggr) \alpha_{t}(x,u), \qquad\bar{\gamma} _{t}(x,u)= \gamma_{t}(x,u), \\& \bar{\rho}_{t}=\rho_{t} \biggl(1+\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr), \\& \bar{H}_{t}(u,\cdot)= \biggl(\frac{1}{(g_{t}(u)-u^{T}v_{t})}+\frac {f_{t}(u)+u^{T}z_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)H_{t}(u,\cdot),\qquad \bar{d}_{t}(x,u)= \biggl( \frac{d_{t}(x,u)}{g_{t}(x)-x^{T}v_{t}} \biggr), \\& \bar{C}_{(x,u)}(a)= \biggl(\frac{f_{t}(u)+u^{T}z_{t} +g_{t}(u)-u^{T} v_{t}}{(g_{t}(u)-u^{T}v_{t})^{2}} \biggr)C_{x,u} \biggl( \frac{(g_{t}(u)-u^{T} v_{t})^{2} a}{f_{t}(u)+u^{T}z_{t}+g_{t}(u)-u^{T}v_{t}} \biggr), \end{aligned}$$

and

$$a= \nabla \biggl(\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr)+\nabla_{p} \bar{H}_{t}(u,p). $$

It follows that

$$\begin{aligned} &\frac{1}{\bar{\alpha}_{t}(x,u)} \biggl(\frac {f_{t}(x)+x^{T}z_{t}}{g_{t}(x)-x^{T}v_{t}}-\frac{f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr) \\ &\quad\geq \biggl\{ \bar{C}_{(x,u)} \biggl(\nabla \biggl(\frac {f_{t}(u)+u^{T}z_{t}}{g_{t}(u)-u^{T}v_{t}} \biggr)+ \nabla_{p} \bar{H}_{t}(u,p) \biggr) \\ &\qquad{} +\frac {1}{\bar{\gamma}_{t}(x,u)}\bigl[\bar{H}_{t}(u,s)-s^{T} \nabla_{s}\bar{H}_{t}(u,s)\bigr]+\frac {\bar{\rho}_{t} \bar{d}_{t}(x,u)}{\bar{\alpha}_{t}(x,u)} \biggr\} . \end{aligned}$$

Hence, \((\frac{f_{t}(\cdot)+(\cdot)^{T}z_{t}}{g_{t}(\cdot)-(\cdot)^{T} z_{t}} )\) is higher-order \((\bar{C},\bar{\alpha}_{t},\bar{\gamma}_{t},\bar{\rho} _{t},\bar{d}_{t})\)-convex at \(u\in X\) with respect to \(\bar{H}_{t}\), p, and s. □

Theorem 3.2

Let \(X\subset R^{n}\) be an open convex set. Let \(\psi_{i} : X\rightarrow R\) be higher-order \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex at one point in X with respect to \(\phi_{i}, \rho_{i}\geq0\), and for all \(x\in X\), \(\phi_{i}(x,\cdot)=-(\cdot)^{T}\nabla\psi_{i}(x)\), \(i=1,2,\ldots,k\); then only one of the following two cases holds:

  1. (i)

    there exists \(x\in X\) such that \(\psi_{i}(x)<0\), \(i=1,2,\ldots,k\);

  2. (ii)

    there exists \(\lambda\in R_{+}^{k}\setminus\{0\}\) such that \(\sum_{i=1}^{k}\lambda_{i} \psi_{i}(x)\geq0\), for all \(x\in X\).

Proof

If (i) has a solution, that is, there exists \(x\in X\) such that \(\psi_{i}(x)< 0\), \(i=1,2,\ldots,k\), then for every \(\lambda\in R_{+}^{k}\setminus\{0\}\), we have \(\sum_{i=1}^{k}\lambda_{i} \psi_{i}(x)< 0\), which implies that (ii) does not hold.

If (i) has no solution, let \(K = \psi(X)+ \operatorname{int}(R_{+}^{k})\), \(\psi= (\psi_{1},\psi_{2},\ldots,\psi_{k})^{T}\), then \(K\cap(-R_{+}^{k})=\phi\). For any \(z^{1}, z^{2}\in K\) and \(\beta\in(0,1)\), there exist \(x^{1}, x^{2}\in X\) and \(s^{1}, s^{2}\in \operatorname{int}(R_{+}^{k})\) such that

$$\beta z^{1} + (1-\beta)z^{2}=\beta\psi\bigl(x^{1} \bigr)+(1-\beta)\psi\bigl(x^{2}\bigr)+\beta s^{1}+ (1- \beta)s^{2}. $$

Since \(u=\beta x^{1}+(1-\beta)x^{2}\in X\) and \(\phi_{i}(x,\cdot)=-(\cdot)^{T}\nabla\psi _{i}(x)\), it follows from the higher-order \((C,\alpha_{i},\gamma_{i},\rho _{i},d_{i})\)-convexity of \(\psi_{i}\) with respect to \(\phi_{i}\), p, and s that we have

$$\begin{aligned} &\frac{1}{\alpha_{i}(x^{j},u)}\bigl[\psi_{i}\bigl(x^{j}\bigr)- \psi_{i}(u)\bigr]\\ &\quad\geq C_{x,u}\bigl[\nabla \psi_{i}(u)+\nabla_{p}\phi_{i}(u,p)\bigr]+ \frac {1}{\gamma_{i}(x^{j},u)}\bigl\{ \phi_{i}(u,s)-s^{T}\bigl[ \nabla_{s}\phi_{i}(u,s)\bigr]\bigr\} + \frac{\rho_{i} d(x^{j},u)}{\alpha_{i}(x^{j},u)} \\ &\quad= \frac{1}{\alpha _{i}(x^{j},u)}\rho_{i} d\bigl(x^{j},u\bigr),\quad j=1,2. \end{aligned}$$

Thus, there exists \(t\geq0\) such that

$$\beta\psi_{i}\bigl(x^{1}\bigr)+(1-\beta)\psi_{i} \bigl(x^{2}\bigr)=\psi_{i}\bigl(\beta x^{1} + (1- \beta)x^{2}\bigr)+t, $$

which implies that

$$\beta z^{1} + (1-\beta)z^{2}=\beta\psi\bigl(x^{1} \bigr)+(1-\beta)\psi\bigl(x^{2}\bigr)+\beta s^{1}+ (1- \beta)s^{2}+(t,t,\ldots,t)^{T}. $$

Since \(R_{+}^{k}\) is a closed convex cone, we have \(\beta z_{1} + (1-\beta )z_{2}\in K\), that is, K is an open convex set. It follows from the convex separated theorem that there exists \(\lambda\in R_{+}^{k}\setminus\{0\}\) such that, for all \(x\in X\), \(\sum_{i=1}^{k}\lambda_{i} \psi_{i}(x)\geq0\). □

We now discuss some optimality conditions for the problem (MFP).

