Abstract
We use -type-I and generalized -type-I functions to establish sufficient optimality conditions and duality results for multiobjective variational problems. Some of the related problems are also discussed.
1. Introduction
The relationship between mathematical programming and classical calculus of variation was explored and extended by Hanson [1]. Thereafter variational programming problems have some attention in the literature. Mond and Hanson [2] obtained optimality conditions and duality results for scalar valued variational problems under convexity assumptions. Motivated by the approach by Bector and Husain [3], Nahak and Nanda [4] and later Bhatia and Mehra [5] extended the results of Mond et al. [6] to multiobjective variational problems involving invex functions and generalized -invex functions, respectively. Analogous results were developed by Zalmai [7] for fractional variational programming problem containing arbitrary norms and by Liu [8] for generalized fractional case involving -convex functions.
Type-I functions were first introduced by Hanson and Mond [9], and Rueda and Hanson [10] defined a class of pseudo-type-I and quasi-type-I functions as generalization of type-I functions. Bhatia and Mehra [5] studied the optimality conditions and duality results for multiobjective variational problems involving generalized -invexity. We use --type-I and generalized --type-I functions to establish sufficient optimality conditions and duality results for multiobjective variational problems.
In Section 2 of this paper, we introduce --type-I and generalized --type-I functions for the continuous case. Using these new classes of functions, we establish various sufficient optimality conditions in Section 3 of this paper. Section 4 is devoted to the duality results. Some related problems are discussed in Section 5. The results obtained in this paper are more general than those obtained in [4, 5, 11].
Let be a real interval, let , and let be continuously differentiable functions with respect to their arguments and , denotes the derivative of with respect . Denote the partial derivative of scalar valued function , with respect to , , and , respectively, by , , and such that , . Similarly we write the partial derivative of the vector functions and using matrices with and rows, respectively, instead of one. Let denote the space of piecewise smooth functions with norm , where the differential operator is given by Therefore, except at discontinuities.
Consider the following multiobjective variational problem: Let .
Efficiency and proper efficiency are defined in their usual sense as defined in [4].
In relation to (), we introduce the following multiobjective problem , for each , each problem with single objective:
The following lemma can be established on the lines of Chankong and Haimes [12].
Definition 1 (see [4]). A point in is said to be an efficient solution of () if for all
Definition 2 (see [13]). A point in is said to be a properly efficient solution of () if for all there exists a scalar such that for all for some such that whenever is in and
Lemma 3. is an efficient solution of if and only if is an optimal solution of for each .
Theorem 4 (see [2]). For every optimal normal solution of , for each , there exist real numbers with and piecewise smooth function such that
2. --Type-I and Generalized --Type-I Functions
Definition 5. Let and be real valued functions. A pair is said to be -type-I at with respect to , , , and if there exist functions , and , with , such that for all If, in the previous definition, (9) is satisfied as a strict inequality, then we say that a pair is -semistrictly-type-I at with respect to , , , and and .
Definition 6. A pair is said to be -quasi-type-I at with respect to , , , if there exist functions , and , with , such that for all
Definition 7. A pair is said to be -strongly pseudo-type-I at with respect to , , , and if there exist functions , and , with , such that for all , Clearly the class of -quasi-type-I functions and the class of -strongly pseudo-type-I functions are more general than the class of -type-I functions.
Definition 8. A pair is said to be -quasi--pseudo-type-I at with respect to , , , and if there exist functions , and , with , such that for all If, in the above definition, inequality (14) is satisfied as then we say that a pair is -quasi--strictly pseudo-type-I at with respect to , , , and and .
Definition 9. A pair is said to be -strongly pseudo--quasi-type-I at with respect to , , , if there exist functions , and , with , such that for all If, in the above definition, inequality (16) is satisfied as then we say that a pair is -strictly pseudo--quasi-type-I at with respect to , , , and and .
Remark 10. Let be a continuously differentiable function with respect to each of the arguments. Let be differentiable with and . Then,
Lemma 11. Every -type-I function is -strongly pseudo-type-I function but the converse is not true.
Proof. It is clear that the statement every -type-I function is -strongly pseudo-type-I function but the converse is not true follows from the following counter example. Example 12. Let be defined by
Let be defined by
Let the functions and be defined by
Let the functions and be defined by
Taking and , we have to show that the pair is -strongly pseudo-type-I but not -type-I with respect to , , , and . First we verify that is -strongly pseudo-type-I function. Case 1. When , it follows that
Again
Case 2. When , it is clear that
From Cases 1 and 2, it is clear that the pair is -strongly pseudo-type-I.
But the pair is not -type-I for because
3. Optimality Conditions
Theorem 13 (necessary optimality conditions). Let be a properly efficient solution for () which is assumed to be normal solution for , for . Then, there exist and a piecewise smooth function such that
Proof. See [5].
Theorem 14. Let be a feasible solution of (VP), and let there exist , , and a piecewise smooth function such that for all Further, if is -type-I at with respect to functions , , , , , and , with , then is a properly efficient solution of ().
