- Research
- Open access
- Published:
Duality results for interval-valued semiinfinite optimization problems with equilibrium constraints using convexificators
Journal of Inequalities and Applications volume 2022, Article number: 128 (2022)
Abstract
This paper deals with the study of interval-valued semiinfinite optimization problems with equilibrium constraints (ISOPEC) using convexificators. First, we formulate Wolfe-type dual problem for (ISOPEC) and establish duality results between the (ISOPEC) and the corresponding Wolfe-type dual under the assumption of \(\partial ^{*} \)-convexity. Second, we formulate Mond–Weir-type dual problem and propose duality results between the (ISOPEC) and the corresponding Mond–Weir-type dual under the assumption of \(\partial ^{*} \)-convexity, \(\partial ^{*} \)-pseudoconvexity, and \(\partial ^{*} \)-quasiconvexity.
1 Introduction
Due to a wide range of applications, the optimization problem with equilibrium constraints has been studied widely by many researchers. This problem regularly emerges in real-world applications, such as engineering designs, robust optimization, image restoration issues, fuzzy sets, and molecular distance geometry problems; see, for example, [1–8]. Optimization with equilibrium is a constrained optimization problem, with certain variational inequality or complementary constraints. The mathematical problem with equilibrium constraints (MPEC) belongs to a class of extremely significant problems. For a survey of recent developments and applications of optimization problems with equilibrium constraints, see [6, 9–15].
In nonsmooth optimization, the concept of convexificators has been shown to be a useful tool for deriving optimality and duality results. In 1994, Demyanov [16] developed the notion of a convexificator. A closed nonconvex convexificator for extended real-valued functions was developed by Jeyakumar and Luc [17]. In terms of convexificators, Wang and Jeyakumar [18] proposed the optimality conditions for optimization problems with equality constraints.
Throughout the last several decades, the study of dual problems connected to MPEC has been utilized to solve optimization issues in various fields, such as fractional problems, complex minimax problems, semiinfinite problems, and so on. Wolfe [19] and Mond–Weir [20] developed dual models for nonlinear programming problems. Rockafellar [21] used a conjugate function to study a fundamental duality theory for convex programming and developed an extended version of Fenchel’s duality theorem. Sun et al. [22] investigated Wolfe-type robust duality between the uncertain convex optimization problem and its uncertain dual problem by proving duality between the deterministic robust counterpart of the primal model and the optimistic counterpart of its dual problem. Sun et al. [23] introduced a mixed-type robust dual problem of this class of uncertain optimization problems and explored robust strong duality relations between them. Further, Sun et al. [24] established necessary and sufficient optimality conditions for robust approximate optimal solutions of this uncertain convex optimization problem, introduced a Wolfe-type robust approximate dual problem, and investigated robust approximate duality relations between them. The duality theory of nonlinear programming problems has received a lot of attention, see Mangasarian [3] and Mishra and Giorgi [25].
Semiinfinite programming is an optimization problem with an infinite number of inequality constraints on a feasible set. Semiinfinite optimization has recently become an active field in applied mathematics. Mishra et al. [26] established duality results for nonsmooth semiinfinite programming problems, and these results were extended for vanishing and equilibrium constraints by Pandey and Mishra [6, 7].
With the advancement of mathematics, an increasing number of researchers are focusing on interval-valued optimization problems. For continuously differentiable interval-valued optimization problems, Wu [27] considered the Wolfe-type dual problem. Tung [28] gave the optimality conditions and duality results for convex semiinfinite multiobjective interval-valued optimization problems. Using the concept of contingent epiderivatives, Su and Dinh [29] investigated the duality for interval-valued pseudoconvex optimization problems with equilibrium constraints. Recently, Wang and Wang [30] established duality theorems for nondifferentiable semiinfinite interval-valued optimization problems with vanishing constraints.
In this work, we consider the duality results for the interval-valued semiinfinite optimization problem with equilibrium constraints ((ISOPEC)) and show the dual relationships between (ISOPEC) and the corresponding Wolfe- and Mond–Weir-type dual models, using convexificators, which are inspired by the literature listed above. The organization of this paper is as follows. In Sect. 1, we recall some needful preliminaries and basic results. In Sect. 2, we establish weak and strong duality results between the (ISOPEC) and the corresponding Wolfe-type dual, using convexificators under the assumption of \(\partial ^{*}\)-convexity. In Sect. 3, we establish weak and strong duality results between the (ISOPEC) and the corresponding Mond–Weir-type dual, using convexificators under the assumption of \(\partial ^{*}\)-convexity, and in the last Sect. 4, we establish weak and strong duality results between the (ISOPEC) and the corresponding Mond–Weir-type dual, using convexificators under the assumption of \(\partial ^{*}\)-pseudoconvexity and \(\partial ^{*}\)-quasiconvexity.