Theorem 3.3

If \(u\in X\) is a weakly efficient solution of (MFP), \(f_{i}(\cdot)+(\cdot)^{T}z_{i} \), and \(-(g_{i}(\cdot)-(\cdot)^{T}v_{i})\) are higher-order \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex functions at \(u\in X\) with respect to \(\phi_{i}(u,\cdot)= -\frac {(g_{i}(u)-u^{T}v_{i})^{2}}{(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}\times(\cdot)^{T}\nabla (\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} )\), and \(\rho_{i}\geq 0\), \(i=1,2,\ldots,k\). Then there exists \(0 \leq\bar{\lambda}\in R^{k}\), \(\bar {\lambda}\neq0\) such that u is an optimal solution of \((\mathrm{FP}_{\bar {\lambda}})\).

Proof

If \(u\in X\) is a weakly efficient solution of (MFP), then there does not exist any \(x\in X\) such that

$$\psi_{j}(x) = \biggl(\frac{f_{j}(x)+x^{T}z_{j}}{g_{j}(x)-x^{T}v_{j}}-\frac {f_{j}(u)+u^{T}z_{j}}{g_{j}(u)-u^{T}v_{j}} \biggr)< 0, \quad j=1,2,\ldots,k. $$

Since \(f_{j}(\cdot)+(\cdot)^{T}z_{j} \) and \(-(g_{j}(\cdot)-(\cdot)^{T}v_{j})\) are higher-order \((C,\alpha_{j},\gamma_{j},\rho_{j},d_{j})\)-convex at u with respect to \(\phi _{j}\), p and \(s, j=1,2,\ldots,k\), it follows from Theorem 3.1 that \(\psi _{j}(x) = ( \frac{f_{j}(x)+x^{T}z_{j}}{g_{j}(x)-x^{T}v_{j}}-\frac {f_{j}(u)+u^{T}z_{j}}{g_{j}(u)-u^{T}v_{j}} )\) is higher-order \((\bar{C},\bar {\alpha}_{j},\bar{\gamma}_{j}, \bar{\rho}_{j}, \bar{d}_{j})\)-convex at u with respect to \(\bar{\phi}_{j}\), p, and \(s, j=1,2,\ldots,k\), where

$$\begin{aligned}& \bar{\alpha}_{j}(x,u)= \biggl(\frac {g_{j}(x)-x^{T}v_{j}}{g_{j}(u)-u^{T}v_{j}} \biggr) \alpha_{j}(x,u), \qquad\bar{\gamma }_{j}(x,u)= \gamma_{j}(x,u), \\& \bar{\rho}_{j}=\rho_{j} \biggl(1+\frac{f_{j}(u)+u^{T}z_{j}}{g_{j}(u)-u^{T}v_{j}} \biggr),\\& \begin{aligned}[b] \bar{\phi }_{j}(u,\cdot)&= \biggl(\frac{1}{g_{j}(u)-u^{T}v_{j}}+\frac {f_{j}(u)+u^{T}z_{j}}{(g_{j}(u)-u^{T}v_{j})^{2}} \biggr)\phi_{j}(u,\cdot)=- (\cdot)^{T}\nabla \biggl( \frac{f_{j}(u)+u^{T}z_{j}}{g_{j}(u)- u^{T}v_{j}} \biggr) \\ &=-(\cdot)^{T}\nabla\psi_{j}(u), \end{aligned}\\& \bar{d}_{j}(x,u)= \biggl( \frac {d_{j}(x,u)}{g_{j}(x)-x^{T}v_{j}} \biggr), \end{aligned}$$

and

$$\begin{aligned}& \bar{C}_{x,u}(a)= \frac{f_{j}(u)+u^{T}z_{j} +g_{j}(u)-u^{T} v_{j}}{(g_{j}(u)-u^{T}v_{j})^{2}}C_{x,u} \biggl( \frac{(g_{j}(u)-u^{T} v_{j})^{2} a}{f_{j}(u)+u^{T}z_{j} + g_{j}(u)-u^{T}v_{j}} \biggr),\\& a= \nabla \biggl(\frac{f_{j}(u)+u^{T}z_{j}}{g_{j}(u)-u^{T}v_{j}} \biggr)+ \nabla_{p} \bar{\phi }_{j}(u,p). \end{aligned}$$

Using Theorem 3.2, there exists \(0 \leq\bar{\lambda}\in R^{k}\), \(\bar {\lambda}\neq0\) such that \(\sum_{j=1}^{k}\bar{\lambda}_{j} \psi_{j}(x)\geq0\), that is,

$$\sum_{j=1}^{k}\bar{\lambda}_{j} \biggl(\frac {f_{j}(x)+x^{T}z_{j}}{g_{j}(x)-x^{T}v_{j}} \biggr)\geq\sum_{j=1}^{k} \bar{\lambda }_{j} \biggl(\frac{f_{j}(u)+u^{T}z_{j}}{g_{j}(u)-u^{T}v_{j}} \biggr), $$

which implies that u is an optimal solution of \((\mathrm{FP}_{\bar{\lambda}})\). □

Theorem 3.4

(Necessary condition) [18]

Assume that is an efficient solution of (MFP) and the Slater constraint qualification is satisfied on X. Then there exist \(\bar{\lambda}\in R^{k}\), \(\bar{\mu}\in R^{m}\), \(\bar{z}_{i}\in R^{n}\), \(\bar{v}_{i}\in R^{n}\), and \(\bar{w}_{j}\in R^{n}\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\), such that

$$\begin{aligned}& \sum_{i=1}^{k}\bar{\lambda}_{i} \nabla \biggl(\frac{f_{i}(\bar {x})+\bar{x}^{T}\bar{z}_{i}}{g_{i}(\bar{x})-\bar{x}^{T}\bar{v}_{i}} \biggr)+ \sum_{j=1}^{m} \bar{\mu}_{j}\nabla\bigl(h_{j}(\bar{x})+\bar {x}^{T}\bar{w}_{j}\bigr)=0, \\& \sum_{j=1}^{m}\bar{\mu}_{j} \bigl(h_{j}(\bar{x})+\bar{x}^{T}\bar{w}_{j} \bigr)=0, \\& \bar{x}^{T}\bar{z}_{i}=S(\bar{x}|C_{i}), \quad i=1,2,\ldots,k, \\& \bar{x}^{T}\bar{v}_{i}=S(\bar{x}|D_{i}), \quad i=1,2,\ldots,k, \\& \bar{x}^{T}\bar{w}_{j}=S(\bar{x}|E_{j}),\quad j=1,2,\ldots,m, \\& \bar{z}_{i}\in C_{i}, \qquad \bar{v}_{i}\in D_{i}, \qquad \bar{w}_{j}\in E_{j}, \quad i=1,2,\ldots,k, j=1,2,\ldots,m, \\& \bar{\lambda}_{i}> 0, \qquad \bar{\mu}_{j}\geq0, \quad i=1,2,\ldots,k, j=1,2,\ldots,m. \end{aligned}$$

