Exercise 4.35 (Separation of disjoint polyhedra)

Consider two noempty polyhedra P = { x n Ax b } and Q = { x n Dx d } . We are interested in finding out whether the two polyhedra have a point in common.

(a) Devise a linear programming problem such that: if P Q is nonempty, it returns a point in P Q ; if P Q is empty, the linear programming problem is infeasible.

(b) Suppose that P Q is empty. Use the dual of the problem you have constructed in part (a) to show that there exists a vector c such that c x < c y for all x P and y Q .

Answers

(a)

max 0 x s.t. Ax b Dx d x  is free .

(b) The dual (D) is given by:

min p b + q d s.t. p A + q D = 0 p 0 , q 0 .

If P Q = , then either (D) is infeasible or unbounded. Since p = 0 and q = 0 produces a feasible solution, then (D) is unbounded.

Morevoer, assuming p = 0 and q = 0 and since (D) is unbouded, then p b + q d < 0 . We can describe P and Q as follows:

Ax b p Ax p b , x 0 Dy d q Dy q d , y 0 .

The constraints in (D) give us that q D = p A . Thus, we can use it to obtain p Ay q d , y 0 . Next, we can sum up both inequalities:

p Ax p Ay = p A ( x y ) p b + q d < 0 .

Let c = p A , then it follows that c x < c y for all x P and y Q .

User profile picture
2024-10-11 14:08
Comments