2 Preliminaries
We recall some basic definitions and results in this section, which will be used in this paper.
We denote a finite-dimensional Euclidean space by X. The notation \(\langle \cdot ,\cdot \rangle \) refers to the inner product. The symbol \(B(\bar{z},\delta ):= \{{z\in X: \| z-\bar{z}\|} < \delta \}\) denotes the open ball of radius δ centered at z̄. Let C be a nonempty subset of X. We write \(\mathrm{co}\ C\), \(\mathrm{cone}\ C\), and \(\mathrm{cl}\ C\) to denote the convex hull, convex cone, and closure of C, respectively, generated by C. Let \(C\neq \emptyset \), then the contingent cone of set C at point z̄ is defined by
Let \(\mathbb{D}\) be the set of all closed and bounded intervals in \(\mathbb{R}\). For any \(C= [c_{1}, c_{2}] \in \mathbb{D}\), \(D = [d_{1}, d_{2}] \in \mathbb{D}\), one can see [31] that
where k denotes a real number. A partial ordering for an interval can be expressed as follows:
Let \(F: X \rightarrow \mathbb{D}\) be a function on X, defined as
where \(F^{L}\) and \(F^{U}\) are functions defined on X with \(F^{L}(z)\leq F^{U}(z)\), \(\forall z\in X\). We consider the interval-valued semiinfinite optimization program with equilibrium constraints (ISOPEC)
where the index set T is an arbitrary nonempty set, not necessarily finite; functions \(g: X \times T \rightarrow \mathbb{R}\), \(h_{j}: X \rightarrow \mathbb{R}\), \(G_{i}: X \rightarrow \mathbb{R}\mathbbm{,} \ \text{and} \ H_{i}: X \rightarrow \mathbb{R}\) are locally Lipschitz on X and \((\cdot )^{T}\) indicates the transpose. The feasible set of problem (ISOPEC) is
Definition 2.1
Suppose \(\bar{z} \in M\).
-
1.
A feasible point z̄ is said to be a locally LU-optimal solution of (ISOPEC) if there exists a neighborhood \(N(\bar{z})\) of z̄ such that there is no \(z \in M \cap N(\bar{z})\), for which
$$ F(z)< _{\mathrm{LU}} F(\bar{z}).$$ -
2.
A feasible point z̄ is said to be a locally weakly LU-optimal solution of (ISOPEC) if there exists a neighborhood \(N(\bar{z})\) of z̄ such that there is no \(z \in M \cap N(\bar{z})\), for which
$$ F(z)< _{\mathrm{LU}}^{s} F(\bar{z}).$$
Let \(\bar{z} \in M \) be a feasible vector for (ISOPEC). We define the following index sets to be used in the sequel:
The set β is called a degenerate set. The vector z̄ is said to satisfy the strict complementarity condition if β is empty.
Let \(f: X \rightarrow \mathbb{R}\cup \{+\infty \}\) be an extended real-valued function and \(z\in X \) be such that \(f(z)\) is finite. Then the lower and upper Dini directional derivatives of f at z in the direction d, respectively, are given as
Definition 2.2
([17])
A function \(f: X \rightarrow \mathbb{R}\cup \{+\infty \}\) is said to admit an upper convexificator \(\partial ^{*}f(z)\) at \(z \in \ X\) if \(\partial ^{*}f(z) \subset X \) is a closed set and, for every \(d \in X\),
The function f is said to admit a lower convexificator \(\partial ^{*}f(z)\) at \(z \in \ X\) if \(\partial ^{*}f(z) \subset X \) is a closed set and, for every \(d \in X\),
A closed set \(\partial ^{*}f(z)\) is said to be a convexificator of f at z if it is both an upper and lower convexificator.
Now, we collect the definition of upper and lower semiregular convexificator from [17, 32].
Definition 2.3
Let \(f : X \rightarrow \mathbb{R} \cup \{+\infty \}\) be an extended real-valued function and let \(z \in X \) be such that \(f(z)\) is finite. The function f is said to admit an upper semiregular convexificator \(\partial ^{*}f (z) \subset X \) at z if \(\partial ^{*} f(z) \) is closed and, for each \(d\in X \),
The function f is said to admit a lower semiregular convexificator \(\partial _{*} f(z) \subset X \) at z if \(\partial _{*} f(z) \) is closed and, for each \(d \in X\),
The set \(\partial ^{*}f(z) \) is said to be a semiregular convexificator, if it is both an upper and lower semiregular convexificator.