Theorem 3.5

(Sufficient condition)

Let u be a feasible solution of (MFP). Assume that there exist \(\lambda_{i} >0\), \(i=1,2,\ldots,k\) and \(\mu_{j}\geq0\), \(j=1,2,\ldots,m\), such that

$$\begin{aligned}& \sum_{i=1}^{k}\lambda_{i}\nabla \biggl(\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+ \sum_{j=1}^{m} \mu_{j}\nabla\bigl(h_{j}(u)+u^{T}w_{j} \bigr)=0, \\& \sum_{j=1}^{m}\mu_{j} \bigl(h_{j}(u)+u^{T}w_{j}\bigr)=0, \\& u^{T}z_{i}=S(u|C_{i}),\quad i=1,2,\ldots,k, \\& u^{T}v_{i}=S(u|D_{i}),\quad i=1,2,\ldots,k, \\& u^{T}w_{j}=S(u|E_{j}),\quad j=1,2,\ldots,m, \\& z_{i}\in C_{i},\qquad v_{i}\in D_{i},\qquad w_{j}\in E_{j}, \quad i=1,2, \ldots,k, j=1,2,\ldots,m. \end{aligned}$$

Let for any \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\),

  1. (i)

    \((f_{i}(\cdot)+(\cdot)^{T}z_{i} )\) and \(- (g_{i}(\cdot)-(\cdot)^{T}v_{i} )\) be higher-order \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex at u, with respect to \(H_{i}\), p, and s,

  2. (ii)

    \((h_{j}(\cdot)+(\cdot)^{T}w_{j} )\) be higher-order \((C,\xi _{j},\delta_{j},\eta_{j},c_{j})\)-convex at u, with respect to \(K_{j}\), q, and r,

  3. (iii)

    \(\sum_{i=1}^{k} \lambda_{i}\bar{\rho}_{i} \frac {\bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)} + \sum_{j=1}^{m} \mu _{j}\eta_{j}\frac{c_{j}(x,u)}{\xi_{j}(x,u)}\geq0\),

  4. (iv)

    \(\bar{\gamma}_{i}(x,u)=\zeta(x,u)\), \(\xi_{j}(x,u)=\sigma (x,u)\), and \(\delta_{j}(x,u)=\sigma^{\prime}(x,u)\),

  5. (v)

    \(\sum_{i=1}^{k} \lambda_{i} (\nabla_{p} \bar {H}_{i}(u,p) ) + \sum_{j=1}^{m} \mu_{j} (\nabla_{q} K_{j}(u,q) )=0\), \(\sum_{i=1}^{k}\lambda_{i} (\bar {H}_{i}(u,s)-s^{T} \nabla_{s} \bar{H}_{i}(u,s) )\geq0 \) and \(\sum_{j=1}^{m} \mu_{j} (K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r) )\geq0\),

where

$$\begin{aligned}& \bar{\alpha}_{i}(x,u)= \biggl(\frac {g_{i}(u)-u^{T}v_{i}}{g_{i}(x)-x^{T}v_{i}} \biggr) \alpha_{i}(x,u), \bar{\gamma} _{i}(x,u)= \gamma_{i}(x,u), \qquad\bar{\rho}_{i}=\rho_{i} \biggl(1+\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr), \\& \bar{H}_{i}(u,\cdot)= \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}}+\frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr) H_{i}(u,\cdot),\qquad \bar {d}_{i}(x,u)= \biggl( \frac{d_{i}(x,u)}{g_{i}(x)-x^{T}v_{i}} \biggr). \end{aligned}$$

Then u is an efficient solution of (MFP).

Proof

Suppose u is not an efficient solution of (MFP), then there exists \(x\in X^{0}\) such that

$$\begin{aligned} \frac{f_{i}(x)+ S(x|C_{i})}{g_{i}(x)- S(x|D_{i})}\leq\frac {f_{i}(u)+ S(u|C_{i})}{g_{i}(u)- S(u|D_{i})},\quad \mbox{for all } i=1,2,\ldots,k, \end{aligned}$$

and

$$\begin{aligned} \frac{f_{r}(x)+ S(x|C_{r})}{g_{r}(x)- S(x|D_{r})}< \frac{f_{r}(u)+ S(u|C_{r})}{g_{r}(u)- S(u|D_{r})}, \quad\mbox{for some } r = 1,2,\ldots,k, \end{aligned}$$

which implies

$$\begin{aligned} &\frac{f_{i}(x)+ x^{T}z_{i}}{g_{i}(x)- x^{T}v_{i}}\leq \frac {f_{i}(x)+ S(x|C_{i})}{g_{i}(x)- S(x|D_{i})}\leq\frac{f_{i}(u)+ S(u|C_{i})}{g_{i}(u)- S(u|D_{i})}= \frac{f_{i}(u)+ u^{T}z_{i}}{g_{i}(u)- u^{T}v_{i}}, \\ &\quad\mbox{for all } i=1,2,\ldots,k, \end{aligned}$$
(1)

and

$$\begin{aligned} &\frac{f_{r}(x)+ x^{T}z_{r}}{g_{r}(x)- x^{T}v_{r}}\leq \frac {f_{r}(x)+ S(x|C_{r})}{g_{r}(x)- S(x|D_{r})}< \frac{f_{r}(u)+ S(u|C_{r})}{g_{r}(u)- S(u|D_{r})}= \frac{f_{r}(u)+ u^{T}z_{r}}{g_{r}(u)- u^{T}v_{r}}, \\ & \quad\mbox{for some } r = 1,2,\ldots,k. \end{aligned}$$
(2)

Since \(\frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)} > 0\), \(i=1,2,\ldots,k\), the inequalities (1) and (2) give

$$\begin{aligned} \sum_{i=1}^{k} \frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)} \biggl(\frac{f_{i}(x)+ x^{T}z_{i}}{g_{i}(x)- x^{T}v_{i}}-\frac{f_{i}(u)+ u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) < 0. \end{aligned}$$
(3)

