Optimal Discriminating Designs


This package provides several functions suitable for efficient numerical construction of optimal discriminative designs.


At the current state this package provides the routine tpopt for the construction of T_P-optimal designs, the routine KLopt.lnorm for the calculation of KL-optimal designs (for lognormal errors) and several auxiliary procedures to represent the results. Function tpopt is based on the algorithms that were developed in [7]. Function KLopt.lnorm is based on the methodology proposed in [8]. See the references for more details.

It is planned to add several new routines for different types of discriminative designs.


[1] Atkinson A.C., Fedorov V.V. (1975) The design of experiments for discriminating between two rival models. Biometrika, vol. 62(1), pp. 57–70.

[2] Atkinson A.C., Fedorov V.V. (1975) Optimal design: Experiments for discriminating between several models. Biometrika, vol. 62(2), pp. 289–303.

[3] Dette H., Pepelyshev A. (2008) Efficient experimental designs for sigmoidal growth models. Journal of statistical planning and inference, vol. 138, pp. 2–17.

[4] Dette H., Melas V.B., Shpilev P. (2013) Robust T-optimal discriminating designs. Annals of Statistics, vol. 41(4), pp. 1693–1715.

[5] Braess D., Dette H. (2013) Optimal discriminating designs for several competing regression models. Annals of Statistics, vol. 41(2), pp. 897–922.

[6] Braess D., Dette H. (2013) Supplement to “Optimal discriminating designs for several competing regression models”. Annals of Statistics, online supplementary material.

[7] Dette H., Melas V.B., Guchenko R. (2014) Bayesian T-optimal discriminating designs. ArXiv link.

[8] Dette H., Guchenko R., Melas V.B. (2015) Efficient computation of Bayesian optimal discriminating designs. ArXiv link.

comments powered by Disqus