Estimates of the coverage of parameter space by Latin Hypercube and Orthogonal sampling: connections between Populations of Models and Experimental Designs

Autor:	Donovan, Diane, Burrage, Kevin, Burrage, Pamela, McCourt, Thomas A, Thompson, Harold Bevan, Yazici, Emine Sule
Rok vydání:	2015
Předmět:	Mathematics - Statistics Theory 62K25
Druh dokumentu:	Working Paper
Popis:	In this paper we use counting arguments to prove that the expected percentage coverage of a $d$ dimensional parameter space of size $n$ when performing $k$ trials with either Latin Hypercube sampling or Orthogonal sampling (when $n=p^d$) is the same. We then extend these results to an experimental design setting by projecting onto a 2 dimensional subspace. In this case the coverage is equivalent to the Orthogonal sampling setting when the dimension of the parameter space is two. These results are confirmed by simulations. The ideas presented here have particular relevance when attempting to perform uncertainty quantification or when building populations of models. Comment: 15 pages, 2 figures. arXiv admin note: text overlap with arXiv:1502.06559
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/1510.03502 Zobrazit plný text záznamu View this record from Arxiv