From Theorem 3.1, for each i, \(1\leq i \leq k\), \(( \frac{f_{i}(\cdot)+(\cdot)^{T}z_{i}}{g_{i}(\cdot)-(\cdot)^{T} v_{i}} )\) is higher-order \((\bar {C},\bar{\alpha}_{i},\bar{\gamma}_{i},\bar{\rho}_{i},\bar{d}_{i})\)-convex at \(u\in X^{0}\) with respect to \(\bar{H}_{i}\), p, and s, therefore

$$\begin{aligned} &\frac{1}{\bar{\alpha}_{i}(x,u)} \biggl(\frac {f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) \\ &\quad\geq\bar{C}_{x,u} \biggl(\nabla \biggl(\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\nabla_{p} \bar{H}_{i}(u,p) \biggr) \\ &\qquad{}+\frac {1}{\bar{\gamma}_{i}(x,u)}\bigl[\bar{H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr]+\frac {\bar{\rho}_{i} \bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)}, \end{aligned}$$
(4)

where

$$\begin{aligned}& \bar{\alpha}_{i}(x,u)= \biggl(\frac {g_{i}(u)-u^{T}v_{i}}{g_{i}(x)-x^{T}v_{i}} \biggr) \alpha_{i}(x,u),\qquad \bar{\gamma} _{i}(x,u)= \gamma_{i}(x,u),\\& \bar{\rho}_{i}=\rho_{i} \biggl(1+\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr), \\& \bar{H}_{i}(u,\cdot)= \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}}+\frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr)H_{i}(u,\cdot),\qquad \bar{d}_{i}(x,u)= \biggl( \frac{d_{i}(x,u)}{g_{i}(x)-x^{T}v_{i}} \biggr), \end{aligned}$$

and

$$\begin{aligned}& \bar{C}_{x,u}(a)= \biggl(\frac{f_{i}(u)+u^{T}z_{i} +g_{i}(u)-u^{T} v_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr) C_{x,u} \biggl( \frac{(g_{i}(u)-u^{T} v_{i})^{2} a}{f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i}} \biggr), \\& a= \nabla \biggl(\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\nabla_{p} \bar {H}_{i}(u,p). \end{aligned}$$

Also, by the higher-order \((C,\xi_{j},\delta_{j},\eta _{j},c_{j})\)-convexity of \((h_{j}(\cdot)+ (\cdot)^{T}w_{j})\) at u with respect to \(K_{j}\), q, and \(r, j=1,2,\ldots,m\), we have

$$\begin{aligned} &\frac{1}{\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq C_{x,u} \bigl(\nabla \bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &\qquad{} +\frac{1}{\delta _{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$
(5)

Let \(\tau= \sum_{i=1}^{k}\lambda_{i}\frac {(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}{(g_{i}(u)-u^{T}v_{i})^{2}}+ \sum_{j=1}^{m}\mu_{j} > 0\).

Adding the inequalities obtained by multiplying (4) by \(\frac{\lambda_{i}}{\tau}\) and (5) by \(\frac{\mu_{j}}{\tau}\), we get

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau} \frac {(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}{(g_{i}(u)-u^{T}v_{i})^{2}}C_{x,u} \Biggl[\biggl(\frac{(g_{i}(u)-u^{T}v_{i})^{2}}{f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i}} \biggr) \\ &\qquad{}\times\biggl( \nabla \biggl(\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+ \biggl(\frac {1}{g_{i}(u)-u^{T}v_{i}}+ \frac{f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr)\nabla_{p} H_{i}(u,p) \biggr) \\ &\qquad{} +\sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac {1}{\bar{\gamma}_{i}(x,u)} \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}}+ \frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr) \bigl(H_{i}(u,s)-s^{T} \nabla_{s}H_{i}(u,s)\bigr)\Biggr]\\ &\qquad{}+ \sum _{i=1}^{k}\frac{\lambda_{i}}{\tau}\frac{\rho_{i} d_{i}(x,u)}{\alpha _{i}(x,u)} \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}} +\frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr)\\ &\qquad{}+ \sum_{j=1}^{m} \frac{\mu _{j}}{\tau}C_{x,u} \bigl(\nabla\bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr)\\ &\qquad{}+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau}\frac{1}{\delta_{j}(x,u)} \bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau}\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

Further, using the convexity of C, we have

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq C_{x,u} \Biggl(\frac{1}{\tau} \Biggl(\sum _{i=1}^{k}{\lambda_{i}}\nabla \biggl( \frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{i=1}^{k}{ \lambda _{i}}\nabla_{p} \bar{H}_{i}(u,p)\\ &\qquad{}+\sum _{j=1}^{m}\mu_{j}\nabla \bigl(h_{j}(u)+u^{T}w_{j}\bigr)+\sum _{j=1}^{m} \mu_{j}\nabla_{q} K_{j}(u,q) \Biggr) \Biggr) \\ &\qquad{} +\sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac{1}{\bar{\gamma }_{i}(x,u)}\bigl(\bar{H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr)+ \sum _{i=1}^{k}\frac {\lambda_{i}}{\tau}\frac{\bar{\rho}_{i} \bar{d}_{i}(x,u)}{\bar{\alpha }_{i}(x,u)} \\ &\qquad{}+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau} \frac{1}{\delta _{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau }\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

It follows from hypotheses (iii)-(v) that

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau \sigma(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq C_{x,u} \Biggl(\frac{1}{\tau} \Biggl(\sum _{i=1}^{k}{\lambda_{i}}\nabla \biggl( \frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{j=1}^{m} \mu_{j}\nabla \bigl(h_{j}(u)+u^{T}w_{j} \bigr) \Biggr)\Biggr). \end{aligned}$$

Using the fact \(C_{x,u}(0)=0\) and \(\sum_{j=1}^{m}\mu _{j}(h_{j}(u)+u^{T}w_{j}) = 0\), we get

$$\begin{aligned} \sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) +\sum_{i=1}^{m} \frac {\mu_{j}}{\tau\sigma(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}\bigr] \geq0. \end{aligned}$$

Finally, using feasibility of the primal problem (MFP), we have

$$\begin{aligned} \sum_{i=1}^{k}\frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\geq0, \end{aligned}$$

which contradicts (3). Therefore, u is an efficient solution of (MFP). □

Duality model

Consider the following higher-order Mond-Weir type dual (MFD) of (MFP):

$$\begin{aligned} \begin{aligned} &\textstyle\begin{array}{@{}l@{\quad}l@{\quad}l} (\mathbf{MFD}) & \mbox{maximize}\quad \displaystyle G(u)= \biggl\{ \frac {f_{1}(u)+u^{T}z_{1}}{g_{1}(u)-u^{T}v_{1}},\frac {f_{2}(u)+u^{T}z_{2}}{g_{2}(u)-u^{T}v_{2}},\ldots,\frac {f_{k}(u)+u^{T}z_{k}}{g_{k}(u)-u^{T}v_{k}} \biggr\} \\ & \mbox{subject to}\\ &\quad \displaystyle\sum_{i=1}^{k}\lambda_{i}\nabla \biggl(\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{j=1}^{m} \mu _{j}\nabla\bigl(h_{j}(u)+u^{T}w_{j}\bigr)\\ &\displaystyle\qquad{}+\sum_{i=1}^{k}\lambda_{i}\nabla_{p} \bar{H}_{i}(u,p)+\sum_{j=1}^{m}\mu_{j}\nabla_{q} K_{j}(u,q) =0, \end{array}\displaystyle \end{aligned} \end{aligned}$$
(6)
$$\begin{aligned} &\sum_{j=1}^{m} \mu_{j}\bigl[h_{j}(u)+u^{T}w_{j}+ K_{j}(u,r)-r^{T} \nabla_{r}K_{j}(u,r) \bigr]\geq0, \end{aligned}$$
(7)
$$\begin{aligned} & \sum_{i=1}^{k} \lambda_{i}\bigl[\bar{H}_{i}(u,s)-s^{T} \nabla_{s} \bar {H}_{i}(u,s)\bigr]\geq0, \\ &z_{i}\in C_{i}, \qquad v_{i}\in D_{i},\qquad w_{j}\in E_{j}, \quad i=1,2, \ldots,k, j=1,2,\ldots,m, \\ &\lambda_{i}> 0, \qquad\mu_{j}\geq0, \quad i=1,2,\ldots,k, j=1,2,\ldots,m. \end{aligned}$$
(8)

We now discuss the duality results for the primal-dual pair (MFP) and (MFD).

