Understanding the Lagrangian in Constrained Optimization

Abstract

A new approach to teaching constrained optimization is proposed, where the Lagrange multipliers in the Lagrangian are viewed as undetermined coefficients, and the Lagrangian are viewed as a weighted sum of multiple objective functions. This perspective facilitates a more intuitive and natural derivation of the Karush-Kuhn-Tucker (KKT) conditions, including the stationary and complementary slackness conditions, assisting students in appreciating their significance in determining the minimizer. This approach helps students understand how the KKT conditions come into being, allowing them to better grasp and master the optimality conditions, ultimately enabling them to apply the theoretical knowledge to real-world problems.