This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 8 Phase 1 of the simplex method Let us summarize the method we sketched in the previous section, for finding a feasible basis for the constraint system (8.1) n X j =1 a ij x j = b i ( i = 1 , 2 , . . . , m ) x j ≥ ( j = 1 , 2 , . . . , n ) . We can assume each righthand side b i is nonnegative: otherwise we multiply the corresponding equation by 1. The idea of “Phase 1” of the simplex method is as follows: correspond ing to each constraint, introduce an “artificial variable” x n + i ≥ 0 (for i = 1 , 2 , . . . , m ), which we think of as the error in that constraint. We therefore consider the system ( S ) X j a ij x j + x n + i = b i ( i = 1 , 2 , . . . , m ) . We then try to force the values of these artificial variables x j , j > n , to zero, by using the simplex method to minimize their sum: (8.2) min n m X i =1 x n + i : ( S ) holds , x ≥ o We transform this Phase 1 problem to standard equality form by multiplying the objective function by 1. An initial feasible tableau is readily available for this problem: we choose x n +1 , x n +2 , . . . , x n + m as the initial list of basic variables, and eliminate them from the top equation defining the Phase 1 objective function using row operations. We then begin the simplex method. If, after some iteration of the Phase 1 simplex method, we arrive at a tableau ( T ) where one of the new variables, x n + r say, has become nonbasic, then before proceeding to the next iteration we can delete all the appearances of the artificial variable x n + r in ( T ) to obtain a system ( T ) . To justify this process, notice that it is equivalent to setting x n + r = 0 throughout ( T ). Consider the effect of editing the system ( S ) by setting the variable x n + r to zero. This new system, ( S ) say, shares the key motivating property with ( S ): the original system of constraints (8.1) has a feasible solution exactly 42 when the the linear program min n X i 6 = r x n + i : ( S ) holds , x j ≥ 0 ( j = 1 , /ldots, n ) , x n + i ≥ 0 ( i = 1 , /ldots, m, i 6 = r ) o has a feasible solution with objective value zero, and furthermore the system ( T ) is a feasible tableau for this linear program....
View
Full
Document
This note was uploaded on 02/20/2009 for the course ORIE 3300 taught by Professor Todd during the Fall '08 term at Cornell.
 Fall '08
 TODD

Click to edit the document details