Theorem 4.1

(Weak duality theorem)

Let \(x\in X^{0}\) and \((u,z,v,\mu,\lambda,w,p,q,r,s)\) be feasible for (MFD). Suppose that:

  1. (i)

    \((f_{i}(\cdot)+(\cdot)^{T}z_{i} )\) and \(- (g_{i}(\cdot)-(\cdot)^{T}v_{i} )\) are higher-order \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex at u, with respect to \(H_{i}\), p, and \(s, i=1,2,\ldots,k\),

  2. (ii)

    \((h_{j}(\cdot)+(\cdot)^{T}w_{j})\) is higher-order \((C,\xi_{j},\delta _{j},\eta_{j},c_{j})\)-convex at u, with respect to \(K_{j}\), q, and \(r, j=1,2,\ldots,m\),

  3. (iii)

    \(\sum_{i=1}^{k} \lambda_{i}\bar{\rho}_{i} \frac {\bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)} + \sum_{j=1}^{m} \mu _{j}\eta_{j}\frac{c_{j}(x,u)}{\xi_{j}(x,u)}\geq0\),

  4. (iv)

    \(\bar{\gamma}_{i}(x,u)=\zeta(x,u)\) and \(\xi _{j}(x,u)=\delta_{j}(x,u)=\sigma(x,u)\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\),

where

$$\begin{aligned}& \bar{\alpha}_{i}(x,u)= \biggl(\frac {g_{i}(x)-x^{T}v_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) \alpha_{i}(x,u), \qquad\bar{\rho}_{i}=\rho_{i} \biggl(1+\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr),\\& \bar {d}_{i}(x,u)= \biggl( \frac{d_{i}(x,u)}{g_{i}(x)-x^{T}v_{i}} \biggr). \end{aligned}$$

Then the following cannot hold:

$$\begin{aligned} {\frac{f_{i}(x)+S(x|C_{i})}{g_{i}(x)-S(x|D_{i})}\leq\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}}}, \quad\textit{for all } i=1,2,\ldots,k, \end{aligned}$$
(9)

and

$$\begin{aligned} {\frac{f_{r}(x)+S(x|C_{r})}{g_{r}(x)-S(x|D_{r})}< \frac {f_{r}(u)+u^{T}z_{r}}{g_{r}(u)-u^{T}v_{r}}}, \quad \textit{for some } r=1,2,\ldots, k. \end{aligned}$$
(10)

Proof

Suppose that (9) and (10) hold, then using \(\frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)}> 0\), \(x^{T}z_{i}\leq S(x|C_{i})\), \(x^{T}v_{i}\leq S(x|D_{i})\), \(i=1,2,\ldots,k\), we have

$$\begin{aligned} \sum_{i=1}^{k} \frac{\lambda_{i}}{\bar{\alpha }_{i}(x,u)} \biggl({\frac{f_{i}(x)+ x^{T} z_{i}}{g_{i}(x)- x^{T}v_{i}}- \frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}}} \biggr)< 0. \end{aligned}$$
(11)

From hypothesis (i) and Theorem 3.1, \((\frac {f_{i}(\cdot)+(\cdot)^{T}z_{i}}{g_{i}(\cdot)-(\cdot)^{T}v_{i}} )\) is higher-order \((\bar{C},\bar {\alpha}_{i},\bar{\gamma}_{i},\bar{\rho}_{i},\bar{d}_{i})\)-convex at u with respect to \(\bar{H}_{i}\), p, and \(s, i=1,2,\ldots,k\), therefore

$$\begin{aligned} &\frac{1}{\bar{\alpha}_{i}(x,u)} \biggl(\frac {f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) \\ &\quad\geq \bar{C}_{x,u} \biggl(\nabla \biggl(\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\nabla_{p} \bar{H}_{i}(u,p) \biggr) \\ &\qquad{}+\frac {1}{\bar{\gamma}_{i}(x,u)}\bigl[\bar{H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr]+\frac {\bar{\rho}_{i} \bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)}, \end{aligned}$$
(12)

and by the higher-order \((C,\xi_{j},\delta_{j},\eta_{j},c_{j})\)-convex of \((h_{j}(\cdot)+ (\cdot)^{T}w_{j})\) at u with respect to \(K_{j}\), q, and r, \(j=1,2,\ldots,m\), we have

$$\begin{aligned} &\frac{1}{\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq C_{x,u} \bigl(\nabla \bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &\qquad{}+\frac {1}{\delta_{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$
(13)

Let \(\tau= \sum_{i=1}^{k}\lambda_{i}\frac {(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}{(g_{i}(u)-u^{T}v_{i})^{2}}+ \sum_{j=1}^{m}\mu_{j} > 0\).

Multiply (12) by \(\frac{\lambda_{i}}{\tau}\) and (13) by \(\frac{\mu_{j}}{\tau}\) and add them, to get

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq \sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac {(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}{(g_{i}(u)-u^{T}v_{i})^{2}}C_{x,u} \Biggl[ \biggl( \frac{(g_{i}(u)-u^{T}v_{i})^{2}}{f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i}} \biggr) \\ &\qquad{}\times\biggl(\nabla \biggl(\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+ \biggl(\frac {1}{g_{i}(u)-u^{T}v_{i}} +\frac{f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr)\nabla_{p} H_{i}(u,p) \biggr)\\ &\qquad{}+\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau} \frac {1}{\bar{\gamma}_{i}(x,u)} \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}}+\frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr) \bigl(H_{i}(u,s)-s^{T}\nabla_{s}H_{i}(u,s) \bigr)\Biggr] \\ &\qquad{}+ \sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac{\rho_{i} d_{i}(x,u)}{\alpha_{i}(x,u)} \biggl(\frac{1}{g_{i}(u)-u^{T}v_{i}}+ \frac {f_{i}(u)+u^{T}z_{i}}{(g_{i}(u)-u^{T}v_{i})^{2}} \biggr)\\ &\qquad{}+ \sum_{j=1}^{m} \frac{\mu _{j}}{\tau}C_{x,u} \bigl(\nabla\bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &\qquad{} +\sum_{j=1}^{m} \frac{\mu_{j}}{\tau}\frac{1}{\delta _{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau }\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

