Projet SAMOURAI (ANR-20-CE46-0013)

Summary of the project
Metamodeling and optimization under uncertainty for complex simulators
Towards frugal simulation methods for robust design applicationsIncreasing the efficiency of simulation-based industrial processes requires improvements in uncertainty quantification and numerical optimization. A major difficulty arises from the “black box” nature of the simulator, a direct consequence of the growing complexity of the applications targeted. The main objectives of the project, driven by application needs, are to develop efficient optimization methods based on metamodels, pushing back their current limits of performance and applicability. The ambition is to meet the following four challenges:
(i) Work package 1: design metamodels adapted to large-scale problems (typically around 100 input variables) in the context of a limited budget of simulations (around 500);
(ii) Work package 2: adapt sequential enrichment strategies to large-scale problems in the context of optimal and reliable design ;
(iii) Work package 3: develop metamodeling and optimization methods involving non-continuous input variables such as discrete or categorical variables;
(iv) Work package 4: take into account the presence of simulator failures by learning the associated hidden constraints and integrating them into the optimization procedure.
Variable selection, Gaussian process models and active learning adapted to expensive simulators for optimization under uncertainty
To build efficiently predictive metamodels from a limited set of simulations, the study of new sensitivity indices to identify influential variables, metamodels adapted to high dimensions and a robust optimization method for their hyperparameters were proposed. The second objective of the project was to identify the set of controlled input values of a simulator enabling to reach a target risk level for an output of interest also dependent on uncertain variables, using an adapted enrichment criterion. Problems with a low probability of failure were solved by a sequence of problems associated with a progressively reduced failure region. For the third objective, metamodels with a kernel adapted to sets of vectors were developed and coupled to a Bayesian optimization method with an enrichment criterion optimized by a specific evolutionary algorithm. Finally, an active learning method was developed to predict the probability of simulation failure, and coupling strategies were used to take these hidden constraints into account in “black box” optimization methods.
The project has resulted in a dozen of papers and around 30 oral communications in international journals and conferences, as well as the sharing of code whose sources are available on public repositories (CRAN, GitHub or GitLab).
In addition, the results are being used for applications in the field of renewable and low-carbon energies and air transport to reduce CO2 emissions: in the reliable and optimal design of wind turbines and engine turbine blades, for the optimal layout of wind farms and the analysis of uncertainties in thermal-hydraulic models for nuclear safety.
The Samourai project is a collaborative research-enterprise project (PRCE) coordinated by IFPEN. It also involves CEA, CentraleSupélec, EDF R&D, EMSE, Safran Tech and Polytechnique Montréal.
Publications of the project
WP1Journal articles (published or preprint)
- Marrel, Iooss and Chabridon. The ICSCREAM methodology: Identification of penalizing configurations in computer experiments using screening and metamodel -- Applications in thermal-hydraulics. Nuclear Science and Engineering, 2022, 196, pp.301-321, DOI:10.1080/00295639.2021.1980362
- Marrel and Iooss. Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation. Reliability Engineering and System Safety, 2024a, 247, pp.110094, DOI:10.1016/j.ress.2024.110094
- Marrel and Iooss. Probabilistic surrogate modeling by Gaussian process: A new estimation algorithm for more robust prediction. Reliability Engineering and System Safety, 2024b, 247, pp.110120, DOI:10.1016/j.ress.2024.110120
- Carpintero Perez, Da Veiga, Garnier and Staber. Gaussian process regression with Sliced Wasserstein Weisfeiler-Lehman graph kernels. Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:1297-1305, hal-04440186
- Sarazin, Marrel, Da Veiga and Chabridon. New insights into the feature maps of Sobolev kernels: application in global sensitivity analysis. 2023. cea-04320711
- Carpintero Perez, Da Veiga, Garnier and Staber. Learning signals defined on graphs with optimal transport and Gaussian process regression. 2024, hal-04740924
- Sarazin, Marrel, Da Veiga and Chabridon. What is hidden behind the Sobolev kernels involved in the HSIC-ANOVA decomposition?. 2022 SAMO Conference - 10th International Conference on Sensitivity Analysis of Model Output, Florida State University, Mar 2022, Tallahassee, United States. cea-03701074
- Sarazin, Marrel, Da Veiga and Chabridon. Test d'indépendance basé sur les indices HSIC-ANOVA d'ordre total. 53èmes Journées de Statistique de la SFdS, Société Française de Statistique (SFdS); Université Claude Bernard Lyon 1, Jun 2022, Lyon, France. cea-03701170
- Marrel, Iooss and Chabridon. The ICSCREAM methodology: identification of penalizing configurations in computer experiments using screening and metamodel applications in thermal-hydraulics. SIAM Conference on Uncertainty Quantification (UQ22), Apr 2022, Atlanta, United States. cea-03700747
- Marrel and Iooss. New estimation algorithm for more reliable prediction in Gaussian process regression: application to an aquatic ecosystem model. ENBIS 23 - The 23th annual conference of the European Network for Business and Industrial Statistics, Sep 2023, Valence, Spain. cea-04216148
- Marrel and Iooss. Hidden But Essential Recipes for Successful Gaussian Process Metamodeling to Support Uncertainty Quantification in Numerical Simulation. SIAM Conference on Uncertainty Quantification (UQ24), Feb 2024, Trieste, Italy. cea-04506392
- Iooss and Marrel. Gaussian process regression: new hyperparameter estimation algorithm for more reliable prediction. SIAM Conference on Uncertainty Quantification (UQ24), Feb 2024, Trieste, Italy. cea-04506405
- Carpintero Perez, Da Veiga, Garnier and Staber. Regression par processus Gaussiens pour des entrées graphes en grande dimension. 53èmes Journées de Statistique de la SFdS, Société Française de Statistique (SFdS); Université Claude Bernard Lyon 1, Jun 2022, Lyon, France.