Definition 2.4
([33])
Let \(f: X \rightarrow \mathbb{R}\cup \{+\infty \}\) be an extended real-valued function and suppose it admits an upper semiregular convexificator at \(z \in X\). Then the function f is said to be \(\partial ^{*}\)-convex at x̄ iff
Example 2.1
Consider the function \(f:\mathbb{R}^{2}\rightarrow \mathbb{R}\) given by
The set \(\partial ^{*} f(0,0)=\{(-1,1),(-1,-1),(1,1), (1,-1)\} \) is an upper semiregular convexificator of \(f(z)\) at \(\bar{z}=(0,0)\). Then,
Thus, we conclude that f is \(\partial ^{*}\)-convex.
Definition 2.5
([34])
Let \(f: X \rightarrow \mathbb{R}\cup \{+\infty \}\) be an extended real-valued function and suppose it admits an upper semiregular convexificator at \(z \in X\). Then the function f is said to be \(\partial ^{*}\)-pseudoconvex at x̄ iff
Example 2.2
([17])
Consider the function \(f:\mathbb{R}\rightarrow \mathbb{R}\) given by
The set \(\partial ^{*} f(0)=[1, +\infty )\) is an upper convexificator of f at 0 and
Also f is \(\partial ^{*}\)-pseudoconvex on \(\mathbb{R}\), but not \(\partial ^{*}\)-convex.
We extend the following sets of indicators and definitions for interval-valued semiinfinite mathematical programming problem with equilibrium constraints (ISOPEC), which will be used in the sequel, that were given for semiinfinite mathematical programming problem with equilibrium constraints (SIMPEC) by Pandey and Mishra [7]:
where \(t_{1},t_{2},\dots ,t_{k}\in T_{g}(\bar{z})\leq n+1\).
Definition 2.6
If all functions have upper convexificators at a feasible point z̄ of (ISOPEC), we say that the generalized standard Abadie constraints condition (GS-ACQ) is satisfied at z̄ if at least one of the dual sets used in the definition of \(\Gamma (\bar{z})\) is nonempty and
Definition 2.7
(Generalized alternatively stationary point)
Let z̄ be a feasible point of (SIOPEC). A generalized alternatively stationary point, denoted by GA-stationary, is said to be at z̄ if there exist \(\alpha ^{L},\alpha ^{U} \in \mathbb{R}\), \(\eta = (\eta ^{g}, \eta ^{h}, \eta ^{G}, \bar{\eta}^{H})\in \mathbb{R}^{k+p+2l}\), \(\mu = (\mu ^{h}, \mu ^{G}, \mu ^{H})\in \mathbb{R}^{p+2l}\), and indices \(t_{1},t_{2},\dots ,t_{k} \in T_{g}(\bar{z})\), \(k\leq n+1 \) such that the following conditions hold:
Definition 2.8
(Generalized strong stationary point)
Let z̄ be a feasible point of (ISOPEC). A generalized alternatively stationary point, denoted by GA-stationary, is said to be at z̄ if there exist \(\alpha ^{L},\alpha ^{U} \in \mathbb{R}\) and \(\eta = (\eta ^{g}, \eta ^{h}, \eta ^{G}, \eta ^{H})\in \mathbb{R}^{k+p+2l}\), \(\ \mu = (\mu ^{h}, \mu ^{G}, \mu ^{H})\in \mathbb{R}^{p+2l}\), and indices \(t_{1},t_{2},\dots ,t_{k} \in T_{g}(\bar{z})\), \(k\leq n+1\), satisfying conditions (3)–(5) together with the following conditions:
3 Duality
In this section we formulate Wolfe- and Mond–Weir-type dual model for (ISOPEC) and establish the weak and strong duality results in terms of convexificators. Let \(\bar{z} \in M\) be any feasible solution of (ISOPEC). The following index sets will be used in the sequel:
Definition 3.1
(Wolfe dual)
The Wolfe dual (WDISOPEC) for the interval-valued semiinfinite optimization problem with equilibrium constraints (ISOPEC) is defined by
where \(\eta = (\eta ^{g}, \eta ^{h}, \eta ^{G}, \eta ^{H}) \in \mathbb{R}^{k+p+2l}\), \(\mu \in (\mu ^{h},\mu ^{G},\mu ^{H}) \in \mathbb{R}^{p+2l}\) and \(t_{1},t_{2},\dots ,t_{k} \in T_{g}(\bar{z})\), \(k \leq n+1\).
