Mehar Methods for Fuzzy Optimal Solution and Sensitivity Analysis of Fuzzy Linear Programming with Symmetric Trapezoidal Fuzzy Numbers

The drawbacks of the existing methods to obtain the fuzzy optimal solution of such linear programming problems, in which coefficients of the constraints are represented by real numbers and all the other parameters as well as variables are represented by symmetric trapezoidal fuzzy numbers, are pointed out, and to resolve these drawbacks, a new method (named as Mehar method) is proposed for the same linear programming problems. Also, with the help of proposed Mehar method, a new method, much easy as compared to the existing methods, is proposed to deal with the sensitivity analysis of the same type of linear programming problems.


Introduction
Linear programming is one of the most successively applied operation research techniques.Real world situations are represented by using any linear programming model which involves a lot of parameters, whose values are assigned by experts.However, both experts and decision maker frequently do not precisely know the value of those parameters.Therefore, it is useful to consider the knowledge of experts about the parameters as fuzzy data [1].In this section, the work which has been done in recent years for such linear programming problems, in which the imprecise parameters are represented by symmetric triangular/trapezoidal fuzzy numbers, is reviewed.
Ganesan and Veeramani [2] proposed new product of symmetric trapezoidal fuzzy numbers and proposed a method to find the fuzzy optimal solution of such singleobjective linear programming problems in which coefficients of the constraints are represented by real numbers and all the other parameters as well as variables are represented by symmetric trapezoidal fuzzy numbers.
Nasseri and Mahdavi-Amiri [3] proposed the dual for the existing symmetric fuzzy linear programming problems [2] and proposed some duality results.Nasseri et al. [4] proved weak and strong duality theorems for the existing symmetric fuzzy linear programming problems [2].
Ebrahimnejad et al. [5] extended the existing symmetric fuzzy linear programming problem [2] to bounded symmetric fuzzy linear programming problem and proposed a method for solving these problems.
Ebrahimnejad [6] pointed out that existing method [2] can be used only if it is easily possible to find a basic feasible solution of primal problem.However, if this condition is not satisfied then existing method [2] cannot be used.To overcome these limitations, Ebrahimnejad [6] proposed a primal dual simplex algorithm for solving these problems.
Ebrahimnejad and Nasseri [7] proposed a duality approach to solve linear programmes with trapezoidal fuzzy numbers.Kumar and Kaur [8] proposed an alternative method for solving existing symmetric fuzzy linear programming problems [2].
Ebrahimnejad [9] proved that if fuzzy linear programming problem considered by Ganesan and Veeramani [2] had 2 Mathematical Problems in Engineering a fuzzy feasible solution, then it also has fuzzy basic feasible solution, and if it has an optimal fuzzy solution, then it has an optimal fuzzy basic solution.
Ebrahimnejad and Verdegay [10] proposed a new method to deal with the sensitivity analysis of that kind of fuzzy linear programming problems in which the elements of the coefficient matrix of the constraints are represented by real numbers and remaining parameters are represented by symmetric trapezoidal fuzzy numbers.
Kheirfam and Verdegay [11] proposed an approach, based on dual simplex method, for existing symmetric fuzzy linear programming problems [2] and also proposed a method to deal with the sensitivity analysis of these problems.

