## R Packages

I am author, co-author or contributor of the following R packages:

- KrigInv (Kriging Based Inversion): joint work with Victor Picheny and David Ginsbourger.
- DiceView (Plot methods for computer experiments design and surrogate): joint work with Yann Richet and Yves Deville.

- Contributor of the DiceKriging package (Kriging methods for computer experiments): joint work with Olivier Roustant, David Ginsbourger and Yves Deville.
- Contributor of the DiceOptim package (Kriging-based optimization for computer experiments): joint work with David Ginsbourger, Victor Picheny and Olivier Roustant.

### Modeling non-stationary extreme dependence with stationary max-stable processes and multidimensional scaling

This is a joint work with Olivia Martius and David Ginsbourger.

We here publicly share some R code to reproduce the results of the article. The associated article is available on arXiv; and a description is available in the Supplementary material.

The following zip files can be downloaded:

- Extremal coefficient maps (.7z file) or Extremal coefficient maps (.zip file): All 219 extremal coefficient maps. The folder contains 219 .png files.
- Public R code (.7z file) or Public R code (.zip file): This code enables to reproduce all the figures and results of the article 'Modeling non-stationary extreme dependence with stationary max-stable processes and multidimensional scaling', as well as the results in the Supplementary Material.

### Fast computation of best linear predictors when the dataset is large

I am co-author of a recent algorithm for computing approximated best linear predictors in the case where the number, n, of observations is large.

This is a joint work with Nicolas Durrande, Didier Rullière and François Bachoc. The implemented algorithms can be used when n is in the interval [10^5 , 10^6 ], or lower. The code is meant to be run easily in R. It requires the Rcpp and RcppArmadillo package.

The associated article is available here.

The following files can be downloaded and run in R:

- readme.txt : Help file to read before running the R and Rcpp code.
- CodeCpp_fastpredict_functions.cpp : C++ implementation of our fast best linear predictor -- useful when the number of observations n is large, i.e. larger than 10^5
- CodeR_gradientoptim_functions.R : A tailored algorithm to estimate the parameters of the covariance function.
- ToyExamples_LargeDataset.R : Read the instructions and run the code in this file to use the implemed algorithms. This code performs many calls to the functions implemented in 'CodeCpp_fastpredict_functions.cpp' and 'CodeR_gradientoptim_functions.R'.