Optimal Designs for Fourier Regession Models
classical Q-optimal designs and minimax designs under A-, D-, and Q-optimal criteria. We have proved the equivalency of Q-optimal and A-optimal designs. It is shown that A- and D-optimal designs are not sensitive to the misspecified order of model assumed. D-optimal designs for estimating pairs of coefficients are also constructed in this paper. Furthermore, D-optimal designs for a commonly-used reduced Fourier model have been derived. The comparison with other competitors show that our designs incorporating the reduced models are much more efficient.
D-optimality A-optimality Q-optimality Trigonometric regression Symmetric designs Minimax designs Sensitivity
XIAOJIAN XU XIAOLI SHANG
DEPARTMENT OF MATHEMATICS, BROCK UNIVERSITY, ST. ATHARINES, ON L2S 3A1, CANADA DEPARTMENT OF MATHEMATICS, BROCK UNIVERSITY, ST. CATHARINES, ON L2S 3A1, CANADA In this article, the
国际会议
北京
英文
889-895
2011-06-20（万方平台首次上网日期，不代表论文的发表时间）