LINEAR PROGRAMMING AND ITS EXTENSIONS (NPTEL) ASSIGNMENT Consider the standard LPP(3) with extreme points X,,X,,X4,X, and extreme directions d,,d, and d, such that peers wt = pees ot a Maes Cc X)=5, CX, =7, Cc X3= =C XpC d,=0, Cd, =0 and Cid; =4. Characterise all the optimal solutions, of the LPP.. Note: Finite optimal solutions exist even though directions are present in the feasible region. Try to explain why ? 4 Use a linear programming formulation to show that the constraints 2x a) 7x3 424 +5 <3 3x4 +4, +443 —5x 4-25 s+, 7 20Vj imply ~6xy+8x,+7x5—9x4 5x5 >—18. Obtain the set of alternate optimal solutions given the following optimal tableau: eee ae X, x X, RHS Oita -Oss2 0: 0 2 3: 9 1 0. |2 -l -1 1 4 0 feild: i 2: 3. B4, Consider the problem: Min 2x, -X5 ~5x3-3xy St. x +2x, + 4x, x4 = 2x, $39 — 43 +4 =12 x 4x3 txya4 HpXy0X30%4 20 Find a bfs with the basic variables as XpXy & Xy: Is this solution optimal? 5. The starting and current tableaux of a given LPP are shown. Find the values of the unknowns a through ¢ Starting Tableau ke a RHS 0 [0 [0 Tent) ota 6 of fi Current Tableau Z_[x |x. [x [x _[x, [RHS ze [1 [0 ft fi Kee peal aaeala I Orie stl s eee Ose ee ee eel 2 Z 6. Read each of the following statements carefully and check whether it is true or false. Justify your answer by constructing asimple illustrative example, if possible, or by a simple proof. In all these statements, problem (1) is the LP. Minimize 2(x)=cx Subject to Ax—b x20 where 4 is a matrix of order mxn. K is the set of feasible solutions of this problem (1) f(6) is the minimum objective value in this problem, as a function of the right — hand side constant vector, b. a The number of positive x by BFS of (1) can never exceed the rank of A. The unboundedness criterion will never be satisfied while solving the Phase I problem. If (1) is feasible, termination occurs in solving the Phase I problem associated with (1) only when all the artificial variables leave the basic vector. If a tie occurs in the pivot row choice during a pivot step while solving (1) by the simplex algorithm, the BFS obtained after this pivot step is degenerate. Simplex algorithm moves from an extreme point of K only in ano degenerate pivot step. In solving an LP by the simplex algorithm, a new feasible solution is generated after every pivot step. Let the rank of A in (1) be m. If (1) has an optimum solution, there must exist an optimum basis. Every basis consists of m column vectors and a pivot step changes one column vector in a basis. Hence starting from a feasible basis, the simplex algorithm finds an optimum basis in at most m pivot steps. Every feasible solution of (1) in which m variables are positive is a BFS.20. 2 If ¥ and j are two adjacent BFSs of (1), the total number of variables that are positive in either ¥ or ¥, or both , is at most m+1. No feasible solution of (1) in which m+2 or more variables are positive can lie on an edge of K. Every convex set has an extreme point. Every nonempty convex polytope has an extreme point. If (1) has an optimum solution, K must be a bounded set. An LP can have more than one optimum solution iff it is degenerate. The total number of optimum feasible solutions of any LP is always finite. The total number of BFSs of any LP is always finite. The set of optimum solutions of any LP is always a bounded set. If (1) has an optimum solution in which m+1 or more variables are positive, it must have an uncountable number of optimum solutions. If rank of A is m, and (1) is nondegenerate, every feasible solution in which exactly m variables are positive, must be a BFS. While solving (1) by the simplex algorithm, if the BFS in the beginning of a pivot step is degenerate, the objective value remains unchanged in this pivot step. . Can a vector which has just lett the basis in the simplex algorithm recenter on the very next pivot? Hint. If xp is the out going variable, compute its i. new C. Zp and show that it will be positive. By. r