Quant GT
Browse all lessons
Section 18 · Lesson 18.3

Convex Optimization

Why convex problems are tractable and most others aren't.

A function ff is convex if f(λx+(1λ)y)λf(x)+(1λ)f(y)f(\lambda x + (1-\lambda) y) \le \lambda f(x) + (1-\lambda) f(y) for all x,yx, y and λ[0,1]\lambda \in [0, 1]. A set is convex if it contains every line segment between its points.

Convex optimization minimizes a convex function over a convex set. The defining property: any local minimum is a global minimum. There are no spurious basins to get stuck in.

Convex problems are solvable by dependable polynomial-time algorithms (interior-point methods, gradient-based methods with global guarantees). Linear programs, quadratic programs, semidefinite programs, second-order cone programs are all convex.

Recognizing convexity matters: x2\|x\|^2 is convex; logx\log x is concave (so logx-\log x is convex); sums and pointwise maxima of convex functions are convex; affine compositions preserve convexity. In finance, mean-variance portfolio optimization is convex; option pricing via static replication is a linear program.