论文信息 - Selective Random Search for Optimal Experiment Designs

Selective Random Search for Optimal Experiment Designs

The most time consuming task in numerical search for optimum experiment design is in finding (by multidimensional global maximization) candidate points to be entered to a current design. To overcome this difficulty it is proposed to use a selective random search. The key idea is to use the probability distribution proportional to the prediction variance of the present design as the density of random search. In each iteration of the proposed algorithm additional optimization of the current design weights is performed.

Ewaryst Rafajłowicz | E. Rafajłowicz

[1] E. Ziegel. Optimal design and analysis of experiments , 1990 .

[2] L. Devroye. Non-Uniform Random Variate Generation , 1986 .

[3] H. Wynn. The Sequential Generation of $D$-Optimum Experimental Designs , 1970 .

[4] Andrej Pázman,et al. Foundations of Optimum Experimental Design , 1986 .

[5] S. Silvey,et al. An algorithm for optimal designs on a design space , 1978 .

[6] W. J. Studden,et al. Theory Of Optimal Experiments , 1972 .

[7] Anthony C. Atkinson,et al. Optimum Experimental Designs , 1992 .

[8] Ben Torsney. Algorithms for a constrained optimization problem with applications in statistics and optimum design , 1981 .

[9] Johan Fellman,et al. An empirical study of a class of iterative searches for optimal designs , 1989 .

[10] B. Torsney. A Moment Inequality and Monotonicity of an Algorithm , 1983 .

[11] H. Wynn,et al. The Convergence of General Step-Length Algorithms for Regular Optimum Design Criteria , 1978 .

[12] Corwin L. Atwood,et al. Convergent Design Sequences, for Sufficiently Regular Optimality Criteria , 1976 .