Cluster Gauss Newton method

A computationally efficient algorithm to find multiple solutions of nonlinear least squares problems.

Standard methods such as the Levenberg-Marquardt method can find a solution of a nonlinear least squares problem that does not have a unique solution. However, the parameter found by the algorithm depends on the choice of the initial iterate. To reduce the analysis bias due to the initial iterate used for the algorithm, it is a good practice to repeatedly use the algorithm with various initial iterates to gain the understanding of the influence of the initial iterates. On the other hand, due to the computational challenges and cost, it is rarely done in practice. Cluster Gauss-Newton method is made to remedy this issue by efficiently finding multiple solutions. (please read https://arxiv.org/abs/1808.06714 for detail)