Groupement De Recherche Mascot-Num

Software

This page provides a collection of links to software of interest for the MASCOT-NUM community—and hopefully beyond—on the topics of computer experiments, uncertainty and sensitivity analysis, Gaussian processes, polynomial chaos expansions, … Some (but not all) of these have been developped by, are maintained by, or include contributions from members of GdR MASCOT-NUM.

Table of Contents

R packages

  • Most of these packages can be installed using CRAN facilities (ie. install.packages).
  • The Sensitivity project for the sensitivity package - a collection of tools to compute sensitivities
  • CRAN Task View: Design of Experiments & Analysis of Experimental Data link
  • CompModSA - Sensitivity Analysis for Complex Computer Models. This package is useful for conducting sensitivity analysis of complex computer codes when model evaluations are somewhat expensive (e.g. take longer than a couple seconds to run) but a reasonable number (50 or more) of model evaluations can be obtained at sampled input values.
  • Authors' page (which provides an old version)
  • Unofficial updated version
  • lhs - for various Latin Hypercube Samples
  • BACCO - Bayesian analysis of computer code software
  • tgp - Treed Gaussian processes
  • DiceDesign, DiceEval, DiceKriging, DiceOptim - the four metamodeling packages of the Dice Consortium.
  • multisensi and modelcf - Sensitivity analysis and metamodeling for multidimensional and functional outputs
  • SMURFER: Sensitivity, Metamodel, Uncertainty and Reliability FEatured by R - sources - Documentation in French

Matlab/Octave

  • DACE - Design and Analysis of Computer Experiments. A matlab kriging toolbox.
  • FERUM - Finite Element Reliability Using Matlab. General-purpose structural reliability code.
  • GPML - Gaussian Processes for Machine Learning.
  • SA interface - for sensitivity analysis
  • STK - Small (Matlab/GNU Octave) Toolbox for Kriging
  • scalaGAUSS - Matlab kriging toolbox with a focus on large datasets
  • Sensitivity analysis routines - from Joint Research Centre

Scilab

  • DACE-Scilab - Scilab port of the DACE kriging matlab toolbox
  • krigeage - Kriging toolbox for Scilab
  • KRISP - Kriging based regression and optimization package for Scilab
  • NISP (Non Intrusive Spectral Projection) - Scilab toolbox to perform sensitivity analysis, based on polynomial chaos decomposition (CEA, Digiteo)

Python

  • scikit-learn - machine learning in Python
  • OpenTURNS - Open source initiative to Treat Uncertainties, Risks’N Statistics (Python/C++)

Others

  • A partial list proposed by B. Iooss and N. Devictor (CEA, 2008)
  • Promethee project (IRSN): a distributed computing engine designed for parametric calculation with heavy computing software. Coupled with R as algorithm provider (uncertainties propagations, sensitivity analysis, optimizations) for design of experiments and statistical analysis.
  • URANIE platform (CEA) in ROOT
  • DAKOTA (Design Analysis Kit for Optimization and Terascale Applications) - A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis
  • Neuro Pex : a software dedicated to the design of experiments for neural networks, algebraic nonlinear models, ordinary differential equations and some computer codes. Neuro Pex calculates D-optimal design, X-optimal design (Vila & Gauchi), true D-efficiency (Torsney) and parameter curvatures (Bates & Watts).
CNRS