Skip to content

Optimization for Data Science

Theory of Convex Optimization

First-order Characterization of Convexity

We have a lemma to prove a function is convex using first-order gradient.
Suppose \(D(f)\) is open and that \(f\) is differentiable; in particular the gradient

\[ \nabla f(x) = \left( \frac{\partial f}{\partial x_1}, \dots, \frac{\partial f}{\partial x_d} \right) \]

exists at every point \(x \in D(f)\).

Then \(f\) is convex iff \(D(f)\) is convex and

\[ f(y) \ge f(x) + \nabla f(x)^T (y - x) \]

holds for all \(x, y \in D(F)\).

Comments