Further, using convexity on C, (6)-(8), and hypotheses (iii)-(iv), we have

$$\begin{aligned} \sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{i=1}^{m} \frac{\mu_{j}}{\tau \sigma(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}\bigr] \geq0. \end{aligned}$$

Since x is a feasible solution of (MFP), it follows that

$$\begin{aligned} \sum_{i=1}^{k}\frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\geq0. \end{aligned}$$

This contradicts (11). Hence we have the result. □

Theorem 4.2

(Strong duality theorem)

Assume ū is an efficient solution of (MFP) and let the Slater constraint qualification be satisfied on X. Also, if

$$\begin{aligned}& \bar{H}(\bar{u},0)=0,\qquad K(\bar{u},0)=0, \qquad\nabla_{p} \bar{H}( \bar {u},0)=0, \qquad\nabla_{q} K(\bar{u},0)=0,\\& \nabla_{s}\bar{H}(\bar{u},0)=0,\qquad \nabla _{r}K( \bar{u},0)=0, \end{aligned}$$

then there exist \(0 < \bar{\lambda}\in R^{k}\), \(\bar{\mu}\in R^{m}\), \(\bar {z}_{i}\in R^{n}\), \(\bar{v}_{i}\), and \(\bar{w}_{j}\in R^{n}\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\), such that \((\bar{u},\bar{z},\bar {v},\bar{\mu},\bar{\lambda},\bar{w},\bar{p}=0, \bar{q}=0, \bar {s}=0, \bar{r}=0)\) is a feasible solution of (MFD) and the objective function values of (MFP) and (MFD) are equal. Furthermore, if the conditions of Theorem  4.1 hold for all feasible solutions of (MFP) and each feasible solution \((u',z',v',\mu',\lambda',w',p'=0, q'=0, r'=0, s'=0)\) of (MFD), then \((\bar{u},\bar{z},\bar{v},\bar{\mu},\bar{\lambda},\bar {w},\bar{p}=0, \bar{q}=0, \bar{r}=0, \bar{s}=0)\) is an efficient solution of (MFD).

Proof

Assume ū is an efficient solution of (MFP) and the Slater constraint qualification is satisfied on X. Then, from Theorem 3.4, there exist \(\bar{\lambda}_{i}> 0\), \(\bar{\mu}\in R^{m}\), \(\bar {z}_{i}\in R^{n}\), \(\bar{v}_{i}\in R^{n}\), and \(\bar{w}_{j}\in R^{n}\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\), such that

$$\begin{aligned}& \sum_{i=1}^{k}\bar{\lambda}_{i} \nabla \biggl(\frac{f_{i}(\bar {u})+\bar{u}^{T}\bar{z}_{i}}{g_{i}(\bar{u})-\bar{u}^{T}\bar{v}_{i}} \biggr)+ \sum_{j=1}^{m} \bar{\mu}_{j}\nabla\bigl(h_{j}(\bar{u})+\bar {u}^{T}\bar{w}_{j}\bigr)=0, \end{aligned}$$
(14)
$$\begin{aligned}& \sum_{j=1}^{m}\bar{\mu}_{j} \nabla\bigl(h_{j}(\bar{u})+\bar {u}^{T}\bar{w}_{j} \bigr)=0, \end{aligned}$$
(15)
$$\begin{aligned}& \bar{u}^{T}\bar{z}_{i}=S(\bar{u}|C_{i}),\quad i=1,2,\ldots,k, \end{aligned}$$
(16)
$$\begin{aligned}& \bar{u}^{T}\bar{v}_{i}=S(\bar{u}|D_{i}), \quad i=1,2,\ldots,k, \end{aligned}$$
(17)
$$\begin{aligned}& \bar{u}^{T}\bar{w}_{j}=S(\bar{u}|E_{j}),\quad j=1,2,\ldots,m, \end{aligned}$$
(18)
$$\begin{aligned}& \bar{z}_{i}\in C_{i}, \qquad\bar{v}_{i}\in D_{i}, \qquad\bar{w}_{j}\in E_{j},\quad i=1,2, \ldots,k, j=1,2,\ldots,m, \end{aligned}$$
(19)
$$\begin{aligned}& \bar{\lambda}_{i}> 0,\qquad \bar{\mu}_{j}\geq0, \quad i=1,2,\ldots,k, j=1,2,\ldots,m. \end{aligned}$$
(20)

Thus, \((\bar{u},\bar{z},\bar{v},\bar{\mu},\bar{\lambda},\bar{w},\bar {p}=0,\bar{q}=0, \bar{r}=0, \bar{s}=0)\) is feasible for (MFD) and from (16)-(18), the objective function values of (MFP) and (MFD) are equal.

We now show that \((\bar{u},\bar{z},\bar{v},\bar{\mu},\bar{\lambda},\bar {w},\bar{p}=0, \bar{q}=0, \bar{r}=0, \bar{s}=0)\) is an efficient solution of (MFD). If not, then there exists \((u',z',v',\mu',\lambda', w',p', q', r', s')\) of (MFD) such that

$$\begin{aligned} {\frac{f_{i}(\bar{u})+\bar{u}^{T}\bar{z}_{i}}{g_{i}(\bar {u})-\bar{u}^{T}\bar{v}_{i}}\leq\frac{f_{i}(u')+ u^{\prime T}z'_{i}}{g_{i}(u')- u^{\prime T}v'_{i}}},\quad i=1,2,\ldots,k, \end{aligned}$$

and

$$\begin{aligned} {\frac{f_{r}(\bar{u})+\bar{u}^{T}\bar{z}_{r}}{g_{r}(\bar {u})-\bar{u}^{T}\bar{v}_{r}}< \frac{f_{r}(u')+ u^{\prime T}z'_{r}}{g_{r}(u')- u^{\prime T}v'_{r}}}, \quad\mbox{for some } r=1,2, \ldots,k. \end{aligned}$$

This contradicts the weak duality theorem. Hence we have the result. □

Theorem 4.3

(Strict converse duality theorem)

Let x be a feasible solution for (MFP) and \((u,z,v,\mu,\lambda ,w,p,q,r,s)\) be feasible for (MFD). Suppose that:

  1. (i)

    \((f_{i}(\cdot)+(\cdot)^{T}z_{i} )\) and \(- (g_{i}(\cdot)-(\cdot)^{T}v_{i} )\) are higher-order strictly \((C,\alpha_{i},\gamma_{i},\rho_{i},d_{i})\)-convex at u, with respect to \(H_{i}\), p, and \(s, i=1,2,\ldots,k\),

  2. (ii)

