HAL-Polytechnique
Not a member yet
51406 research outputs found
Sort by
Distributional encoding for Gaussian process regression with qualitative inputs
No full textGaussian Process (GP) regression is a popular and sample-efficient approach for many engineering applications, where observations are expensive to acquire, and is also a central ingredient of Bayesian optimization (BO), a highly prevailing method for the optimization of black-box functions. However, when all or some input variables are categorical, building a predictive and computationally efficient GP remains challenging. Starting from the naive target encoding idea, where the original categorical values are replaced with the mean of the target variable for that category, we propose a generalization based on distributional encoding (DE) which makes use of all samples of the target variable for a category. To handle this type of encoding inside the GP, we build upon recent results on characteristic kernels for probability distributions, based on the maximum mean discrepancy and the Wasserstein distance. We also discuss several extensions for classification, multi-task learning and incorporation or auxiliary information. Our approach is validated empirically, and we demonstrate state-of-the-art predictive performance on a variety of synthetic and real-world datasets. DE is naturally complementary to recent advances in BO over discrete and mixed-spaces.</div
A theory of shape regularity for local regression maps
No full textWe introduce the concept of shape-regular regression maps as a framework to derive optimal rates of convergence for various non-parametric local regression estimators. Using Vapnik-Chervonenkis theory, we establish upper and lower bounds on the pointwise and the sup-norm estimation error, even when the localization procedure depends on the full data sample, and under mild conditions on the regression model. Our results demonstrate that the shape regularity of regression maps is not only sufficient but also necessary to achieve an optimal rate of convergence for Lipschitz regression functions. To illustrate the theory, we establish new concentration bounds for many popular local regression methods such as nearest neighbors algorithm, CART-like regression trees and several purely random trees including Mondrian trees
Can investors curb greenwashing?
No full textInternational audienceWe show how investors with pro-environmental preferences and who penalize revelations of past environmental controversies impact corporate greenwashing practices. Through a dynamic equilibrium model, we characterize firms' optimal environmental communication, green investments, and greenwashing policies, and we explain the forces driving them. Notably, under a condition that we explicitly characterize, companies greenwash to inflate their environmental rating above their fundamental environmental value, with an effort and impact increasing with investors' proenvironmental preferences. However, investment decisions that penalize greenwashing, policies increasing transparency, and environment-related technological innovation contribute to mitigating corporate greenwashing. We provide empirical support for our results
Parametric Study of the Performance of an Electrostatic Analyzer With an Hemispheric Field‐of‐View Based on the Donut Topology
No full textInternational audienceWe carried out a parametric study of the optical performance of an electrostatic analyzer based on the donut topology. The instantaneous hemispheric field‐of‐view of the optics eliminates the need of electrostatic deflectors, which are usually added to the energy analyzer in other designs to cover such a wide field‐of‐view. Parametrization of the donut topology and the use of parallel computing have enabled a wide parametric study of the instrument's performance as a function of the angle resolution of the instrument. We have identified a limited number of geometric parameters, including the outer radius of the detection system, which determine the geometric factor and energy resolution of the instrument for a given angular resolution. The average geometric factor per pixel of this 3D plasma camera varies in the range with an energy resolution between 9% and 14% for an energy limit of 20 keV. Our results suggest that a wide range of space missions could benefit from this new instrument concept. A low‐angular‐resolution version of the instrument could be installed on a nano‐satellite platform, for example, for space weather monitoring on low‐Earth orbit. For space missions requiring high‐angular‐resolution measurements, a high‐temporal‐resolution plasma camera would be able to provide the detailed distribution function of charged particles on larger platforms
Joint evolution of irrigation, the water cycle and water resources under a strong climate change scenario from 1950 to 2100 in the IPSL-CM6
No full textInternational audienceIrrigation, a key activity for food security, uses local water resources to increase evapotranspiration, creating feedback loops with the atmosphere and water resources. With climate change, it is unclear how irrigation will evolve in the future and how it may influence the evolution of water resources and the water cycle. It is also unclear whether irrigation may be constrained by climate change or water resource shortages. Here, we compare two surface-atmosphere simulations performed with the IPSL-CM6 model from 1950-2100: one with irrigation and one without irrigation. In both simulations, the evolutions of atmospheric radiative forcing, land use, and irrigated areas are taken from CMIP6, which uses a historical dataset for the data before 2014 and the SSP5-RCP8.5 dataset for data after 2014. The two simulations reveal strong global warming and precipitation increases between 1950-2000 and 2050-2100 average values (+5.6 °C and +8.1 %, on average, over land with irrigation). Over the same period, our results indicate an increase in irrigation (+76 % increase in irrigation in the 2050-2100 compared to the 1950-2000 period), which is in line with an important expansion of irrigated areas. The influence of irrigation on evapotranspiration in irrigated areas is greater in 2050-2100 than in 1950-2000 (+12 % vs. +8 %, respectively). Evapotranspiration has also been found to increase in non-irrigated areas near irrigated zones owing to an increase in precipitation under historical and future climate conditions. Water depletion due to irrigation is more intense in the future than in the historical period, although climate change increases water storages and river discharge due to more precipitation in the future. We also identified areas where future environmental conditions can limit irrigation or where irrigation can increase tensions over water use (approximately one-third of irrigated areas, including the Mediterranean basin, California, and Southeast Asia). Our results highlight the importance of considering irrigation in climate projections and future water resources assessments
