chap6_handout

chap6_handout - Chapter 6: Duality and Sensitivity Analysis...

Info iconThis preview shows pages 1–11. Sign up to view the full content.

View Full Document Right Arrow Icon
I Chapter 6: Duality and Sensitivity Analysis 1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Introduction Introduction I Consider the following LP: maximize Z = 30 x 1 + 20 x 2 subject to x 1 + 2 x 2 6 (1) 2 x 1 + x 2 8 (2) x 2 2 (3) - x 1 + x 2 1 (4) x 1 , x 2 0 I Here is a feasible solution: ( x 1 , x 2 ) = (2 , 1) with Z = 80. I Is it optimal? If not, How far from optimal is it? I We could answer by solving the LP using simplex. Is there another way? 2
Background image of page 2
Introduction Lower and Upper Bounds I Since (2 , 1) is a feasible solution with Z = 80, 80 is a lower bound on the optimal objective value. I Why? I ( 10 3 , 4 3 ) is also a feasible solution, with Z = 380 3 = 126 . 67, so 126.67 is also a LB. I What if I told you that 160 is an upper bound on the optimal objective value? I Then ( 10 3 , 4 3 ) is no more than 160 - 126 . 67 = 33 . 33 units away from optimal. I If I could prove that 126.67 is also an upper bound, then ( 10 3 , 4 3 ) must be optimal. I Why? I How can we find such an upper bound? 3
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Introduction An Upper Bound max Z = 30 x 1 + 20 x 2 s.t. x 1 + 2 x 2 6 (1) 2 x 1 + x 2 8 (2) x 2 2 (3) - x 1 + x 2 1 (4) I If we multiply (2) by 20, we get 40 x 1 + 20 x 2 160 . I So any feasible solution satisfies 40 x 1 + 20 x 2 160. I The objective function is 30 x 1 + 20 x 2 . I So any feasible solution must satisfy 30 x 1 + 20 x 2 160. I Therefore 160 is an upper bound on the optimal objective value. 4
Background image of page 4
Introduction Another Upper Bound max Z = 30 x 1 + 20 x 2 s.t. x 1 + 2 x 2 6 (1) 2 x 1 + x 2 8 (2) x 2 2 (3) - x 1 + x 2 1 (4) I Multiply (1) and (2) by 10 and add them: 30 x 1 + 30 x 2 140 . I Since 30 x 1 + 20 x 2 30 x 1 + 30 x 2 140 , 140 is another (even better) upper bound. I How can we find the best (i.e., smallest) upper bound? 5
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Introduction Finding the Best Upper Bound max Z = 30 x 1 + 20 x 2 s.t. x 1 + 2 x 2 6 (1) 2 x 1 + x 2 8 (2) x 2 2 (3) - x 1 + x 2 1 (4) I We want multipliers for each constraint. I Call them y 1 , y 2 , y 3 , y 4 . I The multipliers must be chosen such that: I multipliers times x 1 ’s constraint coefficients 30. I multipliers times x 2 ’s constraint coefficients 20. I We also want the sum of the multipliers times the RHSs to be as small as possible. I Also, the multipliers have to be nonnegative. 6
Background image of page 6
Introduction Finding the Best UB, cont’d max Z = 30 x 1 + 20 x 2 s.t. x 1 + 2 x 2 6 (1) 2 x 1 + x 2 8 (2) x 2 2 (3) - x 1 + x 2 1 (4) I In other words, we want to minimize W = 6 y 1 + 8 y 2 + 2 y 3 + y 4 subject to y 1 + 2 y 2 - y 4 30 2 y 1 + y 2 + y 3 + y 4 20 y 1 , y 2 , y 3 , y 4 0 7
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Introduction The Dual Problem I This new problem is called the dual problem . I The original problem is called the primal problem . I The optimal objective function value for the dual problem gives an upper bound on the optimal objective function value for the primal problem: Z * W * . I How close is the UB? I We’ll answer this shortly. 8
Background image of page 8
The Essence of Duality Theory (Section 6.1) Formulating the Dual Formulating the Dual I In standard form, the primal and dual problems look like this: Primal: (P) maximize Z = n P j =1 c j x j subject to n P j =1 a ij x j b i i = 1 , . . . , m x j 0 j = 1 , . . . , n Dual: (D) minimize W = m P i =1 b i y i subject to m P i =1 a ij y i c j j = 1 , . . . , n y i 0 i = 1 , . . . , m 9
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
The Essence of Duality Theory (Section 6.1)
Background image of page 10
Image of page 11
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 77

chap6_handout - Chapter 6: Duality and Sensitivity Analysis...

This preview shows document pages 1 - 11. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online