Theorem 3.2
(Weak duality)
Let z̄ and \((w,\alpha ^{L},\alpha ^{U},\eta )\) be feasible points for the interval-valued semiinfinite optimization problem (ISOPEC) and the dual problem (WDISOPEC), respectively, and let the index set \(T_{g}\), as well as α, β, γ, be defined accordingly. Suppose the functions \(F^{L};\, F^{U}; \, g(\cdot ,t)\), for \(t\in T\); \(\pm h_{j}\), for \(j=1,\dots ,p;\, -G_{i}\), for \(i\in \nu _{1}\), and \(-H_{i}\), for \(i\in \nu _{2}\), admit bounded upper semiregular convexificators and \(\partial ^{*}\)-convex functions at w. If \(L_{\mu} = \phi \), then
Proof
Based on the initial hypothesis, we have the following system:
Since \(F^{L}\) and \(F^{U} \) are convex functions,
Similarly, we have
If \(L_{\mu} = \phi \), multiplying (13)–(19) by \(\alpha ^{L},\alpha ^{U},\eta _{i}^{g} \geq 0\), \(t_{i}\in T_{g}\), \(\eta _{j}^{h}> 0\), \(j=1,\dots ,p\), \(\mu _{j}^{h}> 0\), \(j=1,\dots ,p\), \(\eta _{i}^{G}> 0\), \(i\in \nu _{1}\), \(\eta _{i}^{H}> 0\), \(i\in \nu _{2}\), respectively, and adding (13)–(19), we have
where \(\xi _{\alpha}=\alpha ^{L}\xi _{1} + \alpha ^{U}\xi _{2} \).
From (7), there exist \(\bar{\xi _{1}} \in \mathrm{co}\,\partial ^{*}F^{L}(w)\), \(\ \bar{\xi _{2}} \in \mathrm{co}\,\partial ^{*}F^{U}(w)\), \(\bar{\xi}_{i}^{g} \in \mathrm{co}\,\partial ^{*}g(w,t_{i})\), \(\bar{\zeta}_{j}^{h_{1}} \in \mathrm{co}\,\partial ^{*}h_{j}(w)\), \(\bar{\zeta}_{j}^{h_{2}}\in \mathrm{co}\,\partial ^{*}(-h_{j})(w)\), \(\bar{\xi}_{i}^{G} \in \mathrm{co}\,\partial ^{*}(-G_{i})(w)\), and \(\bar{\xi}_{i}^{H} \in \mathrm{co}\,\partial ^{*}(-H_{i})(w)\), such that
Therefore, it follows from (20) that
Now, combining (9)–(12), we get
This, together with equation (21), leads to
In fact, if (8) does not hold, then there would be
By definition, we have
or
or
Since \(\alpha ^{L},\alpha ^{U}\in (0,1)\) and \(\alpha ^{L}+\alpha ^{U} =1\), the strict inequality holds:
which contradicts (22) and hence the proof is complete. □
Example 3.1
Consider the following \((\mathit{WDISOPEC})\) problem in \(\mathbb{R}^{2}\):
The feasible set, denoted by M, is defined by
Clearly, \(z^{*}= (0,0)\) is a feasible point. Then, \(T_{g}(0,0)=\{0\}\), \(\beta (0,0)=\{1\}\).
Now consider
subject to
where
From all the above, we conclude that the conditions on the multipliers can be imposed as follows: \(\eta _{1}^{g}\geq 0\), \(\ \mu _{1}^{H}=0\), \(\eta _{1}^{G}= 0\), \(\alpha ^{L} w_{1}=-\eta _{1}^{G}=0\), and \(\alpha ^{U} w_{2}=\eta _{1}^{g}\).
Then, we get the conclusions
Hence, Theorem 3.2 is verified.
Now, we present an example to show that the obtained results may fail to hold when the convexificator is not available.
Example 3.2
Consider the following \((\mathit{WDISOPEC})\) problem in \(\mathbb{R}^{2}\):
The feasible region of the problem is
The Wolfe-type dual problem \((\mathit{WDISOPEC}(\bar{z}))\) for the primal problem is
and
If we use a set which is not a convexificator then we show that the weak duality theorem does not hold. For this, let \(\partial ^{*}F^{L}(w)=(1,2)\), \(\partial ^{*} F^{U}(w) =(2,3)\). Then,
Now, we conclude that
and w from the Wolfe dual feasible region.
As a direct consequence of Theorem 3.2, the following corollary follows.
Corollary 3.3
Let z̄ be feasible for (ISOPEC), where all constraint functions \(g(\cdot ,t),\, t\in T;\ \pm h_{j}\), \(j=1,\dots ,p;\, -G_{i}\), \(i \in \nu _{1};\,-H_{i}\), \(i\in \nu _{2}\) are affine and the index set \(T_{g}\), as well as α, β, γ, is defined accordingly. Then, for any z feasible for (ISOPEC) and \((w,\alpha ^{L},\alpha ^{U}, \eta )\) feasible for (WDISOPEC), we have
Now, to obtain the strong duality result for the (WDISOPEC), we give the following corollary.
Corollary 3.4
Let z̄ be a local solution of (ISOPEC). Suppose that \(F^{L}\), \(F^{U}\) are locally Lipschitz around z̄ also that \(F^{L}\), \(F^{U}\) and effective constraint functions admitting bounded upper semiregular convexificators at z̄. If GS-ACQ is satisfied at z̄, then z̄ is a GS-stationary point.
