Linear Programming Homework

Homework 0 2
Lecturer: Solution 1 Given two convex sets S1 and S2 in Rn . S1 S2 is convex: Observe that for any vectors x, y S1 S2 and for any [0, 1], we have that x + (1 )y S1 and x + (1 )y S2 since S1 and S2 are convex. Hence, x + (1 )y S1 S2 and so S1 S2 is convex. S1 S2 = {x1 + x2 : x1 S1 , x2 S2 } is convex: Observe that for any vectors x, y S1 S2 and for any [0, 1], there exists x1 , y1 S1 and x2 , y2 S2 such that x1 + x2 = x S1 S2 and y1 + y2 = y S1 S2 . Then since S1 and S2 are convex we have that (x1 + (1 )y1 ) S1 and (x2 + (1 )y2 ) S2 . So, for any [0, 1]: (x1 + (1 )y1 ) + (x2 + (1 )y2 ) = (x1 + x2 ) + (1 )(y1 + y2 ) = x + (1 )y Year: Fall 2012
Hence, since we wrote x + (1 )y in terms of vectors in S1 and S2 and the fact that sum of two convex functions is convex we have that x + (1 )y S1 S2 . S1 S2 = {x1 x2 : x1 S1 , x2 S2 } is convex:
Observe that for any vectors x, y S1 S2 and for any [0, 1], there exists x1 , y1 S1 and x2 , y2 S2 such that x1 x2 = x S1 S2 and y1 y2 = y S1 S2 . Then since S1 and S2 are convex we have that (x1 + (1 )y1 ) S1 and (x2 + (1 )y2 ) S2 . So, for any [0, 1]: (x1 + (1 )y1 ) + ((x2 + (1 )y2 )) = (x1 + (1 )y1 ) (x2 + (1 )y2 ) = (x1 x2 ) + (1 )(y1 y2 ) = x + (1 )y
Hence, since we wrote x + (1 )y in terms of vectors in S1 and S2 and the fact that sum of two convex functions is convex we have that x + (1 )y S1 S2 .
0-1
Homework 2
Srikanth Potukuchi
Solution 2 Variables: pi , Mi for i {X, Y, Z} Constraints: p i Mi q 1 + q 2

i{X,Y,Z} 5
q1 + q2
i=1
dk .
Objective Function: min

j{P1 ,P2 } i{X,Y,Z}
ci,j pi Mi
0-2
Homework 2
Srikanth Potukuchi
Solution 3 We want to solve the following problem using the graphical method and then write the problem in standard form: min 5x1 + x2 s.t x1 + 3x2 6 x1 + 6x2 6 x1 x2 2 x2 0. The standard form with x1 = x+ x since x1 is unconstrained in sign and adding or subtracting one slack 1 1 variable to each constraint accordingly: min 5x+ 5x + x2 1 1 s.t x+ x + 3x2 + x3 = 6 1 1 x+ x + 6x2 x4 = 6 1 1 x+ x x2 + x5 = 2 1 1 x+ , x , x2 , x3 , x4 , x5 0. 1 1
0-3
Homework 2
Srikanth Potukuchi
Solution 4 We want to prove that the following functions are convex: Given f (x) = max{f1 (x), f2 (x), , fk (x)}, with fi (x) convex for i = 1, 2, , k. Observe that we need to show for arbitrary vectors x, y Rn and for any [0, 1]: f (x + (1 )y) f (x) + (1 )f (y). Now observe for arbitrary vectors x, y Rn and for any [0, 1]: f (x + (1 )y) = max{fi (x + (1 )y)} i max{fi (x) + (1 )fi (y)} ( Here i is max and xed and the inequality is due to convexity) max{fi (x)} + max{(1 )fi (y)} ( property of max function)
i i
= max {fi (x)} + (1 ) max {fi (y)} ( property of max function)

i=1, ,k i=1, ,k
= f (x) + (1 )f (y) We want to show that f (x) = |x| is convex: Observe that we can take f1 (x) = x and f2 (x) = x where both are convex functions. Thus the maximum of these functions gives us f (x) and from previous part of this problem we can conclude that f (x) = |x| is convex.
0-4
Homework 2
Srikanth Potukuchi
Solution 5 We have two problems to solve: Problem 1.4: We want to reformulate the following problem as a LP problem: minimize 2x1 + 3|x2 10| subject to |x1 + 2| + |x2 | 5 We dene z1 = |x2 10|, z2 = |x1 + 2|, z3 = |x2 |. So we have the following inequalities: z1 x2 10 z1 x2 + 10 z2 x2 + 2 z2 x2 2 z3 x2 z3 x2 So Reformulation: minimize subject to z1 x2 10 z1 x2 + 10 z1 + z3 5 z2 x2 + 2 z2 x2 2 z3 x2 z3 x2 Problem 2.1:For each of the following sets, determine whether it is a polyhedron: (a) Given : The set of all (x, y)
2
2x1 + 3z1
satisfying the constraints
x cos + y sin 1, [0, /2] x 0, y 0. This cannot be a polyhedron since we have innitely possible values for the constraint x cos + y sin 1, [0, /2] and A is a m n matrix with nite entries. I think if we pick a special matrix A with some entries as cos() and sin() then then may be we will have a polyhedron. (b) Given: The set of all x satisfying the constraint x2 8x + 15 0. No. (c) Given: The empty set. Yes. Observe that if A is nite and vector b is innite then the empty set will satisfy Ax = b.
0-5

Linear Programming Homework

Uploaded by

Linear Programming Homework

Uploaded by

Homework 0 2

Solution 2 Variables: pi , Mi for i {X, Y, Z} Constraints: p i Mi q 1 + q 2

Objective Function: min

= max {fi (x)} + (1 ) max {fi (y)} ( property of max function)

satisfying the constraints

You might also like