Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
min −x1 − x2
s.t. x1 ≥ 0
x2 ≥ 0
x21 + 21 x22 ≤ 1.
1
Now 2λ3 x1 = 1 = λ3 x2 , and we know that λ3 %= 0, so 2x1 = x2 .
Therefore x21 + 2x21 = 1, which (coupled with x1 ≥ 0) implies that
1
x1 = √ .
3
Back substitution then gives us
√
2 3
x2 = √ , λ3 = .
3 2
(d) The only active constraint at the KKT point is the third constraint.
Therefore the only active gradient is
! "
2x1
∇g3 (x) = .
x2
2
(a)
% &
α 1
Pα (x) = −x1 − x2 + (−x1 )2+ + (−x2 )2+ + (x21 + x22 − 1)2+ .
2 2
(b)
−1 + 12 x1 (x21 + 21 x22 − 1)
! "
∇P1/4 (x) = .
−1 + 14 x2 (x21 + 21 x22 − 1)
3. [5 marks] Show that any point on a line drawn between any two
feasible points in a convex program is itself feasible. (In other
words, that the feasible region is a convex set).
If the points are x and y, then any point on the line between then can be
expressed as αx + (1 − α)y for some α ∈ [0, 1]. Since gi (x) is convex,
since α ∈ [0, 1] and both x and y are feasible. Furthermore, since h(x) is
affine,
h(αx + (1 − α)y) = αh(x) + (1 − α)h(y) = 0.
Therefore any point on the line is feasible.