Let \mathcal{G} = \left\{f_{1}\left(x\right), f_{0}\left(x\right) \mid x \in \mathcal{D}\right\} be the set of achievable constraint/objective pairs. The Lagrange Dual Function g\left(\lambda\right) = \text{inf}_{t,u \in G} \left(t + \lambda u\right). We will push this line all the way down, and then push up its slope as much as possible.