Local and Global Optima in Convex Optimization
Theorem: Global vs. local optima in convex optimization
For convex problems, any locally optimal point is globally optimal. In addition, the optimal set is convex.
Proof: Let be a local minimizer of
on the set
, and let
. By definition,
bf dom
. We need to prove that
. There is nothing to prove if
, so let us assume that
bf dom
. By convexity of
and
, we have
, and:
Since is a local minimizer, the left-hand side in this inequality is nonnegative for all small enough values of
. We conclude that the right hand side is nonnegative, i.e.,
, as claimed.
Also, the optimal set is convex, since it can be written
This ends our proof.