    \((h_{j}(\cdot)+(\cdot)^{T}w_{j})\) is higher-order \((C,\xi_{j},\delta _{j},\eta_{j},c_{j})\)-convex at u, with respect to \(K_{j}\), q, and \(r, j=1,2,\ldots,m\),

  3. (iii)

    \(\bar{\gamma}_{i}(x,u)=\zeta(x,u)\) and \(\xi _{j}(x,u)=\delta_{j}(x,u)=\sigma(x,u)\), \(i=1,2,\ldots,k\), \(j=1,2,\ldots,m\),

  4. (iv)

    \(\sum_{i=1}^{k} \lambda_{i}\bar{\rho}_{i} \frac {\bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)} + \sum_{j=1}^{m} \mu _{j}\eta_{j}\frac{c_{j}(x,u)}{\xi_{j}(x,u)}\geq0\),

  5. (v)

    \(\sum_{i=1}^{k}\frac{\lambda_{i}}{\bar{\alpha }_{i}(x,u)} (\frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} )\leq0\),

where

$$\begin{aligned}& \bar{\alpha}_{i}(x,u)= \biggl(\frac {g_{i}(x)-x^{T}v_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) \alpha_{i}(x,u),\qquad \bar{\rho}_{i}=\rho _{i} \biggl(1+\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr), \\& \bar {d}_{i}(x,u)= \biggl( \frac{d_{i}(x,u)}{g_{i}(x)-x^{T}v_{i}} \biggr). \end{aligned}$$

Then \(x =u\).

Proof

Suppose that \(x\neq u\) and exhibit a contradiction. Let \((u,z,v,\mu,\lambda,w,p,q,r,s)\) be feasible for (MFD), then

$$\begin{aligned} C_{x,u}(0) =&C_{x,u} \Biggl( \sum _{i=1}^{k}\lambda_{i}\nabla \biggl( \frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{j=1}^{m} \mu_{j}\nabla\bigl(h_{j}(u)+u^{T}w_{j} \bigr) \\ &{}+\sum_{i=1}^{k}\lambda_{i} \nabla_{p} \bar{H}_{i}(u,p)+\sum _{j=1}^{m}\mu_{j}\nabla_{q} K_{j}(u,q) \Biggr)=0. \end{aligned}$$
(21)

Since \((\frac{f_{i}(\cdot)+(\cdot)^{T}z_{i}}{g_{i}(\cdot)-(\cdot)^{T}v_{i}} )\) is higher-order strictly \((\bar{C},\bar{\alpha}_{i},\bar{\gamma}_{i},\bar {\rho}_{i},\bar{d}_{i})\)-convex at u with respect to \(\bar{H}_{i}\), p, and s, we have

$$\begin{aligned} &\frac{1}{\bar{\alpha}_{i}(x,u)} \biggl(\frac {f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\quad> \bar{C}_{x,u} \biggl(\nabla \biggl(\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+ \nabla_{p} \bar{H}_{i}(u,p) \biggr) \\ &\qquad{}+\frac {1}{\bar{\gamma}_{i}(x,u)}\bigl[\bar{H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr]+\frac {\bar{\rho}_{i} \bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)}. \end{aligned}$$

Let \(\tau= \sum_{i=1}^{k}\lambda_{i}\frac {(f_{i}(u)+u^{T}z_{i}+g_{i}(u)-u^{T}v_{i})}{(g_{i}(u)-u^{T}v_{i})^{2}}+ \sum_{j=1}^{m}\mu_{j} > 0\).

Multiplying by \(\frac{\lambda_{i}}{\tau}\) in the above inequality, we get

$$\begin{aligned} &\frac{\lambda_{i}}{\tau} \biggl[\frac{1}{\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr) \\ &\quad>\bar{C}_{x,u} \biggl( \nabla \biggl(\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\nabla_{p} \bar{H}_{i}(u,p) \biggr) \\ &\qquad{} +\frac {1}{\bar{\gamma}_{i}(x,u)}\bigl[\bar{H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr]+\frac {\bar{\rho}_{i} \bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)} \biggr]. \end{aligned}$$
(22)

By the higher-order \((C,\xi_{j},\delta_{j},\eta_{j},c_{j})\)-convex of \((h_{j}(\cdot)+ (\cdot)^{T}w_{j})\) at u with respect to \(K_{j}\), q, and r, \(j=1,2,\ldots,m\), we obtain

$$\begin{aligned} \frac{1}{\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr]\geq{}& C_{x,u} \bigl(\nabla \bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &{}+\frac {1}{\delta_{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

It follows that

$$\begin{aligned} &\frac{\mu_{j}}{\tau} \biggl[\frac{1}{\xi _{j}(x,u)} \bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad\geq C_{x,u} \bigl(\nabla \bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &\qquad{}+\frac {1}{\delta_{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)} \biggr]. \end{aligned}$$
(23)

Taking summation over i in (22) and over j in (23), we get

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau\xi _{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad> \sum_{i=1}^{k}\frac{\lambda_{i}}{\tau} \bar{C}_{x,u} \biggl(\nabla \biggl(\frac{f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+ \nabla_{p} \bar {H}_{i}(u,p) \biggr)\\ &\qquad{}+\sum _{i=1}^{k}\frac{\lambda_{i}}{\tau}\frac{1}{\bar {\gamma}_{i}(x,u)}\bigl[ \bar{H}_{i}(u,s)-s^{T}\nabla_{s} \bar{H}_{i}(u,s)\bigr] \\ &\qquad{} + \sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac{\bar{\rho}_{i} \bar {d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)}+ \sum_{j=1}^{m} \frac{\mu_{j}}{\tau }C_{x,u} \bigl(\nabla\bigl(h_{j}(u)+u^{T}w_{j} \bigr)+\nabla_{q} K_{j}(u,q) \bigr) \\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau}\frac{1}{\delta _{j}(x,u)}\bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau }\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

Further, using convexity on C and (21), we obtain

$$\begin{aligned} &\sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)\\ &\qquad{}+\sum_{j=1}^{m} \frac{\mu_{j}}{\tau \xi_{j}(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}-h_{j}(u)-u^{T}w_{j} \bigr] \\ &\quad > \sum_{i=1}^{k}\frac{\lambda_{i}}{\tau} \frac{1}{\bar{\gamma}_{i}(x,u)}\bigl[\bar {H}_{i}(u,s)-s^{T} \nabla_{s}\bar{H}_{i}(u,s)\bigr]+\sum _{j=1}^{m}\frac{\mu_{j}}{\tau }\frac{1}{\delta_{j}(x,u)} \bigl[K_{j}(u,r)-r^{T} \nabla_{r} K_{j}(u,r)\bigr]\\ &\qquad{}+ \sum_{i=1}^{k} \frac{\lambda_{i}}{\tau}\frac{\bar{\rho }_{i} \bar{d}_{i}(x,u)}{\bar{\alpha}_{i}(x,u)} + \sum_{j=1}^{m} \frac{\mu _{j}}{\tau}\frac{\eta_{j} c_{j}(x,u)}{\xi_{j}(x,u)}. \end{aligned}$$

