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: Advanced Linear Programming: The Exercises The answers are sometimes not written out completely. 1.5 a) min c T x + d T y s . t . Ax + By ≤ b y =  x  (1) First reformulation, using z smallest number satisfying x ≤ z and x ≤ z : min c T x + d T z s . t . Ax + Bz ≤ b x ≤ z x ≤ z. (2) Second reformulation, using x = x + x and  x  = x + + x : min c T x + c T x + d T x + + d T x s . t . Ax + Ax + Bx + + Bx ≤ b x + ≥ x ≥ . (3) 1.5 b) Suppose (1) has feasible solution x (hence y =  x  ), then choosing x + i = x i and x i = 0 if x i ≥ 0 and x + i = 0 and x i = x i if x i ≤ 0 gives a feasible solution of (3). Moreover, the solution values coincide, implying that the optimal value of (3) is at most that of (1). Suppose (3) has feasible solution x + ,x with the property that for all i = 1 ,...,n , x + i > ⇒ x i = 0 and x i > ⇒ x + i = 0. Then setting x i = x + i x i and z = x + i + x i gives a feasible solution for (2) with equal value, implying that the optimal value of (2) is at most that of (3). If (3) has a feasible so lution that does not have the property then there are variables x + i > 0 and x i > 0. Setting ˆ x + i = x + i min { x + i ,x i } and ˆ x i = x i min { x + i ,x i } gives c i (ˆ x + i ˆ x i ) = c i ( x + i x i ) and, since d i ≥ 0 and the ith row of B is non negative, we obtain again a feasible solution and d i (ˆ x + i ˆ x i ) < d i ( x + i x i ). Hence a better feasible solution. Suppose (2) has feasible solution x,z satisfying for all i = 1 ,...,n that z i = x i or z i = x i . Then, setting y = z and keeping x as it is is a feasible solution of (1) with equal value, implying that the optimal value of (1) is at most that of (2). If (2) has a feasible solution that does not have the property then there exist variables z i > x i and z i > x i . Clearly, since d i ≥ 0 and the ith row of B is nonnegative, diminishing the value of z i yields another feasible solution with better objective value. 1 1.5 c) The set { x ∈ IR  x 2  x  ≤  1 ,  x  ≤ 1 } is not even connected, let alone convex. Minimizing x gives local optimum x = 1, whereas x = 1 is the global optimum. ——————— 1.15 a) min (9 1 . 2) x 1 + (8 . 9) x 2 s . t . 1 4 x 1 + 1 3 x 2 ≤ 90 1 8 x 1 + 1 3 x 2 ≤ 80 x 1 ≥ , x 2 ≥ (4) 1.15 b(i)) min (9 1 . 2) x 1 + (8 . 9) x 2 7 x 3 7 x 4 s . t . 1 4 x 1 + 1 3 x 2 x 3 ≤ 90 1 8 x 1 + 1 3 x 2 x 4 ≤ 80 x 3 + x 4 ≤ 50 x 1 ≥ , x 2 ≥ , x 3 ≥ , x 4 ≥ (5) 1.15 b(ii)) The easiest way is to solve the LP with the discounted price for raw material. If the optimal solution yields a total bill for raw materials of at least of $300 then this is the best decision. If not, solve the LP as in a) and solve the LP with the extra restriction that the total raw material cost is $300. Choose the best of these two solutions....
View
Full
Document
This note was uploaded on 02/15/2011 for the course EECS 6.231 taught by Professor Bertsekas during the Spring '10 term at MIT.
 Spring '10
 Bertsekas

Click to edit the document details