Proof. As is -type-I at , for all , we have Since , we write (33) as Adding (32) and (34), we have Using Remark 10, (35) becomes By (29) and , we have As , it follows that that is, which implies that minimizes over with . Hence is a properly efficient solution of () on account of Theorem 1 of Bector and Husain [3].
Theorem 15. Let be a feasible solution of , and let there exist , , and a piecewise smooth function such that, for every , conditions, (29)–(31) of Theorem 14 are satisfied. Further, if is -semistrictly-type-I at with respect to functions , , , , , and , with for all , then is a properly efficient solution of .
Proof. Let be not; an efficient solution of (), then there exist and an index such that the following inequalities hold. Because and , (40) imply Further in view of (30) As is -semistrictly-type-I at with respect to functions , , , and , we have Adding (43), we have By Remark 10, we have which contradicts (29) and . Hence is a properly efficient solution of ().
Theorem 16. Let be a feasible solution of (VP), and let there exist , , and a piecewise smooth function such that conditions (29) of Theorem 14 are satisfied by . Further, if is -strongly pseudo--quasi-type-I at with respect to functions , , , , and , , then is a properly efficient solution of (VP).
Proof. From (30), we have
As is -strongly pseudo--quasi-type-I at with respect to functions , , , and ,
Using Remark 10, (45) becomes
Equation (48) along with (29) gives
Using Remark 10, for all , (48) becomes
As , we have
Equation (51) along with the fact that is -strongly pseudo--quasi-type-I at gives
Since , it follows that
Hence minimizes over with .
Therefore is a properly efficient solution of () ([3, Theorem 1]).
Theorem 17. Let be a feasible solution of , and let there exist , , and a piecewise smooth function such that the conditions (29)–(31) of Theorem 14 are satisfied by . Further, if is -quasi--strictly pseudo-type-I at with respect to functions , , , , , and , with for all , then is a properly efficient solution of .
Proof. If is not an efficient solution of (), then there exist and an index , , such that Because and , the previous relations give which along with the fact that is -quasi--strictly pseudo-type-I at with respect to functions , , , and gives Using Remark 10, (56) becomes that is, Now using (29) in (58), we have It follows that As , we have Again using Remark 10, we obtain which in view of a given hypothesis implies that Since , we obtain which contradicts (30). Hence is a properly efficient solution of ().
4. Duality
The Mond-Weir-type dual problem associated with () is given by We now establish duality results between () and () under generalized -type-I conditions.
Theorem 18 (weak duality). Let be a feasible solution of , and let be a feasible solution of (VD ). Let either the following conditions hold. (i) is -semi-strictly-type-I at with respect to functions , , , , and .(ii) and is -strongly pseudo--quasi-type-I at with respect to functions , , , and with , and .(iii) is -quasi--strictly pseudo-type-I at with respect to functions , , , and , with , , and , Then
Proof. (i) Let, if possible,
Then there exists an index , , such that
Because and , the above relations give
that is,
which is the same as (41) with replaced by and replaced by . The rest of the proof runs on the same line of that of Theorem 15 and hence is omitted.
(ii) From (67), we get
Since , we have
which implies that
on account of hypothesis (ii). This is the same as (45) with replaced by and replaced by . Again proceeding on the lines of Theorem 16, we get the result.
(iii) Proof of this part follows on the lines of Theorem 17 and hence is omitted.
Theorem 19 (strong duality). Let be a properly efficient solution of . Assume that is normal for each , . Then there exist and a piecewise smooth function such that is feasible for (). Further, if, for each feasible of (), any of the conditions of Theorem 18 hold, then is a properly efficient solution of ().
Proof. Because is a properly efficient solution of (), it follows that, from Theorem 13 there exist and a piecewise smooth function such that (28) hold. Moreover, , and hence the feasibility of for () follows. As the weak duality (Theorem 18) holds between () and (), is an efficient solution of (). If is not a properly efficient solution of (), then, proceeding on the lines to that of [3, Theorem 1], we get a contradiction to the weak duality.
Theorem 20 (strict converse duality). Let be feasible for (), and let be feasible for (VD ) such that Further, let be -strictly pseudo--quasi-type-I at with respect to functions , , , , , and , , for all . Then is a properly efficient solution of ().
Proof. We assume that and get a contradiction. Feasibility of for () implies that As , we have Equation (80) along with the fact that is -strictly pseudo--quasi-type-I at implies that Using Remark 10 (with replaced by ) in (81), we get Equation (82) along with (66) gives Since , we have Again using Remark 10 (with replaced by ) in (84), we obtain which in view of the given hypothesis implies that As , it follows that But this contradicts (78), hence the result.
5. Some Related Problems
All the optimality conditions and duality results developed for () in the previous sections can easily be modified for several other classes of variational problems.
5.1. Natural Boundary Value Problems
Omitting the boundary conditions for the fixed end points as was done by Mond and Hanson [2], we obtain The corresponding Mond-Weir type dual is given by If the problems and are independent of , then these problems essentially reduce to the static case and of multiobjective nonlinear programs:
Acknowledgments
The authors wish to thank the referees and the Editor for their valuable suggestions which improved the presentation of the paper.