Abstract
A method is described for determining an optimal straight-line segment approximation to specified functions for constrained and unconstrained endpoints.