Heuristic optimization of experimental designs

Abstract We propose to integrate different algorithms for constructing D -optimum designs for linear models. Our emphasis is on efficiency gain and on applicability to larger models than those currently considered in the literature. We implement a one-exchange algorithm and use a generalized simulated annealing. This method does not require to construct or to enumerate each point of the candidate set, whose size grows exponentially with the number of variables. In order to handle more complex problems, we develop a procedure generating guided starting designs. A comparison of our results with those found in the literature shows that the simultaneous integration of these algorithms turns out to be very effective. As compared to results from the literature, our algorithmic process allows an increase in efficiency while, for larger models (up to 20 parameters), we attain a 90% D -efficiency level.

[1]  O. Dykstra.  The Augmentation of Experimental Data to Maximize [X′X] , 1971 .

[2]  R. K. Meyer,et al.  The Coordinate-Exchange Algorithm for Constructing Exact Optimal Experimental Designs , 1995 .

[3]  Cecilia R. Aragon,et al.  Optimization by Simulated Annealing: An Experimental Evaluation; Part I, Graph Partitioning , 1989, Oper. Res..

[4]  R. C. St.  D-Optimality for Regression Designs: A Review , 1975 .

[5]  M. J. Box,et al.  Factorial Designs, the |X′X| Criterion, and Some Related Matters , 1971 .

[6]  Miguel A. Lejeune.  A “Coordinate-Columnwise” Exchange Algorithm for the Construction of Supersaturated, Saturated, and Non-Saturated Experimental Designs , 2003 .

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

[8]  M. E. Johnson,et al.  Generalized simulated annealing for function optimization , 1986 .

[9]  Toby J. Mitchell,et al.  Computer Construction of “D-Optimal” First-Order Designs , 1974 .

[10]  M. E. Johnson,et al.  Some Guidelines for Constructing Exact D-Optimal Designs on Convex Design Spaces , 1983 .

[11]  J. Kiefer,et al.  Time- and Space-Saving Computer Methods, Related to Mitchell's DETMAX, for Finding D-Optimum Designs , 1980 .

[12]  P. Laycock,et al.  Optimum Experimental Designs , 1995 .

[13]  Toby J. Mitchell,et al.  An algorithm for the construction of “ D -optimal” experimental designs , 2000 .

[14]  L. Haines.  The application of the annealing algorithm to the construction of exact optimal designs for linear-regression models , 1987 .

[15]  R. K. Meyer,et al.  Constructing exact D-optimal experimental designs by simulated annealing , 1988 .

[16]  R. C. St. John,et al.  D-Optimality for Regression Designs: A Review , 1975 .

[17]  Van Schalkwyk,et al.  On the design of mixture experiments , 1971 .

[18]  Ehl Emile Aarts,et al.  Simulated annealing and Boltzmann machines , 2003 .