Proof
If \(F^{L}\), \(F^{U}\), and effective constraint functions admit bounded upper semiregular convexificators, then, with the help of [9, Sect. 4.5] and [7, Theorem 3.1], z̄ is a GS-stationary point by Definition 2.8. □
Theorem 3.5
(Strong Duality)
Let z̄ be a local LU-optimal solution of (ISOPEC) and suppose \(F^{L}\) and \(F^{U}\) are locally Lipschitz near z̄. Suppose that \(F^{L}\), \(F^{U}\); \(g(\cdot ,t)\), \(t\in T\); \(\pm h_{j}\), \(j=1,\dots ,p\); \(-G_{i}\), \(i\in \nu _{1}\); \(-H_{i}\), \(i\in \nu _{2}\) admit bounded upper semiregular convexificators and are \(\partial ^{*}\)-convex functions at z̄. If GS-ACQ is satisfied at z̄, then there exist \((\bar{\alpha}^{L}, \bar{\alpha}^{U}) \in \mathbb{R}\) and \(\bar{\eta} = (\bar{\eta}^{g}, \bar{\eta}^{h}, \bar{\eta}^{G}, \bar{\eta}^{H})\in \mathbb{R}^{k+p+2l}\) such that \((\bar{z},\bar{\alpha}^{L}, \bar{\alpha}^{U},\bar{\eta})\) is an LU-optimal solution of (WDISOPEC) and respective objective values are equal.
Proof
Since z̄ is a locally LU-optimal solution of (ISOPEC) and the GS-ACQ condition holds at z̄, by Corollary 3.4, \(\exists \ (\bar{\alpha}^{L}, \bar{\alpha}^{U})\in \mathbb{R}\), \(\bar{\eta} = (\bar{\eta}^{g}, \bar{\eta}^{h}, \bar{\eta}^{G}, \bar{\eta}^{H})\in \mathbb{R}^{k+p+2l}\), \(\bar{\mu} =(\bar{\mu}^{h}, \bar{\mu}^{G}, \bar{\mu}^{H})\in \mathbb{R}^{p+2l}\), and indices \(t_{1},t_{2},\dots ,t_{k} \in T_{g}(\bar{z})\), \(k\leq n+1 \) such that the GS-stationarity conditions for (ISOPEC) hold, that is, there exist \(\bar{\xi}_{1} \in \mathrm{co}\,\partial ^{*}F^{L}(w) + \bar{\xi}_{2} \in \mathrm{co}\,\partial ^{*}F^{U}(w)\), \(\bar{\xi}_{i}^{g} \in \mathrm{co}\,\partial ^{*}g(w,t_{i})\), \(\bar{\zeta}_{j}^{h_{1}} \in \mathrm{co}\,\partial ^{*}h_{j}(w)\), \(\bar{\zeta}_{j}^{h_{2}}\in \mathrm{co}\,\partial ^{*}(-h_{j})(w)\), \(\bar{\xi _{i}}^{G} \in \mathrm{co}\,\partial ^{*}(-G_{i})(w)\), and \(\bar{\xi}_{i}^{H} \in \mathrm{co}\,\partial ^{*}(-H_{i})(w)\) such that, for each \(\bar{\xi}_{\alpha}=\bar{\alpha}^{L}\bar{\xi}_{1} + \bar{\alpha}^{U} \bar{\xi}_{2}\), one obtains
By the definition of (WDISOPEC), we suppose that \(( \bar{z},\bar{\alpha}^{L}, \bar{\alpha}^{U},\bar{\eta})\) is a feasible vector of that problem. By virtue of Theorem 3.2, we have
for any feasible solution \((z,\alpha ^{L},\alpha ^{U},\eta )\) for the Wolfe dual (WDISOPEC). Since \(t_{i} \in T_{g}(\bar{z})\), \(g(\bar{z}, t_{i}) = 0\), \(h_{j}(\bar{z})=0\), \(\ G_{i}(\bar{z})=0\), \(\forall i \in \nu _{1}\), \(H_{i}(\bar{z})=0\), \(\forall i \in \nu _{2} \), we have
Now, from (29) and (30), we have
Thus, the vector \((\bar{z},\bar{\alpha}^{L}, \bar{\alpha}^{U}),\bar{\eta})\) is an optimal solution for (WDISOPEC) and the respective objective values are equal, which completes the proof. □
By using convexificators, we establish duality results that relate the Mond–Weir-type dual problem (MWISOPEC) to the optimization problem with equilibrium constraints (ISOPEC).
