Volume-3, Issue-6, December-2013, ISSN No.: 2250-0758 International Journal of Engineering and Management Research Available at: www.ijemr.net Page Number: 73-75 A New Variant of Simplex Method Dr. R.G. Kedia Assistant Professor & Head, Department of Statistics, Govt. Vidarbha Institute of Science & Humanities, Amravati, INDIA ABSTRACT In this paper a new variant of simplex method is proposed to solve a class of linear programming problem having all constraints≥ type. In the proposed method, instead of using artificial basis techniques we start with infeasible solution obtained by converting all the constraints into≤ type. Then by judicious selection of entering and leaving vector in and from the initial simplex table, feasible solution is obtained in the first iteration only and optimal solution then can be obtained using regular simplex method. Keywords: Linear programming problem, Simplex method, artificial variable, dual simplex method, I. INTRODUCTION Linear programming model is extensively used to solve the variety of problems in Engineering and Management field particularly for production planning and optimum resources allocation. To solve linear programming problem, Dantzig devised simplex method in 1947. ASimplex algorithm of Dantzig is one of the top 10 algorithms of 20th century [Barry Cipra (3)]. The computational complexity in simplex method depends on the number of variables and the number of constraints and is directly proportional to both. To reduce the complexity, several variants of simplex method are developed by various researchers. For details Gass (2) can be referred. If in the given linear programming problem, one or more constraints are ≥ type or equality, simplex method with artificial basis techniques is used. Big-M method and two phase simplex method are quite commonly used. In these methods artificial variables are used to get the standard basis artificially. Artificial variables are then forced to leave the basis step by step. Once all the artificial variables are removed from the basis, optimal solution is then obtained by using regular simplex method. However, use of artificial variables to get the standard basis increases the computational complexity due to the reason mentioned above. To avoid the use of artificial variables, the≥ type constraints are converted into ≤ type constraints by multiplying by -1. Then using slack variables only, initial basic solution is obtained that is not feasible. In other words, one starts with infeasible solution. Methods are developed to solve linear programming problem by starting with infeasible solution and force the solution to be feasible as well as optimal at some iteration. One such method namely dual simplex method devised by Lemke (5) is most popular. In dual simplex method, starting from infeasible solution feasible solution is obtained step by step. Another basis-exchange pivoting algorithm is the criss-cross algorithm by Terlaky, Tamás (4) .There are polynomial-time algorithms for linear programming that use interior point methods: These include Khachiyan's ellipsoidal algorithm, Karmarkar's projective algorithm, and pathfollowing algorithms. Interested readers can refer Robert J. Vanderbeib (1). In the proposed method in this article, we start with initial basic solution that is infeasible. Then by judicious selection of entering and leaving vector in and from the initial simplex table, feasible solution is obtained in the first iteration only and optimal solution then can be obtained using regular simplex method. Notations: The proposed method can be used to solve the linear programming problem of the type For further discussion, following notations will also be used: a j will denote the jth column in A matrix. Bi will denote ith basis vector (row). x Bi will denote ith solution term. s i will denote the slack variable in ith constraint. Other notations will have their usual meaning. II. PROPOSED METHOD Let the given linear programming problem be We assume that there is no equality constraint Rewrite the given problem as 73 This completes the proof. III. PROPOSED METHOD The working of the method is as follows. The corresponding problem is standard linear programming Let the given linear programming problem be The given problem can be converted into maximization type as follows: First we prove the following theorem. Theorem: Let x B = -b be the initial basic infeasible solution to the problem. Now, select the non-basic vector all the elements of which are negative to enter the basis all the elements of which are negative. If there are more than such vectors then among these, the vector for which z j –c j (= c j , as z j = 0 for all j) is least positive is selected to enter the next basis. This is because the objective function is to be minimized and cj with minimum value will contribute least. Let a s be entering vector. Next, we compute the ratioӨ i = x Bi /-a is = -b i /-a is = b i /a is . The basis vector for whichӨ i is most positive is selected to leave the current basis. ӨLet r is most positive. Therefore, r th basis vector i.e. Br leave the current basis and -a rs is the pivotal element. The elements in the next iteration are computed as follows: If there exist at least one non basis vector a j such that all a ij < 0 then basic feasible solution can be obtained in the first iteration only. Proof: Let a s be non basis vector such that all a is < 0. Select a s to enter in the basis of first iteration. Compute the ration Ө i = x Bi /-a is = -b i /-a is = bi /a is for basis vector and select the basis vector to leave the current basis for which Ө i is maximum positive. Let Br be the leaving vector. The new solution will be The solution obtained in first iteration using (3.1) and (3.2) will be feasible and now regular simplex method can be used. The steps in the proposed method can be stated as follows: Since we want ⏞ 𝑥𝑥 Bi ≥ 0, we must have for all i R 74 Step3. Write down initial simplex table. Step4. Select non-basic vector a s to enter in the basis of first iteration such that all the elements of it are negative. If there exists more than one such non-basic vectors, then the vector for which z j –c j (= c j , as z j = 0 for all j) is least positive. Let it be a r . Step5. For the basis vectors (rows) compute the ratio Step5. Select the outgoing vector from the current basis for which Ө i is most positive. Let it be a r . Therefore, a rs is the pivotal element. Step6. Write down the simplex table for first iteration as in regular simplex method. The elements in the simplex table for first iteration will be computed by equations (3.1), (3.2), (3.3) and (3.4). The new solution is basic feasible solution. Step7. Proceed with regular simplex method till optimal solution is obtained or there is indication that solution does not exists. Example: Consider the problem, Iteration 1: (Note that the solution in first iteration is feasible. So we can proceed with regular simplex method. Since z j –c j (7) is most negative for a 1 , it is entering and s 2 is leaving vector as shown in above table.) Iteration 2: Corresponding standard linear programming problem is REFERENCES [1] Robert J. Vanderbeib: Linear Programming: Foundations and Extensions, 3rd ed., International Series in Operations Research & Management Science, Vol. 114, Springer Verlag, 20, 2008 [2] Gass, S.: Linear programming, methods and Applications, 5th ed. McGraw-Hill, New York, 2003 [3] Barry Cipra: The Best of the 20th Century: Editors Name Top 10 Algorithms SIAM News, Volume 33, Number 4, page 1, 2000. [4] Terlaky, Tamás : "A finite crisscross method for oriented matroids". Journal of Combinatorial Theory. Series B 42 (3): 319–327, 1987 [5] Lemke, C. E. (1954): The dual simplex method of solving the linear programming problem, Naval Res. Logist. Quart., Vol. 1, 36-47. Copyright © 2011-13. Vandana Publications. All Rights Reserved. 75
© Copyright 2026 Paperzz