Bipyridyl Ruthenium Complexes Featuring P-Ylide Ligands: A Comparative Study of Their Redox and Photophysical Properties
No full textInternational audienceThe electrochemical and photophysical properties of two series of ruthenium bis(bipyridine) complexes featuring either pyridine-phosphonium ylide (PC) or pyridine-iminophosphorane (PN) ligand were synthesized. Three different bipyridines were employed allowing the synthesis of six P-ylide RuII complexes. All the complexes were fully characterized by multinuclear NMR spectroscopy, HR-mass spectrometry, as well as X-ray crystallography and no major difference was observed between them. In cyclic voltammetry, the measured potentials agree with the electronic properties of the bipyridine ligands and underline the strong electron donation of the phosphonium ylide. Indeed, the oxidation is facilitated and the reductions become more difficult compared to the corresponding Ru(Rbpy)3 complexes. The absorption spectra of the PC and PN RuII complexes are similar with a bathochromic shift compared to polypyridine RuII complexes, which is amplified in presence of electron-withdrawing substituents on the bipyridine. Regarding luminescence, the PC complexes emit at lower energy than the PN analogues. In addition, the associated lifetime depends both on the nature of the P-ylide and the substituents on the bipyridine ligand. This was related to different adiabatic 3MC/3MLCT gap calculated at the DFT level which is larger for two iminophosphorane complexes exhibiting a lifetime of the order of a few hundred of nanoseconds
Origin, evolution, and fate of Titan’s polar clouds
No full textInternational audienceSaturn’s moon Titan exhibits a rich interplay of atmospheric chemistry and meteorology, including the seasonal formation of polar stratospheric clouds, first observed by Voyager and later by the Cassini mission. Using a global climate model with haze and cloud microphysics, we investigate their origin, evolution, and fate. We find that cloud formation begins in early autumn, triggered by rapid radiative cooling and chemical enrichment within the stratospheric polar vortex. Initially extending to 336 km altitude and composed of benzene and hydrogen cyanide ices, these clouds descend to lower altitudes during autumn and winter, evolving in composition as additional species condense, before dissipating in spring. This process reflects a unified, hemisphere-spanning seasonal mechanism shaping Titan’s climate. Our model explains observed seasonal patterns and predicts the onset of the next northern polar cloud in 2027. These results provide a predictive framework for future Titan observations, missions, and long-term climate evolution studies
On the Robustness of Nelder-Mead to Positive and Negative Noise Outliers with Heavy-Tails on the BBOB Test Suite
No full textInternational audienceWe investigate the robustness to noise and outliers of the Nelder-Mead derivative-free optimization algorithm implemented in the scipy.optimize Python library. Using the noisifier module from the COCO platform, we investigate the impact of adding or subtracting a positive Cauchy noise to the function value with a certain probability p. This probability is varied from 0 till 0.4 allowing to appraise the impact of the noise. We find that Nelder-Mead has highly asymmetric performances with respect to adding a positive noise (i.e. degrading artificially possibly good solutions) or adding a negative noise (i.e. making possibly bad solutions appear good). When adding positive noise, Nelder-Mead shows some robustness in low dimensions (≤ 5). However adding negative noise has very detrimental consequences for the performances of Nelder-Mead, even with a low value of p or in low dimension
Une couverture contre la précarité énergétique
No full textInternational audienceThe French government implemented energy price shields in the fall of 2021 to protect the French households and their purchasing power. We propose another measure to cover households against fuel poverty: an insurance that guarantees a minimum level of energy to households. It applies to non-fuel-poor households and is differentiated by income and housing characteristics. For illustration, we focus on non-fuel poor households that live in all-electric dwellings and whose electricity price is regulated. We estimate that 2.06 TWh are needed to guarantee the minimum level of electricity to these households. This measure will be less expensive than the cost of the tariff shield for those households as long as the sourcing price of electricity stays under 803€/MWh!Fin 2021, le gouvernement français a mis en place des boucliers tarifaires pour protéger le pouvoir d'achat des ménages français. Ces boucliers n'ont hélas pas empêché certains ménages de tomber dans la précarité énergétique. Nous proposons une mesure alternative de court terme pour couvrir les ménages non contraints financièrement du risque de précarité énergétique. Cette couverture leur garantit un niveau minimum d'énergie leur permettant de vivre décemment. Ce niveau minimum et donc l'assurance sont spécifiques aux caractéristiques du ménage et de son logement. À titre d'illustration, nous calibrons une assurance pour les ménages non contraints financièrement qui vivent dans des logements tout-électrique. Nous estimons qu'en 2022, une quantité de 2,06 TWh d'électricité aurait été nécessaire pour couvrir ces ménages. Cette assurance énergie aurait été moins coûteuse, pour ces ménages, que le bouclier tarifaire pour tout prix d'approvisionnement en électricité inférieur à 803€/MWh