Definition 3.6
(Mond–Weir dual)
The Mond–Weir dual (MWDISOPEC) for the interval -valued semiinfinite optimization problem with equilibrium constraints (ISOPEC) is defined by
where \(\eta = (\eta ^{g}, \eta ^{h}, \eta ^{G}, \eta ^{H}) \in \mathbb{R}^{k+p+2l}\), \(\ \mu \in (\mu ^{h},\mu ^{G},\mu ^{H}) \in \mathbb{R}^{p+2l}\), and \(t_{1},t_{2},\dots ,t_{k} \in T_{g}(\bar{z})\), \(k \leq n+1\).
Theorem 3.7
(Weak duality)
Let z̄ and \((w,\alpha ^{L},\alpha ^{U},\eta )\) be the feasible points for the interval-valued semiinfinite optimization problem (ISOPEC) and the dual problem (MWDISOPEC), respectively, and let the index set \(T_{g}\), as well as α, β, γ, be defined accordingly. Suppose the functions \(F^{L}\), \(F^{U}\); \(g(\cdot ,t)\), \(t\in T\); \(\pm h_{j}\), \(j=1,\dots ,p\); \(-G_{i}\), \(i\in \nu _{1}\); \(-H_{i}\), \(i\in \nu _{2}\) admit bounded upper semiregular convexificators and are \(\partial ^{*}\)-convex functions at w. If \(L_{\mu} = \phi \), then
Proof
The proof is similar to that of Theorem 3.2. □
As a direct consequence of Theorem (3.7), the following corollary follows.
Corollary 3.8
Suppose z̄ is feasible for (ISOPEC) where all constraint functions \(g(\cdot ,t),\,t\in T;\,\pm h_{j}\), \(j=1,\dots ,p;\, -G_{i}\), \(i\in \nu _{1};\,-H_{i}\), \(i\in \nu _{2}\) are affine and the index set \(T_{g}\), as well as α, β, γ, is defined accordingly. Then, for any z feasible for (ISOPEC) and w, α, η feasible for (WMDISOPEC), we have
Theorem 3.9
(Strong duality)
Suppose z̄ is a local LU-optimal solution of (ISOPEC) and \(F^{L}\) and \(F^{U}\) are locally Lipschitz at z̄. Suppose that \(F^{L}\), \(F^{U}\); \(g(\cdot ,t)\), \(t\in T\); \(\pm h_{j}\), \(j=1,\dots ,p\); \(-G_{i}\), \(i\in \nu _{1}\); \(-H_{i}\), \(i\in \nu _{2}\) admit bounded upper semiregular convexificators and are \(\partial ^{*}\)-convex functions at z̄. If GS-ACQ holds at z̄, then there exist \((\bar{\alpha}^{L}, \bar{\alpha}^{U}) \in \mathbb{R}\) and \(\bar{\eta} = (\bar{\eta}^{g}, \bar{\eta}^{h}, \bar{\eta}^{G}, \bar{\eta}^{H})\in \mathbb{R}^{k+p+2l}\) such that \((\bar{z},\bar{\alpha}^{L}, \bar{\alpha}^{U},\bar{\eta})\) is an LU-optimal solution of (WMDISOPEC) and respective objective values are equal.
Proof
The proof is similar to that of Theorem 3.5. □
Now, we derive weak and strong results for the (ISOPEC) and its Mond–Weir-type dual problem under generalized \(\partial ^{*}\)-convexity assumptions.
Theorem 3.10
(Weak duality)
Let z̄ and \((w,\alpha ^{L}, \alpha ^{U},\eta )\) be feasible points for the interval-valued semiinfinite optimization problem (ISOPEC) and the dual problem (MWDISOPEC), respectively, and the index set \(T_{g}\), as well as α, β, γ, be defined accordingly. Suppose the functions \(F^{L}\) and \(F^{U} \) are \(\partial ^{*}\)-pseudoconvex at w, \(g(\cdot ,t)\), \(t\in T\); \(\pm h_{j}\), \(j=1,\dots ,p\); \(-G_{i}\), \(i\in \nu _{1}\); \(-H_{i}\), \(i\in \nu _{2}\) admit bounded upper semiregular convexificators and are \(\partial ^{*}\)-quasiconvex functions at w. If \(L_{\mu} = \phi \), then
Proof
Suppose that \(F(z) <^{s}_{\mathrm{LU}} F(w) \), then we have
For each \(t_{i}\in T_{g}\),
similarly,
Combining (36)–(40) and using \(\partial ^{*}\)-quasiconvexity of the above functions, we have
that is,
Using the above inequality and (7), there exist \(\xi _{1}\in \partial ^{*}F^{L}(w)\) and \(\xi _{2}\in \partial ^{*}F^{U}(w)\) such that
By (35) and the \(\partial ^{*}\)-pseudoconvexity of \(F^{L}\) and \(F^{U}\), it follows that
Then \(\langle \alpha ^{L}\xi _{1}+\alpha ^{U}\xi _{2},z-w \rangle <0\) and \(\alpha ^{L}, \alpha ^{U} \in [0,1]\), \(\alpha ^{L}+\alpha ^{U}=1\), which contradicts (47). This completes the proof. □
Theorem 3.11
(Strong duality)