Drawbacks of Existing Methods
Ganesan and Veeramani [2] proposed a product of symmetric trapezoidal fuzzy numbers and proposed a method to find fuzzy optimal solution of symmetric fuzzy linear programming problem (P 1 ).Since then the different methods [4][5][6][7][9][10][11] have been proposed for the same linear programming problems.
In the crisp environment, an alternative optimal solution of a crisp linear programming problem will not exist if the relative profit corresponding to none of the basic variables is zero in the optimal table [12].On the same direction, in all the existing methods [2,[4][5][6][7][9][10][11], it is assumed that if the rank of relative fuzzy profit corresponding to none of the fuzzy basic variables is zero in the optimal table, then alternative fuzzy optimal solution will not exist.
for example, Kheirfam and Verdegay [11] solved the problem (P 2 ) to illustrate their proposed method and concluded that Table 1 is the optimal table of this problem.Since, in Table 1, the rank of the fuzzy relative profit corresponding to none of the fuzzy basic variables is zero, according to the existing methods [2,[4][5][6][7][9][10][11] there will not exist any alternative fuzzy optimal solution for this fuzzy linear programming problem.
However, it can be easily verified that all the symmetric trapezoidal fuzzy numbers x1 , x2 , and x3 such that R( x1 ) = 35/2, R( x2 ) = 9, and R( x3 ) = 0 will be the fuzzy optimal solutions of the problem (P 2 ); for example, it can be easily verified that the following four feasible solutions are also fuzzy optimal solutions of problem (P 2 ).
The steps of the proposed Mehar method are as follows.
(P 5 ) Step 4. Use an appropriate existing method [12] to find the optimal solution of the problem (P 5 ).

Proposed Mehar Method to Deal with the Sensitivity Analysis
Kheirfam and Verdegay [11] as well as Ebrahimnejad and Verdegay [10] proposed a method to deal with the sensitivity analysis of symmetric fuzzy linear programming problem (P 1 ).
In this section, an alternative method (named as Mehar method) is proposed for the same.
The steps of the proposed Mehar method are as follows.
Step 1. Use Step 1 to Step 3 of the Mehar method, proposed in Section 3, to convert problem (P 1 ) into problem (P 5 ).
Step Step 3. Use existing sensitivity analysis technique [12] to find the optimal solution of the modified crisp linear programming problem, obtained in Step 2, with the help of the optimal table of the crisp linear programming problem (P 5 ).
Step 4. Use Step 5 of the Mehar method, proposed in Section 3, to find the fuzzy optimal solution of the modified fuzzy linear programming problem with the help of the obtained optimal solution of the modified crisp linear programming problem.
(2) Since, for applying the proposed Mehar methods, there is need to solve crisp linear programming problems, the existing and easily available software such as TORA and LINDO can be used for the same.However, for applying the existing methods [2,[4][5][6][7][9][10][11], there is need to solve fuzzy linear programming problems, so the existing and easily available software such as TORA and LINDO cannot be used and there is need to develop new software.
(3) To find the fuzzy optimal solution and to deal with the sensitivity analysis by using the existing methods [2,[4][5][6][7][9][10][11], there is need to use arithmetic operations of fuzzy numbers.While if the proposed Mehar methods are used for the same, then there is need to use arithmetic operations of real numbers.Since it is much complicated to apply the arithmetic operations of fuzzy numbers as compared to the arithmetic operations of real numbers, so it is much easy to apply the proposed Mehar methods as compared to the existing methods [2, 4-7, 9-11].

