> For constrained optimisation ideally you should be using Lagrange multipliers.
Lagrange multipliers is for ̶u̶n̶c̶o̶n̶s̶t̶r̶a̶i̶n̶e̶d̶ optimization... you're probably thinking of KKT.
EDIT: D'oh... typed too quickly, thanks for the correction. I meant to write they can't handle inequalities, which for some reason is what I always assume when I hear constrained optimization...