论文信息 - The Derivative of |X′X| and Its Uses

The Derivative of |X′X| and Its Uses

The determinant of matrices of the form X′X is used as the D-optimal design criterion for parameter estimation in nonlinear regression and as the estimation criterion for multiresponse parameter estimation. It is helpful to be able to calculate the gradient of this determinant so that sophisticated optimization methods can be used. A method is given to efficiently compute the determinant using modern computational linear algebra and, at the same time, to provide the necessary information to compute the gradient. The method is extended to cases in which a part of the X matrix is fixed. Examples from multiresponse parameter estimation and from experimental design are given.

D. Bates

[1] H. L. Lucas,et al. DESIGN OF EXPERIMENTS IN NON-LINEAR SITUATIONS , 1959 .

[2] George E. P. Box,et al. The Bayesian estimation of common parameters from several responses , 1965 .

[3] G. Golub,et al. Numerical computations for univariate linear models , 1973 .

[4] John M. Chambers,et al. Computational Methods for Data Analysis. , 1978 .

[5] T. Brubaker,et al. Nonlinear Parameter Estimation , 1979 .

[6] David W. Bacon,et al. Singularities in Multiresponse Modelling , 1979 .

[7] R. L. Crane,et al. Solution of the General Nonlinear Programming Problem with Subroutine VMCOM , 1980 .

[8] Peter D. H. Hill. D-Optimal Designs for Partially Nonlinear Regression Models , 1980 .

[9] Eric R. Ziegel,et al. Kinetic Modelling With Multiresponse Data , 1980 .

[10] V. Klema. LINPACK user's guide , 1980 .

[11] Warren E. Stewart,et al. Bayesian Estimation of Common Parameters From Multiresponse Data With Missing Observations , 1981 .

[12] Douglas M. Hawkins,et al. The Cholesky factorization of the inverse correlation or covariance matrix in multiple regression , 1982 .