Theory of functional connections applied to nonlinear programming under equality constraints
Academic Article

Overview

Research

View All

Overview

abstract

This paper introduces an efficient approach to solve quadratic programming problems subject to equality constraints via the Theory of Functional Connections. This is done without using the traditional Lagrange multipliers approach, and the solution is provided in closed-form. Two distinct constrained expressions (satisfying the equality constraints) are introduced. The unknown vector optimization variable is then the free vector $B{g}$, introduced by the Theory of Functional Connections, to derive constrained expressions. The solution to the general nonlinear programming problem is obtained by the Newton's method in optimization, and each iteration involves the second-order Taylor approximation, starting from an initial vector $B{x}^{(0)}$ which is a solution of the equality constraint. To solve the quadratic programming problems, we not only introduce the new approach but also provide a numerical accuracy and speed comparisons with respect to MATLAB's verb"quadprog". To handle the nonlinear programming problem using the Theory of Functional Connections, convergence analysis of the proposed approach is provided.