Tensor Methods of Full-Information Maximum Likelihood Estimation: Estimation with Parameter Constraints.

In this study, we take a method presented in an earlier paper, called a tensor method, and apply it to the computation of constrained FIML estimates. This technique is based upon a fourth order approximation to the log-likelihood function, rather than the second order approximation used in standard methods. The higher order terms are low rank third and fourth order tensors that are computed, at very little storage or computation cost, using information from previous iterations. We discuss interior and exterior point methods for constrained optimization, show how they can be used in conjunction with tensor methods, and then present test results showing that the tensor method is far more efficient than the standard Newton's method for a wide range of constrained FIML estimation problems. Citation Copyright 1995 by Kluwer Academic Publishers.