This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 6 The simplex method We next formalize the method we developed in the previous section. We again consider a general linear program in standard equality form: maximize c T x subject to Ax = b x ≥ . As before, we introduce a new variable z to keep track of the objective value, so the system of equations defining the linear program is z- c T x = 0 Ax = b. (6.1) We break down each iteration of the solution algorithm, which is called the simplex method , into a sequence of steps. Step 0: Initialization. We (somehow) find an initial feasible basis B , form a list N consisting of its complement, and find a corresponding tableau: z- X j ∈ N ¯ c j x j = ¯ v x i + X j ∈ N ¯ a ij x j = ¯ b i ( i ∈ B ) . (6.2) This system is equivalent to the original constraints (6.1), and from it, we can read off the corresponding basic feasible solution: basic variables x i = ¯ b i ( i ∈ B ) nonbasic variables x j = 0 ( j ∈ N ) objective value z = ¯ v. Step 1: Check optimality. If each reduced cost satisfies ¯ c j ≤ 0 (for all indices j ∈ N ), then we stop: the current solution is optimal . To see this, notice that any feasible solution satisfies z = ¯ v + X j ∈ N ¯ c j x j ≤ ¯ v, 35 since ¯ c j ≤ 0 and x j ≥ 0 (for all j ∈ N ). Thus no feasible solution can have objective value larger than ¯ v . But the current solution attains this value, so must be optimal. Step 2: Choose entering index. If we don’t stop in Step 1, we can find an entering index k ∈ N with corresponding reduced cost ¯ c k > 0. Consider now the effect of increasing the value of the corresponding variable x k (called the entering variable ) away from its current value zero to a new value t ≥ 0. At the next iteration of the simplex method, if the value of x k is nonzero, the index k must have entered the basis: hence the term “entering index”....
View Full Document
- Fall '07
- Optimization, basic feasible solution, objective value, basic variables