Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Thierry Klein has a total of 40 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    On the Wasserstein Geodesic Principal Component Analysis of probability measures

    by Nina Vesseron, Elsa Cazelles, Alice Le Brigant, Thierry Klein

    This paper focuses on Geodesic Principal Component Analysis (GPCA) on a collection of probability distributions using the Otto-Wasserstein geometry. The goal is to identify geodesic curves in the space of probability measures that best capture the modes of variation of the underlying dataset. We first address the case of a collection of Gaussian distributions, and show how to lift the computations in the space of invertible linear maps. For the more general setting of absolutely continuous probability measures, we leverage a novel approach to parameterizing geodesics in Wasserstein space with neural networks. Finally, we compare to classical tangent PCA through various examples and provide illustrations on real-world datasets.

  • Open Access

    funGp: An R Package for Gaussian Process Regression with Scalar and Functional Inputs

    by José Betancourt, François Bachoc, Thierry Klein, Déborah Idier, Jérémy Rohmer, Yves Deville

    This article introduces funGp, an R package which handles regression problems involving multiple scalar and/or functional inputs, and a scalar output, through the Gaussian process model. This is particularly of interest for the design and analysis of computer experiments with expensive-to-evaluate numerical codes that take as inputs regularly sampled time series. Rather than imposing any particular parametric input-output relationship in advance (e.g., linear, polynomial), Gaussian process models extract this information directly from the data. The package offers built-in dimension reduction, which helps to simplify the representation of the functional inputs and obtain lighter models. It also implements an ant colony based optimization algorithm which supports the calibration of multiple structural characteristics of the model such as the state of each input (active or inactive) and the type of kernel function, while seeking for greater prediction power. The implemented methods are tested and applied to a real case in the domain of marine flooding.


  • Open Access

    Deviation results for Mandelbrot's multiplicative cascades with exponential tails

    by Thierry Klein, Agnès Lagnoux, P Petit

    Let $W$ be a nonnegative random variable with expectation $1$. For all $r \geqslant 2$, we consider the total mass $Z_r^\infty$ of the associated Mandelbrot multiplicative cascade in the $r$-ary tree. For all $n \geqslant 1$, we also consider the total mass $Z_r^n$ of the measure at height $n$ in the $r$-ary tree. Liu, Rio, Rouault \cite{lrr,liu2000limit,Rouault04} established large deviation results for $(Z_r^n)_{r \geqslant 2}$ for all $n \in [[1,\infty[[$ (resp., for $n = \infty$) in case $W$ has an everywhere finite cumulant generating function $Λ_W$ (resp., $W$ is bounded). Here, we extend these results to the case where $Λ_W$ is only finite on a neighborhood of zero. And we establish all deviation results (moderate, large, and very large deviations). It is noticeable that we obtain nonconvex rate functions. Moreover, our proof of upper bounds of deviations for $(Z_r^\infty)_{r \geqslant 2}$ rely on the moment bound instead of the standard Chernoff bound.

    • 19 October 2023
    • Published by HAL in HAL
    Open Access

    Deviation results for Mandelbrot's multiplicative cascades with exponential tails

    by Thierry Klein, Agnès Lagnoux, Pierre Petit

    Let W be a nonnegative random variable with expectation 1. For all r ⩾ 2, we consider the total mass Z ∞ r of the associated Mandelbrot multiplicative cascade in the r-ary tree. For all n ⩾ 1, we also consider the total mass Z n r of the measure at height n in the r-ary tree. Liu, Rio, Rouault [11, 12, 18] established large deviation results for (Z n r) r⩾2 for all n ∈ 1, ∞ (resp. for n = ∞) in case W has an everywhere finite cumulant generating function Λ W (resp. W is bounded). Here, we extend these results to the case where Λ W is only finite on a neighborhood of zero. And we establish all deviation results (moderate, large, and very large deviations). It is noticeable that we obtain nonconvex rate functions. Moreover, our proof of upper bounds of deviations for (Z ∞ r) r⩾2 rely on the moment bound instead of the standard Chernoff bound.

  • Open Access

    A Geometric Approach to Study Aircraft Trajectories: The Benefits of OpenSky Network ADS-B Data

    by Rémi Perrichon, Xavier Gendre, Thierry Klein

    To date, statistical analyses of aircraft trajectories have been under-exploited in the Airspace Traffic Management (ATM) literature. One reason is the need for advanced methods to tackle the high sampling irregularities and temporal correlations that both characterize trajectories. Differential geometry provides a relevant framework to study trajectories. By modeling trajectories as parameterized curves, shape analysis allows us to answer operational questions. This work presents a geodesic distance that rigorously defines and quantifies shape differences between aircraft trajectories. The key idea is to compare how the shape of a given trajectory changes from one popular data set (the Eurocontrol R&D data archive) to another one (OpenSky Network ADS-B data). Distances and geodesic paths are computed for a sample of flights that departed from Toulouse–Blagnac (LFBO) and landed at Paris–Orly (LFPO) in 2019. Its use for clustering purposes is illustrated and discussed.


