1 / 27

# Lecture 25

Lecture 25. Two-Phase Simple Upper Bounded Simplex Algorithm Example: Minimize â€“2x 1 â€“ x 2 Subject to x 1 + x 2 &lt; 6 0 &lt; x 1 &lt; 5 0 &lt; x 2 &lt; 5. The Graph. x 2 bound. x 2. x 1 bound. optimum. x 1. Convert To An Equality Constraint. Minimize â€“2x 1 â€“ x 2

Download Presentation

## Lecture 25

An Image/Link below is provided (as is) to download presentation Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

### Presentation Transcript

1. Lecture 25 • Two-Phase Simple Upper Bounded Simplex Algorithm • Example: • Minimize –2x1 – x2 • Subject to x1 + x2< 6 • 0 < x1< 5 • 0 < x2< 5

2. The Graph x2 bound x2 x1 bound optimum x1

3. Convert To An Equality Constraint • Minimize –2x1 – x2 • Subject to x1 + x2 +x3 = 6 • 0 < x1< 5 • 0 < x2< 5 • 0 < x3< Note Bound on the slack variable

4. Add An Artificial Variable • Minimize –2x1 – x2 • Subject to x1 + x2 +x3 + A= 6 • 0 < x1< 5 • 0 < x2< 5 • 0 < x3< • 0 < A <

5. Phase I Problem • Minimize A • Subject to x1 + x2 +x3 + A= 6 • 0 < x1< 5 • 0 < x2< 5 • 0 < x3< • 0 < A <

6. Step 0. Initialization • BASIC = {A at 6} • NONBASIC = {x1 at 0, x2 at 0, x3 at 0} • B = 1 • B-1 = 1

7. Step 1. Pricing • v = cBB-1 = 1(1) = 1 • cbar1 = c1 – va1 = 0 – 1(1) = -1 • cbar2 = c2 – va2 = 0 – 1(1) = -1 • cbar3 = c3 – va3 = 0 – 1(1) = -1 • All price favorably!

8. Step 2. Optimality • Not optimal • Let p = 3 - that is x3 is allowed to increase from 0

9. Step 3. Direction • y = B-1a3 = 1(1) = 1 •  = 1 Why?

10. Step 4 Step Size • - = min{, (xj-ℓj)/(yj) : yj > 0} • + = min{, (uj-xj)/(-yj) : yj < 0} •  = min{- , + , up - ℓp} • - = min{, (6-0)/1} = 6 • + = min{} =  •  = min{6, , -0} = 6

11. Step 5. New Point • XB = xB - y = 6 – 6(1)(1) = 0 • A = 0 • xp = x3 = ℓ3 +  = 0 + 6 = 6 • Current Point: [x1,x2,x3,A] = [0,0,6,0] • BASIC = {x3 at 6} • NONBASIC = {x1 at 0, x2 at 0, A at 0} • B = 1 B-1 = 1

12. Step 1. Pricing • v = cBB-1 = 0(1) = 0 • cbar1 = c1 – va1 = 0 – 0(1) = 0 • cbar2 = c2 – va2 = 0 – 0(1) = 0 • All price unfavorably!

13. Step 2. Optimality • Optimal For Phase I

14. Phase II Problem • Minimize -2x1 – x2 • Subject to x1 + x2 +x3 = 6 • 0 < x1< 5 • 0 < x2< 5 • 0 < x3<

15. Step 1. Pricing • Current Point: [x1,x2,x3] = [0,0,6] • BASIC = {x3 at 6} • NONBASIC = {x1 at 0, x2 at 0} • B = 1 B-1 = 1 • v = cBB-1 = 0(1) = 0 • cbar1 = c1 – va1 = -2 – 0(1) = -2 • cbar2 = c2 – va2 = -1 – 0(1) = -1 • All price favorably!

16. Step 2. Optimality • Not Optimal • Let p = 1

17. Step 3. Direction • y = B-1a1 = 1(1) = 1 •  = 1 Why?

18. Step 4 Step Size • - = min{, (xj-ℓj)/(yj) : yj > 0} • + = min{, (uj-xj)/(-yj) : yj < 0} •  = min{- , + , up - ℓp} • - = min{, (6-0)/1} = 6 • + = min{} =  •  = min{6, , 5-0} = 5

19. Step 5. New Point • XB = xB - y = 6 – 5(1)(1) = 1 • x3 = 1 • xp = x1 = ℓ1 +  = 0 + 5 = 5 • Current Point: [x1,x2,x3] = [5,0,1] • BASIC = {x3 at 1} • NONBASIC = {x1 at 5, x2 at 0} • B = 1 B-1 = 1 Note: The basis stayed the same.

20. Step 1. Pricing • v = cBB-1 = 0(1) = 0 • cbar1 = c1 – va1 = -2 – 0(1) = -2 Unfavorable Why? • cbar2 = c2 – va2 = -1 – 0(1) = -1 Favorable

21. The Graph x2 bound x2 x1 bound current point x1

22. Step 2. Optimality • Not Optimal • Let p = 2

23. Step 3. Direction • y = B-1a1 = 1(1) = 1 •  = 1 Why?

24. Step 4 Step Size • - = min{, (xj-ℓj)/(yj) : yj > 0} • + = min{, (uj-xj)/(-yj) : yj < 0} •  = min{- , + , up - ℓp} • - = min{, (1-0)/1} = 1 • + = min{} =  •  = min{1, , 5-0} = 1

25. Step 5. New Point • XB = xB - y = 1 – 1(1)(1) = 0 • x3 = 0 • xp = x2 = ℓ2 +  = 0 + 1 = 1 • Current Point: [x1,x2,x3] = [5,1,0] • BASIC = {x2 at 1} • NONBASIC = {x1 at 5, x3 at 0} • B = 1 B-1 = 1

26. Step 1. Pricing • v = cBB-1 = -1(1) = -1 • cbar1 = c1 – va1 = -2 – (-1)(1) = -1 Unfavorable Why? • cbar3 = c3 – va3 = 0 – (-1)(1) = 1 Unfavorable Why? • Optimality Obtained

27. The Graph x2 bound x2 x1 bound current point x1

More Related