Lagrange Multipliers - shivamvats/notes GitHub Wiki

Statement: Let f be a function to be maximized, under m equality constraints: g1,...,gm. Then, the Langrange function is defined to be L(f, k) = f - k(g). Stationary points of this equation correspond to maxima of the f such that they satisfy the constraints as well.

Intuition: The method of Lagrange multipliers is essentially hill climbing. The gradient of f: grad(f) points along the direction of maximum increase in f