Suppose z̄ is a local LU-optimal solution of (ISOPEC) and \(F^{L}\) and \(F^{U}\) are locally Lipschitz at z̄. Suppose that \(F^{L}\), \(F^{U}\) are \(\partial ^{*}\)-pseudoconvex at z̄, \(g(\cdot ,t)\), \(t\in T\); \(\pm h_{j}\), \(j=1,\dots ,p\); \(-G_{i}\), \(i\in \nu _{1}\); \(-H_{i}\), \(i\in \nu _{2}\) admit bounded upper semiregular convexificators and are \(\partial ^{*}\)-quasiconvex functions at z̄. If GS-ACQ holds at z̄, then there exist \((\bar{\alpha}^{L}, \bar{\alpha}^{U}) \in \mathbb{R}\) and \(\bar{\eta} = (\bar{\eta}^{g}, \bar{\eta}^{h}, \bar{\eta}^{G}, \bar{\eta}^{H})\in \mathbb{R}^{k+p+2l}\) such that \((\bar{z},\bar{\alpha}^{L}, \bar{\alpha}^{U},\bar{\eta})\) is an LU-optimal solution of (WMDISOPEC) and respective objective values are equal.
Proof
The proof follows that of Theorem 3.9 and uses Theorem 3.10. □
4 Conclusion
In this paper, we have introduced duality results of interval-valued semiinfinite optimization problems with equilibrium constraints (ISOPEC). We have formulated the Wolfe-type dual (WDISOPEC) and Mond–Weir-type dual (WMDISOPEC) by using a convexificators. We have established weak and strong duality results relating to (ISOPEC) and the corresponding two duals (WDISOPEC) and (WMDISOPEC), under the assumption of \(\partial ^{*} \)-convexity, \(\partial ^{*} \)-pseudoconvexity, and \(\partial ^{*} \)-quasiconvexity. We can extend results of our paper for interval-valued semiinfinite optimization problems under assumptions of generalized convexity motivated by Sun et al. [35, 36].
Availability of data and materials
Not applicable
References
Bot, R.I., Grad, S.M., Wanka, G.: Duality in Vector Optimization. Springer, Berlin (2009)
Bonnel, H., Iusem, A.N., Svaiter, B.F.: Proximal methods in vector optimization. SIAM J. Optim. 15(4), 953–970 (2005)
Mangasarian, O.L.: Nonlinear Programming. McGraw-Hill, New York (1969)
Mangasarian, O.L.: Pseudo-convex functions. J. SIAM Control Ser. A 3, 281–290 (1965)
Movahedian, N., Nobakhtian, S.: Necessary and sufficient conditions for nonsmooth mathematical programs with equilibrium constraints. Nonlinear Anal., Theory Methods Appl. 72(5), 2694–2705 (2010)
Pandey, Y., Mishra, S.K.: Duality for nonsmooth optimization problems with equilibrium constraints, using convexificators. J. Optim. Theory Appl. 171(2), 694–707 (2016)
Pandey, Y., Mishra, S.K.: Optimality conditions and duality for semi-infinite mathematical programming problems with equilibrium constraints, using convexificators. Ann. Oper. Res. 269(1), 549–564 (2018)
Lai, K.K., Hassan, M., Singh, S.K., Maurya, J.K., Mishra, S.K.: Semidefinite multiobjective mathematical programming problems with vanishing constraints using convexificators. Fractal Fract. 6(1), 3 (2022)
Ansari Ardali, A., Movahedian, N., Nobakhtian, S.: Optimality conditions for nonsmooth mathematical programs with equilibrium constraints, using convexificators. Optimization 65(1), 67–85 (2016)
Mishra, S.K., Jaiswal, M.: Optimality conditions and duality for semi-infinite mathematical programming problem with equilibrium constraints. Numer. Funct. Anal. Optim. 36(4), 460–480 (2015)
Hassan, M., Maurya, J.K., Mishra, S.K.: On M-Stationary Conditions and Duality for Multiobjective Mathematical Programs with Vanishing Constraints. Bull. Malays. Math. Sci. Soc. 1–27 (2022)
Kim, Y., Leyffer, S., Munson, T.: MPEC methods for bilevel optimization problems. In: Bilevel Optimization, pp. 335–360. Springer, Cham (2020)
Pandey, Y., Mishra, S.K.: Duality of mathematical programming problems with equilibrium constraints. Pac. J. Optim. 13(1), 105–122 (2017)
Singh, Y., Pandey, Y., Mishra, S.K.: Saddle point optimality criteria for mathematical programming problems with equilibrium constraints. Oper. Res. Lett. 45(3), 254–258 (2017)
Guu, S.M., Mishra, S.K., Pandey, Y.: Duality for nonsmooth mathematical programming problems with equilibrium constraints. J. Inequal. Appl. 2016(1), 1 (2016)
