Exercice 1. Let A be an m×n matrix, let b ∈ Rm , and c ∈ Rn . Solve the optimization
problem: minimize cT x subject to the constraints Ax = b.
Exercice 2. Show that the following sets are convex:
a) { ni=1 λi xi : λ ≥ 0, ni=1 λi = 1} where x1 , . . . , xn are given in Rn . P P b) {x ∈ Rn : ∥x − x0 ∥ ≤ r} where r > 0, x0 ∈ Rn , and ∥ · ∥ is a norm on Rn . c) {x ∈ Rn : aT x = b} and {x ∈ Rn : aT x ≤ b} where a ∈ Rn , b ∈ R. d) {x ∈ Rn : Ax = b, Cx ≤ d}.
Exercice 3. Show that if a nonempty set is an intersection of convex sets then it is
convex. Deduce that the set of semidefinite positive matrices is convex.
Exercice 4. Show that the set of optimal solutions of a convex optimization problem is convex.
Exercice 5. a) Show that if X is convex its interior is convex too.
b) We define the ε-enlargement of set X by X ε = {y : dist(y, X) ≤ ε} for ε > 0 where dist is the distance function:
inf ∥y − x∥ dist(y, X) = x ∈ X. Show that if X is convex then the set X ε is convex too. c) Let BF (0, ε) = {x : ∥x∥ ≤ ε}. Show that if X is convex and closed then X ε = X + BF (0, ε). If X is closed and convex, deduce another proof of the convexity of X ε .
Exercice 6. Let C be a convex set and let α1 , α2 > 0. Show that
(α1 + α2 )C = α1 C + α2 C. Find a set C for which the above relation does not hold.
Exercice 7. Let S, T be two nonempty convex sets in Rn . Show that a separates S, T
if and only if sup aT x ≤ inf aT y x∈S y∈T and inf aT x < sup aT y. x∈S y∈T Show that the separation is strong if and only if sup aT x < inf aT y. x∈S y∈T