Cases for the nugget in modeling computer experiments
- 515 Downloads
Most surrogate models for computer experiments are interpolators, and the most common interpolator is a Gaussian process (GP) that deliberately omits a small-scale (measurement) error term called the nugget. The explanation is that computer experiments are, by definition, “deterministic”, and so there is no measurement error. We think this is too narrow a focus for a computer experiment and a statistically inefficient way to model them. We show that estimating a (non-zero) nugget can lead to surrogate models with better statistical properties, such as predictive accuracy and coverage, in a variety of common situations.
KeywordsComputer simulator Surrogate model Gaussian process Interpolation Smoothing
Unable to display preview. Download preview PDF.
- Gramacy, R.B.: Bayesian treed Gaussian process models. Ph.D. thesis, University of California, Santa Cruz (2005) Google Scholar
- Gramacy, R.B.: tgp: An R package for Bayesian nonstationary, semiparametric nonlinear regression and design by treed Gaussian process models. J. Stat. Softw. 19, 9 (2007) Google Scholar
- Neal, R.M.: Monte Carlo implementation of Gaussian process models for Bayesian regression and classification. Tech. Rep. 9702, Department of Statistics, University of Toronto (1997) Google Scholar
- O’Hagan, A., Kennedy, M.C., Oakley, J.E.: Uncertainty analysis and other inference tools for complex computer codes. In: Bernardo, J.M., Berger, J.O., Dawid, A., Smith, A. (eds.) Bayesian Statistics 6, pp. 503–524. Oxford University Press, Oxford (1999) Google Scholar
- Ranjan, P., Haynes, R., Karsten, R.: Gaussian process models and interpolators for deterministic computer simulators. Department of Mathematics and Statistics, Acadia University (2010) Google Scholar
- Rogers, S.E., Aftosmis, M.J., Pandya, S.A., Chaderjian, N.M., Tejnil, E., Ahmad, J.U.: Automated CFD parameter studies on distributed parallel computers. In: 16th AIAA Computational Fluid Dynamics Conference (2003). AIAA Paper 2003-4229 Google Scholar
- Taddy, M., Lee, H.K.H., Gray, G.A., Griffin, J.D.: Bayesian guided pattern search for robust local optimization. Tech. Rep. ams2008-02, University of California, Santa Cruz, Department of Applied Mathematics and Statistics (2008) Google Scholar