Demyanov, V.F.: Convexification and concavification of positively homogeneous functions by the same family of linear functions. Report 3(208), 802 (1994)
Jeyakumar, V., Luc, D.T.: Nonsmooth calculus, minimality, and monotonicity of convexificators. J. Optim. Theory Appl. 101, 599–621 (1999)
Wang, X., Jeyakumar, V.: A sharp Lagrange multiplier rule for nonsmooth mathematical programming problems involving equality constraints. SIAM J. Optim. 10(4), 1136–1148 (2000)
Wolfe, P.: A duality theorem for non-linear programming. Q. Appl. Math. 19(3), 239–244 (1961)
Mond, B., Weir, T.: Generalized Concavity and Duality, Generalized Concavity in Optimization and Economics pp. 263–279. Academic Press, New York (1981)
Rockafellar, R.T.: Duality theorems for convex functions. Bull. Am. Math. Soc. 70(1), 189–192 (1964)
Sun, X.K., Peng, Z.Y., Guo, X.L.: Some characterizations of robust optimal solutions for uncertain convex optimization problems. Optim. Lett. 10(7), 1463–1478 (2016)
Sun, X., Teo, K.L., Tang, L.: Dual approaches to characterize robust optimal solution sets for a class of uncertain optimization problems. J. Optim. Theory Appl. 182(3), 984–1000 (2019)
Sun, X.K., Teo, K.L., Zeng, J., Guo, X.L.: On approximate solutions and saddle point theorems for robust convex optimization. Optim. Lett. 14(7), 1711–1730 (2020)
Mishra, S.K., Giorgi, G.: Invexity and Optimization. Springer, New York (2008)
Mishra, S.K., Jaiswal, M., An, L.T.H.: Duality for nonsmooth semi-infinite programming problems. Optim. Lett. 6(2), 261–271 (2012)
Wu, H.C.: Wolfe duality for interval-valued optimization. J. Optim. Theory Appl. 138(3), 497–509 (2008)
Tung, L.T.: Karush–Kuhn–Tucker optimality conditions and duality for convex semi-infinite programming with multiple interval-valued objective functions. J. Appl. Math. Comput. 62(1), 67–91 (2020)
Van Su, T., Dinh, D.H.: Duality results for interval-valued pseudoconvex optimization problem with equilibrium constraints with applications. Comput. Appl. Math. 39(2), 1–24 (2020)
Wang, H., Wang, H.: Duality theorems for nondifferentiable semi-infinite interval-valued optimization problems with vanishing constraints. J. Inequal. Appl. 2021(1), 1 (2021)
Moore, R.E.: Methods and Applications of Interval Analysis. SIAM, Philadelphia (1979)
Demyanov, V.F., Jeyakumar, V.: Hunting for a smaller convex subdifferential. J. Glob. Optim. 10, 305–326 (1997)
Suneja, S.K., Kohli, B.: Optimality and duality results for bilevel programming problem using convexifactors. J. Optim. Theory Appl. 150, 1–19 (2011)
Li, X.F., Zhang, J.Z.: Necessary optimality conditions in terms of convexificators in Lipschitz optimization. J. Optim. Theory Appl. 131, 429–452 (2006)
Sun, X., Teo, K.L., Long, X.J.: Some characterizations of approximate solutions for robust semi-infinite optimization problems. J. Optim. Theory Appl. 191(1), 281–310 (2021)
Sun, X., Teo, K.L., Zeng, J., Liu, L.: Robust approximate optimal solutions for nonlinear semi-infinite programming with uncertainty. Optimization 69(9), 2109–2129 (2020)
Funding
The second author was financially supported by “Research Grant for Faculty” (IoE Scheme) under Dev. Scheme NO. 6031. The third author was financially supported by CSIR-UGC JRF, New Delhi, India, through Reference no. 1009/(CSIR-UGC NET JUNE 2018).
Author information
Authors and Affiliations
Contributions
Writing-original draft preparation, K.K.L.,S.K.M. M.H., J.B., and J.K.M. ; writing-review and editing, K.K.L.,S.K.M. M.H., J.B., and J.K.M.
Corresponding author
Ethics declarations
Competing interests
The authors declare 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
Lai, K.K., Mishra, S.K., Hassan, M. et al. Duality results for interval-valued semiinfinite optimization problems with equilibrium constraints using convexificators. J Inequal Appl 2022, 128 (2022). https://doi.org/10.1186/s13660-022-02866-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13660-022-02866-1