Illustrative Examples
In this section, the proposed Mehar methods are illustrated with the help of existing numerical problems.( Solution.Using the Mehar method, proposed in Section 3, the fuzzy optimal solution of the chosen problem can be obtained as follows. Step 1.Using Step 1 of the proposed Mehar method, the chosen problem can be converted into problem (P 6 ): Step 2. Using Step 2 of the proposed Mehar method, the problem (P 6 ) can be converted into problem (P 7 ): ≥ R ( 0) .(P 7 ) Step 3. Using Step 3 of the proposed Mehar method, the problem (P 7 ) can be converted into problem (P 8 ): Step 4. On solving the problem (P 8 ), the obtained optimal solution is Step 5. Using Step 5 of the proposed Mehar method, all the symmetric trapezoidal fuzzy numbers x1 , x2 , and x3 such that R( x1 ) = 35/2, R( x2 ) = 9, and R( x3 ) = 0 will be the fuzzy optimal solutions of the problem (P 2 ); for example, the following four are also fuzzy optimal solutions of the problem (P 2 ).
(  Solution.Using the Mehar method, proposed in Section 3, the fuzzy optimal solution of the problem (P 9 ) can be obtained as follows.
Step 1.Using Step 1 of the proposed Mehar method, the problem (P 9 ) can be converted into problem (P 10 ): Step 2. Using Step 2 of the proposed Mehar method, the problem (P 10 ) can be converted into problem (P 11 ): Step 3. Using Step 3 of the proposed Mehar method, the problem (P 11 ) can be converted into problem (P 12 ): Step 4. On solving the problem (P 12 ), the obtained optimal solution is Step 5. Using Step 5 of the proposed Mehar method, all the symmetric trapezoidal fuzzy numbers x1 , x2 , x3 , and x4 such that R( x1 ) = 15/8, R( x2 ) = 0, R( x3 ) = 0, and R( x4 ) = 5/8 will be the fuzzy optimal solutions of the problem (P 9 ); for example, the following four are also fuzzy optimal solutions of the problem (P 9 ).Solution.Using the Mehar method, proposed in Section 4, the desired range for  can be obtained as follows.
Step 1.Since c2 = (−4, −4, 4, 4) is replaced by c 2 = c2 +  (3,5,6,6) Step 2. Using existing sensitivity analysis technique [12] it can be easily found that if  ≤ 7/8, then the optimal solution of the modified crisp linear programming problem and hence the fuzzy optimal solution of the modified fuzzy linear programming problem will remain unchanged.
Solution.Using the Mehar method, proposed in Section 4, the desired range for  can be obtained as follows.
Example 6 ([11, Example 6, pp.187]).Find range of  so that there is no change in the fuzzy optimal solution of the fuzzy linear programming problem (P 9 ) on replacing the coefficients 1, 2, and 1 of the fuzzy variables x1 of the first, second, and third constraints by 1 + 2, 2, and 1 − , respectively.
Solution.Using the Mehar method, proposed in Section 4, the desired range for  can be obtained as follows.
Step 1.Since the coefficients 1, 2, and 1 of the fuzzy variables x1 of the first, second, and third constraints are replaced by Step 2. Using existing sensitivity analysis technique [12] it can be easily found that if 1 ≤  ≤ 2, then the optimal solution of the modified crisp linear programming problem and hence the fuzzy optimal solution of the modified fuzzy linear programming problem will remain unchanged.Solution.Using the Mehar method, proposed in Section 4, the fuzzy optimal solution of the modified fuzzy linear programming problem can be obtained as follows. Step Step 2. Using existing sensitivity analysis technique [12], optimal solution of modified crisp linear programming problem is
The proposed Mehar method is applicable only for fuzzy linear programming problems with symmetric trapezoidal fuzzy numbers.In future, proposing new methods for solving fuzzy linear programming problems with nonsymmetric trapezoidal fuzzy numbers may be tried.In this case, we simply write (  ,   , , ) ≈ (  ,   , , ) and it is to be noted that   need not be equal to   or   need not be equal to   , but (  ,   , , ) − (  ,   , , ) ≈ (−ℎ, ℎ,  + ,  + ), where ℎ = (  −   ) ≥ 0.
2. Check which of the following cases is occurring.Case i.If in the fuzzy linear programming problem (P 1 ) the fuzzy coefficient c of the th fuzzy variable x is changed by c  , then replace the coefficient   = R(c  ) of the th variable   of the crisp linear programming problem (P 5 ) by    = R(c   ).Go to Step 3. Case ii.If the fuzzy coefficient b of the fuzzy linear programming problem (P 1 ) is changed by b  , then replace the coefficient   = R( b ) of the crisp linear programming problem (P 5 ) by    of the crisp linear programming problem (P 5 ) by    .Go to Step 3. Case v.If a new fuzzy constraint ∑  =1  (+1) x ≈ b+1 is added in the fuzzy linear programming problem (P 1 ), then add a new constraint ∑  =1  (+1)   = R( b+1 ) in the crisp linear programming problem (P 5 ).Go to Step 3.