- Carpintero Perez, Da Veiga, Garnier and Staber. Gaussian Process Regression for High Dimensional Graph Inputs. SIAM Conference on Uncertainty Quantification (UQ24), Feb 2024, Trieste, Italy.
- Bayesian ACCOSSO has been made available to the SAMOURAI partners through a private git repository on the project GitLab's server
- Release package sensitivity on CRAN
- Sliced Wasserstein Weisfeiler Lehman Graph kernel : https://gitlab.com/drti/swwl
Journal articles (published or preprint)
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Vincent Chabridon, Emmanuel Vazquez. Bayesian sequential design of computer experiments for quantile set inversion. Technometrics, 2025, 67(1), pp.112-121. hal-03835704
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Emmanuel Vazquez. Bayesian Active Learning of (small) Quantile Sets through Expected Estimator Modification. In preparation, soon to be released as a preprint on HAL + arXiv.
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Emmanuel Vazquez. Sequential Bayesian inversion of black-box functions in presence of uncertainties. MASCOT-NUM 2022, Jun 2022, Clermont-Ferrand, France. Poster. hal-03694867
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Emmanuel Vazquez. Reliability-based inversion: Stepwise uncertainty reduction strategies? SIAM Conference on Uncertainty Quantification (UQ22), Apr 2022, Atlanta, United States. hal-03694921
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Vincent Chabridon, Emmanuel Vazquez. Estimation of (small) reliable sets using a sequential Bayesian strategy. MASCOT-NUM 2023, Apr 2023, Le Croisic, France. Poster. hal-04370899
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Emmanuel Vazquez. A stepwise uncertainty reduction strategy for the estimation of small quantile sets. MASCOT-NUM 2024, Apr 2024, Hyères, France. hal-04910458
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Vincent Chabridon, Emmanuel Vazquez. Estimation of Small Quantile Sets Using a Sequential Bayesian Strategy. SIAM Conference on Uncertainty Quantification (UQ24), Feb 2024, Trieste, Italy. hal-04501097
- Romain Ait Abdelmalek-Lomenech, Julien Bect, Vincent Chabridon, Emmanuel Vazquez. Active Learning of (small) Quantile Sets. SAMOURAI (workshop of the SAMOURAI ANR project), Dec 2024, Paris Institut Henri Poincaré, France. hal-04910443
- https://github.com/stk-kriging/contrib-qsi
- https://github.com/stk-kriging/qsi-paper-experiments
- The source code for Réf. [2.2] has been made available to the SAMOURAI partners through a private git repository on the project GitLab's server, and will soon be released on GitHub.