  • Open Access

    Thermodynamic Modeling of the Uranium–Tellurium System: Estimation of the Uncertainties by a Bayesian Approach

    by Christine Guéneau, Eva Lawrence, Thierry Klein, Fabrice Gamboa

    Under irradiation, the formation of fission products in the (U,Pu)O2 fuel with time has a substantial effect on its chemistry. In particular, migration of the most volatile fission products (Cs, Te, I, Mo) from the center to the periphery of the fuel pellet is induced by the large radial thermal gradient. To predict the thermodynamic properties of the irradiated fuel, thermodynamic modeling of the complex multi-component (Cs-I-Te-Mo)–(U-Pu)–O system is performed using the CALPHAD method. In this work, the thermodynamic assessment of the U–Te sub-system is performed. The literature review reveals a lack of experimental data as well as scattering and inconsistency of some of the data. In particular, no thermodynamic data exist on the liquid. From this review, input thermodynamic and phase diagram data are carefully selected. The Gibbs energy functions are then adjusted by fitting these data. An overall good agreement is obtained with all the selected data except for the enthalpy of formation for UTe which is underestimated by 13% by our model. This could be due to an inconsistency between the enthalpy of formation and vapor pressure data. In a second step, the uncertainties on the thermodynamic parameters and their propagation on the calculated thermodynamic and phase diagram data are estimated using a Bayesian approach. The analysis shows that there are too many parameters (22) for too few data points (120 points). The uncertainties are thus large on some of the calculated data. Moreover the inconsistency of some of the data and the lack of thermodynamic data for the liquid makes the model uncertain. New experimental data such as heat capacity, enthalpy of formation for the compounds, and chemical potentials or activities for the liquid phase would improve the reliability of our model. Measurements of phase diagram data in the U–UTe2 region are also required. However this work provides the first detailed uncertainty analysis of the U–Te CALPHAD model. Moreover our approach, contrary to other Bayesian methods, provides an analytical posterior probability distribution and analytical credible intervals for the calculated thermodynamic quantities. It also speeds up the simulation of the uncertainty estimations on the phase diagram.


  • Large-deviation results for triangular arrays of semiexponential random variables

    by Thierry Klein, Agnès Lagnoux, Pierre Petit

    Asymptotics deviation probabilities of the sum of independent and identically distributed real-valued random variables have been extensively investigated, in particular when is not exponentially integrable. For instance, Nagaev (1969a, 1969b) formulated exact asymptotics results for with when has a semiexponential distribution. In the same setting, Brosset et al. (2020) derived deviation results at logarithmic scale with shorter proofs relying on classical tools of large-deviation theory and making the rate function at the transition explicit. In this paper we exhibit the same asymptotic behavior for triangular arrays of semiexponentially distributed random variables.


  • Open Access

    Deviation results for sparse tables in hashing with linear probing

    by Thierry Klein, A Lagnoux, P Petit

    We consider the model of hashing with linear probing and we establish the moderate and large deviations for the total displacement in sparse tables. In this context, Weibull-like-tailed random variables appear. Deviations for sums of such heavy-tailed random variables are studied in \cite{Nagaev69-1,Nagaev69-2}. Here we adapt the proofs therein to deal with conditioned sums of such variables and solve the open question in \cite{TFC12}. By the way, we establish the deviations of the total displacement in full tables, which can be derived from the deviations of empirical processes of i.i.d.\ random variables established in \cite{Wu94}..

  • Open Access

    A User-Oriented Local Coastal Flooding Early Warning System Using Metamodelling Techniques

    by Déborah Idier, Axel Aurouet, François Bachoc, Audrey Baills, José Betancourt, Fabrice Gamboa, Thierry Klein, Andrés F. López-Lopera, Rodrigo Pedreros, Jérémy Rohmer, et al.

    Given recent scientific advances, coastal flooding events can be properly modelled. Nevertheless, such models are computationally expensive (requiring many hours), which prevents their use for forecasting and warning. In addition, there is a gap between the model outputs and information actually needed by decision makers. The present work aims to develop and test a method capable of forecasting coastal flood information adapted to users’ needs. The method must be robust and fast and must integrate the complexity of coastal flood processes. The explored solution relies on metamodels, i.e., mathematical functions that precisely and efficiently (within minutes) estimate the results that would provide the numerical model. While the principle of relying on metamodel solutions is not new, the originality of the present work is to tackle and validate the entire process from the identification of user needs to the establishment and validation of the rapid forecast and early warning system (FEWS) while relying on numerical modelling, metamodelling, the development of indicators, and information technologies. The development and validation are performed at the study site of Gâvres (France). This site is subject to wave overtopping, so the numerical phase-resolving SWASH model is used to build the learning dataset required for the metamodel setup. Gaussian process- and random forest classifier-based metamodels are used and post-processed to estimate 14 indicators of interest for FEWS users. These metamodelling and post-processing schemes are implemented in an FEWS prototype, which is employed by local users and exhibits good warning skills during the validation period. Based on this experience, we provide recommendations for the improvement and/or application of this methodology and individual steps to other sites.