The KrigInv package: An efficient and user-friendly R implementation of Kriging-based inversion algorithms
Résumé
Several strategies relying on kriging have recently been proposed for adaptively estimating contour lines and excursion sets of functions when the evaluation budget is severely limited. Here we present the newly released R package KrigInv, offering a sound implementation of most sampling criteria for those kinds of inverse problems. KrigInv bases on the DiceKriging package, and thus benefits from a number of options concerning the underlying kriging models. In this tutorial, the seven implemented sampling criteria are presented and illustrated with graphical examples, and the different functionalities of KrigInv are gradually explained. Additionally, two recently proposed criteria for batch-sequential inversion are presented, enabling advanced users to distribute function evaluations in parallel on clusters or clouds of machines. We finally discuss about the fine tuning of numerical integration and optimization procedures used within the calculation and/or the optimization of the considered criteria.
Origine | Fichiers éditeurs autorisés sur une archive ouverte |
---|