Journal articles (published or preprint)
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Wasserstein-Based Evolutionary Operators for Optimizing Sets of Points : Application to Wind-Farm Layout Design. Applied Sciences, 2024, 14 (17), pp.7916. DOI:10.3390/app14177916
- Babacar Sow. Optimisation et métamodélisation de fonctions définies sur des nuages de points. Thèse. Ecole des Mines de Saint-Etienne, novembre 2024. hal-04894626
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Sanaa Zannane, Merlin Keller. Learning functions defined over sets of vectors with kernel methods. Article in Proceedinfs of the 5th ECCOMAS Thematic Conference on Uncertainty Quantification in Computational Sciences and Engineering (UNCECOMP 2023), Jun 2023, Athène, Greece. emse-04043206
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Optimization and metamodeling of functions defined over clouds of points. workshop SAMOURAI, réseau thématique Uncertainty Quantification (RT UQ), Dec 2024, Paris, France. hal-04830176
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Optimization of functions defined over sets of points in polygons with evolutionary algorithms based on Wasserstein barycenters. FGS Conference on Optimization 2024: French German Spanish conference on optimization, Jun 2024, Gijon, Spain. emse-04645882
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Active learning for the optimization of functions defined over clouds of points. 55es Journées de Statistique de la SFdS (JdS 2024), May 2024, Bordeaux, France. hal-04645977
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Clouds of points optimization in convex polygons with evolutionary algorithms based on Wasserstein barycenters. Poster, MASCOT-NUM 2024, Apr 2024, Giens, France. hal-04645929
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Merlin Keller, Sanaa Zannane. Wasserstein Barycenter-based Evolutionary Algorithm for the optimization of sets of points. PGMO DAYS 2023, Nov 2023, Paris-Saclay, France. emse-04344769
- Babacar Sow, Rodolphe Le Riche, Sanaa Zannane, Merlin Keller, Julien Pelamatti. Cloud of points as discrete measures for Gaussian models and stochastic optimization. MASCOT-NUM 2023, Apr 2023, Le Croisic, France. hal-04102275
- Babacar Sow, Rodolphe Le Riche, Julien Pelamatti, Sanaa Zannane, Merlin Keller. Gaussian Processes Indexed by Clouds of Points : a study . MASCOT-NUM, Jun 2022, Clermont-Ferrand, France. emse-03720276
- Merlin Keller, Sebastien Le Digabel, Josephine Gobert, Antoine Lesage-Landry, Julien Pelamatti. Windfarm optimization with NOMAD. workshop SAMOURAI, réseau thématique Uncertainty Quantification (RT UQ), Dec 2024, Paris, France. Samourai workshop
WP4
Journal articles (published or preprint)
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Estimation of simulation failure set with active learning based on Gaussian process classifiers and random set theory. under review, submitted in Oct. 2024. hal-03848238
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Learning and space-filling a design space avoiding numerical failure of computer experiments. Pre-print to be submitted hal-04974850
- Le Digabel, S., Menz, M., Sinoquet, D., Tribes, C., Handling binary, unrelaxable and hidden constraints in blackbox optimization, in preparation.
- Jacquet, S., Sinoquet, D. Handling Hidden Constraints in Blackbox Optimization. SIAM CSE, Amsterdam, Netherland, february 26–march 3 , 2023.
- Jacquet, S., Munoz Zuniga, M., Sinoquet, D. Hidden constraints surrogates using gaussian processes in blackbox optimization, JOPT, HEC Montréal, Canada, may 29-31, 2023.
- Menz, M., Munoz Zuniga, M., Sinoquet, D., Learning hidden constraints with gaussian process classifiers in the optimization context, SIAM CSE, Amsterdam, Netherland, february 26–march 3, 2023.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Coupling learning of hidden constraints with surrogate optimization, SIAM Opt, Seattle, United States, may 31–june 3 , 2023.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Recovering hidden constrained subsets by a Gaussian Process Classifier active learning method based on the Stepwise Uncertainty Reduction strategy, UNCECOMP, Athens, Greece, june 12-14, 2023.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Hidden Crash Constraints Management for the Robust Design Optimization of Wind Turbines. SIAM UQ, Trieste, Italy, february 27–march 1, 2024.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Optimization under hidden constraints and its application to the robust design of wind turbines. ISMP, Montreal, Canada, july 21-26, 2024.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Estimation of simulation failure set with active learning based on Gaussian Process classifiers and random set theory. ANR SAMOURAI Workshop, IHP Paris, France December 10-11, 2024.
- Menz, M., Munoz Zuniga, M., Sinoquet, D. Dealing with hidden constraints in the DoE and optimization of computer experiments, UQSAY, online, March 21, 2024. (invited)
- Menz, M., Munoz Zuniga, M., Sinoquet, D., Black-box optimization with hidden constraints, OPTIMA (ARC Training Centre, Optimisation Technologies, Integrated Methodologies, and Applications), on-line, 2024, Australia (invited)
- Menz, M., Munoz Zuniga, M., Sinoquet, D., Black-box optimization with hidden constraints, Séminaire à la 44ème journée francilienne de recherche opérationnelle de ROADEF, 2024, Université Paris-Dauphine (invited)
- Menz, M., Munoz Zuniga, M., Sinoquet, D., Black-box optimization with hidden constraints, Séminaire dans le cadre des journées SPOT / PMNL, dans le cadre du GDR ROD (Recherche Opérationnelle et Décision) 2023, Toulouse (invited)
- GPCsign package
- Package Archissur soon on CRAN