Note 5: Duality and Dual Simplex Method

If the primal problem (P) is max Z = n cj xj j=1 n s.t. aij xj bi i = 1, 2, ..., m j=1 xj 0 j = 1, 2, ..., n, then the dual problem (D) is min W = m bi yi i=1 m s.t. aij yi cj j = 1, 2, ..., n i=1 yi 0 i = 1, 2, ..., m.

(1)

(2)

There are m food types and n nutritional ingredients. Each unit of food type i contains aij units of nutritional ingredient j. Food type i sells for bi , while a persons daily requirement of nutritional ingredient j is cj . Let xj be the fair worth of nutritional ingredient j to the person and yi be the number of units of food type i to be bought daily by the person. Then (P) maximizes the total required fair worth of all nutritional ingredient on the basis that the total fair worth of all nutritional ingredient of any food type does not exceed its selling price, while (D) minimizes the total cost of satisfying the daily nutritional needs. In essence, (P)s variable corresponds to (D)s constraint, and (D)s variable corresponds to (P)s constraint; If (P) is a maximization (minimization) problem, then (D) is a minimization (maximization) problem; (P)s free variable corresponds to (D)s = constraint and its nonnegative variable corresponds to (D)s max / constraint (meaning that (D) is a maximization problem and the constraint is of the type) or min / constraint; on the other hand, (D)s free variable corresponds to (P)s = constraint, its nonnegative variable corresponds to (P)s max / or min / constraint. The dual of dual is the primal itself. Weak duality: If x is a feasible solution for the primal problem and y is a feasible solution for the dual problem, then cx bT y. 1

Reason: Because x 0 and AT y cT or equivalently yT A c, we have cx yT Ax. Since Ax = b, we have yT Ax = yT b = bT y. Strong duality: If x is an optimal solution for the primal problem and y is an optimal solution for the dual problem, then cx = bT y . Reason: Let B be the optimal basis for the primal problem, then it satises both B1 b 0 and cN cB B1 N. Let x = (xT , xT )T and y = (cB B1 )T . It is obvious B N that x is primal feasible (feasible for the primal problem) and y is dual feasible (feasible for the dual problem). But it is obvious that cx = bT y. From the weak duality, y must also be dual optimal. Complementary slackness: We always have xj ( m aij yi cj ) 0 for j = i=1 1, 2, ..., n for any primal feasible x and dual feasible y. Also, x ( m aij yj cj ) = 0 i=1 j for j = 1, 2, ..., n if and only if x is primal optimal and y is dual optimal. Reason: The one for the feasible x and y can be derived from the weak duality property. The one for the optimal x and y can be derived from the strong duality property. In plain English, if the primal problem is of the maximization type and the dual problem is of the minimization type, then the primal objective value is always below the dual objective value. But the maximum the primal can achieve is exactly the minimum the dual can achieve. That is why a basis being primal/dual optimal leads to it being dual/primal feasible, and why a basis being primal and dual feasible leads to it being optimal, in both primal and dual senses. Also, an optimal primal/dual variable must be 0 when its corresponding dual/primal constraint is not tight at optimality, and a primal/dual constraint must be tight at optimality when its corresponding dual/primal variable is not 0 at optimality. Duality Theorem: The following are the only possible relationships between the primal and dual problems: 1. If one problem has feasible solutions and a bounded objective function (and so has an optimal solution), then so does the other problem, so both the weak and strong duality properties are applicable. 2. If one problem has feasible solutions and an unbounded objective function (and so no optimal solution), then the other problem has no feasible solutions. 3. If one problem has no feasible solutions, then the other problem has either no 2

feasible solutions or an unbounded objective function. Dual simplex method: When (P) is of the standard form, (D) is of the following form: w = bT y subject to
BT y + ySB = cB ,

(3)

NT y + ySN = cN , y is free, while ySB , ySN are nonnegative.

(4)

Here, ySB can be viewed as the nonbasic variables and y and ySN the basic variables. We have B y = BT (cB yS ), B (5) yN = cN NT BT cB + NT BT yS , S T T T T B w = b B cB b B yS . When cT cT B1 N 0, (D) is feasible. When (B1 b)i < 0, we may increase w by B N increasing ySi . To guarantee the nonnegativity of ySN , the extent to which ySi can increase is bounded by = min{ cT cT B1 N N B | (B1 N)ij < 0}. 1 N) (B ij (6)

When (B1 N)i is nonnegative, (D) is unbounded and (P) is infeasible. The detailed steps of Dual Simplex Method: Step 1. Is the right-hand side of each constraint nonnegative? If so, an optimal solution has been found; if not, at least one constraint has a negative right-hand side, and we go to step 2. Step 2. Choose the most negative basic variable as the variable to leave the basis. The row in which the variable is basic will be the pivot row. To select the variable that enters the basis, we compute the following ratio for each variable xj that has a negative coecient in the pivot row: (Coecient of xj in row 0)/(Coecient of xj in pivot row). Choose the variable with the smallest ratio (absolute value) as the entering variable. This form of the ratio test maintains a dual feasible tableau (all variables in row 0 has nonnegative coecients). Now use eros to make the entering 3

variable a basic variable in the pivot row. Step 3. If there is any constraint in which the right-hand side is negative and each variable has a nonnegative coecient, the LP has no feasible solution. If no constraint indicating feasibility is found, return to Step 1. Sensitivity analysis suitable for dual simplex method: Addition of a new constraint and change of right-hand side. Example: max Z = 60x1 + 30x2 + 20x3 s.t. 8x1 + 6x2 + x3 48 4x1 + 2x2 + 1.5x3 20 2x1 + 1.5x2 + 0.5x3 8.