A review of least-squares methods for solving partial differential equations



Continuous (integral) and discrete (point-matching) least-squares methods are presented for linear and non-linear problems in boundary-value, eigenvalue, and initial-value form. The history is traced, and important theoretical and practical results are summarized. A comprehensive sample of the literature is presented, indexed to show type of application, version of least squares used, and results of comparison studies. The advantages of least-squares methods are discussed, including convenience in formulation and error evaluation, generality of mixed and local (finite element) versions, and performance that is competitive with other methods.