This paper introduces two R packages available on the Comprehensive R Archive network. The main application concerns the study of computer code output. Package DiceDesign is dedicated to numerical design of experiments, from the construction to the study of the design properties. Package DiceEval deals with the fit, the validation and the comparison of metamodels.After a brief presentation of the context, we focus on the architecture of these two packages. A two-dimensional test function will be a running example to illustrate the main functionalities of these packages and an industrial case study in five dimensions will also be detailed.

A kinetic model for the dehydration of lithium sulfate monohydrate is proposed in order to account for experimental data obtained on single crystals by thermogravimetry at 80 1C under fixed water vapour pressure, and by optical microscopy. This model is based on the assumptions of Mampel's model, the nucleation takes place randomly at the surface of the solid and is followed by isotropic growth toward the centre of the crystal. Calculated rates da/dt are obtained by means of Monte-Carlo simulations and compared to the experimental ones, which leads to the determination of two kinetic constants: the areic frequency of nucleation (in number of nuclei m À2 s À1 ) and the areic reactivity of growth (in mol m À2 s À1 ).

We consider the optimization of a computer model where each simulation either fails or returns a valid output performance. We first propose a new joint Gaussian process model for classification of the inputs (computation failure or success) and for regression of the performance function. We provide results that allow for a computationally efficient maximum likelihood estimation of the covariance parameters, with a stochastic approximation of the likelihood gradient. We then extend the classical improvement criterion to our setting of joint classification and regression. We provide an efficient computation procedure for the extended criterion and its gradient. We prove the almost sure convergence of the global optimization algorithm following from this extended criterion. We also study the practical performances of this algorithm, both on simulated data and on a real computer model in the context of automotive fan design.

International audienceA microscopic model based on the appearance, diffusion, and aggregation of point defects allows to calculate the time of appearance of the first nucleus on a surface during a reaction between a solid and a gas. Calculated distributions of these times of appearance of the first nucleus are qualitatively compared to experimental ones, previously determined. The appearance of the point defects seems to be the most influential step on the time of appearance of the first nucleus. Moreover the comparison between experimental and calculated distributions allows to conclude that the frequency of appearance of the defects is higher on the edges than on the faces of the single crystal

No abstract

No abstract

Kriging‐based exploration strategies often rely on a single Kriging model whose parametric covariance kernel is selected a priori or on the basis of an initial data set. As choosing an unadapted kernel can radically harm the results, we wish to reduce the risk of model misspecification. Here, we consider the simultaneous use of multiple kernels within Kriging. We give the equations of discrete mixtures of Kriging and derive a multikernel version of the expected improvement optimization criterion. We finally provide an illustration of the efficient global optimization algorithm with mixed exponential and Gaussian kernels, where the parameters are estimated by maximum likelihood and the mixing weights are likelihood ratios. Copyright © 2008 John Wiley & Sons, Ltd.

This paper proposes a new methodology to model uncertainties associated with functional random variables. This methodology allows to deal simultaneously with several dependent functional variables and to address the specific case where these variables are linked to a vectorial variable, called covariate. In this case, the proposed uncertainty modelling methodology has two objectives: to retain both the most important features of the functional variables and their features which are the most correlated to the covariate. This methodology is composed of two steps. First, the functional variables are decomposed on a functional basis. To deal simultaneously with several dependent functional variables, a Simultaneous Partial Least Squares algorithm is proposed to estimate this basis. Second, the joint probability density function of the coefficients selected in the decomposition is modelled by a Gaussian mixture model. A new sparse method based on a Lasso penalization algorithm is proposed to estimate the Gaussian mixture model parameters and reduce their number. Several criteria are introduced to assess the methodology performance: its ability to approximate the functional variables probability distribution, their dependence structure and their features which explain the covariate. Finally, the whole methodology is applied on a simulated example and on a nuclear reliability test case.