On Constrained Programming Problems with Singular Designs via Super Convergent Line Series

The study extends the Super Convergent Line Series Algorithm to a case where the determinant information matrix of the design is zero by employing the Moore-Penrose inverse approach. The algorithm is tested using a numerical example on a constrained programming problem. The optimal solution obtained by the algorithm compares favorably with the one obtained by an existing Frank-Wolfe method and the value of the optimizer satisfies the given constraint equation.