This, together with (7)-(8) and hypotheses (iii)-(iv), shows

$$\begin{aligned} \sum_{i=1}^{k}\frac{\lambda_{i}}{\tau\bar{\alpha }_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)+\sum_{i=1}^{m} \frac{\mu_{j}}{\tau \sigma(x,u)}\bigl[h_{j}(x)+x^{T}w_{j}\bigr]> 0. \end{aligned}$$

Since x is a feasible solution of (MFP), it follows that

$$\begin{aligned} \sum_{i=1}^{k} \frac{\lambda_{i}}{\bar{\alpha}_{i}(x,u)} \biggl( \frac{f_{i}(x)+x^{T}z_{i}}{g_{i}(x)-x^{T}v_{i}}-\frac {f_{i}(u)+u^{T}z_{i}}{g_{i}(u)-u^{T}v_{i}} \biggr)> 0. \end{aligned}$$

This contradicts the hypothesis (v). Hence the proof is completed. □

Conclusions

In this article, we consider a class of fractional programming problem having k-objectives in which each numerator and denominator of the objective function is nondifferentiable in terms of the support function of a compact convex set. The important property that the ratio of higher-order \((C,\alpha,\gamma,\rho,d)\)-convex function is also a higher-order \((\bar{C},\bar{\alpha},\bar{\gamma},\bar{\rho},\bar {d})\)-convex function is obtained. We also derive sufficient optimality conditions for an efficient solution for this problem. Furthermore, a higher-order Mond-Weir type dual is formulated and appropriate duality relations are obtained under higher-order \((C,\alpha,\gamma,\rho ,d)\)-convexity assumptions.

References

  1. Mangasarian, OL: Second and higher-order duality in nonlinear programming. J. Math. Anal. Appl. 51, 607-620 (1975)

    MathSciNet  Article  Google Scholar 

  2. Chen, X: Higher-order symmetric duality in nondifferentiable multiobjective programming problems. J. Math. Anal. Appl. 290, 423-435 (2004)

    MATH  MathSciNet  Article  Google Scholar 

  3. Kassem, MA: Higher-order symmetric duality in vector optimization problem involving generalized cone-invex functions. Appl. Math. Comput. 209, 405-409 (2009)

    MATH  MathSciNet  Article  Google Scholar 

  4. Suneja, SK, Srivastava, MK, Bhatia, M: Higher order duality in multiobjective fractional programming with support functions. J. Math. Anal. Appl. 347, 8-17 (2008)

    MATH  MathSciNet  Article  Google Scholar 

  5. Ying, G: Higher-order symmetric duality for a class of multiobjective fractional programming problems. J. Inequal. Appl. 2012, 142 (2012)

    Article  Google Scholar 

  6. Yuan, DH, Chinchuluun, A, Pardalos, PM: Nondifferentiable minimax fractional programming problems with \((C,\alpha,\rho,d)\)-convexity. J. Optim. Theory Appl. 185, 185-199 (2006)

    MathSciNet  Article  Google Scholar 

  7. Chinchuluun, A, Yuan, D, Pardalos, PM: Optimality conditions and duality for nondifferentiable multiobjective fractional programming with generalized convexity. Ann. Oper. Res. 154, 133-147 (2007)

    MATH  MathSciNet  Article  Google Scholar 

  8. Long, XJ: Optimality conditions and duality for nondifferentiable multiobjective programming problems with \((C,\alpha,\rho,d)\)-convexity. J. Optim. Theory Appl. 148, 197-208 (2011)

    MATH  MathSciNet  Article  Google Scholar 

  9. Hanson, MA: On sufficiency of the Kuhn-Tucker conditions. J. Math. Anal. Appl. 80, 545-550 (1981)

    MATH  MathSciNet  Article  Google Scholar 

  10. Gupta, SK, Danger, D, Ahmad, I: On second order duality for nondifferentiable minimax fractional programming problems involving type-I functions. ANZIAM J. 55, 479-494 (2014)

    Google Scholar 

  11. Pandey, S: Duality for multiobjective fractional programming involving generalized η-convex functions. Opsearch 28, 31-43 (1991)

    Google Scholar 

  12. Gulati, TR, Saini, H: Higher-order \((F,\alpha,\beta,\rho,d)\)-convexity and its application in fractional programming. Eur. J. Pure Appl. Math. 4, 266-275 (2011)

    MathSciNet  Google Scholar 

  13. Liang, ZA, Huang, HX, Pardalos, PM: Optimality conditions and duality for a class of nonlinear fractional programming problem. J. Optim. Theory Appl. 110, 611-619 (2001)

    MATH  MathSciNet  Article  Google Scholar 

  14. Gupta, SK, Kailey, N, Sharma, MK: Multiobjective second-order nondifferentiable symmetric duality involving \((F,\alpha,\rho,d)\)-convex function. J. Appl. Math. Inform. 28, 1395-1408 (2010)

    MATH  MathSciNet  Google Scholar 

  15. Chen, X: Sufficiency conditions and duality for a class of multiobjective fractional programming problems with higher-order \((F,\alpha,\rho,d)\)-convexity. J. Appl. Math. Comput. 28, 107-121 (2008)

    MATH  MathSciNet  Article  Google Scholar 

  16. Egudo, RR: Multiobjective fractional duality. Bull. Aust. Math. Soc. 37, 367-378 (1988)

    MATH  MathSciNet  Article  Google Scholar 

  17. Geoffrion, AM: Proper efficiency and the theory of vector maximization. J. Math. Anal. Appl. 22, 618-630 (1968)

    MATH  MathSciNet  Article  Google Scholar 

  18. Gulati, TR, Geeta: Duality in nondifferentiable multiobjective fractional programming problem with generalized invexity. J. Appl. Math. Comput. 35, 103-118 (2011)

    MATH  MathSciNet  Article  Google Scholar 

Download references

Acknowledgements

The authors are thankful to the reviewers for their valuable and constructive suggestions. The authors are also grateful to Deanship of Scientific Research (DSR), University of Tabuk, Saudi Arabia under Grant No. S-0131-1436, for their support to carry out this work.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Shiv K Gupta.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors contributed equally to the writing of this paper. All authors read and approved the final manuscript.

Rights and permissions

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.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Dubey, R., Gupta, S.K. & Khan, M.A. Optimality and duality results for a nondifferentiable multiobjective fractional programming problem. J Inequal Appl 2015, 354 (2015). https://doi.org/10.1186/s13660-015-0876-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13660-015-0876-0

MSC

  • 90C26
  • 90C30
  • 90C32
  • 90C46

Keywords

  • duality results
  • multiobjective programming problem
  • support function
  • KKT conditions
  • efficient solution