HAL Evry
Not a member yet
19237 research outputs found
Sort by
Greedy window stochastic optimization algorithm for reducing data center energy consumption
No full textInternational audienceThis paper presents a stochastic model for Dynamic Power Management (DPM) in data centers-a strategy that involves dynamically activating and deactivating servers to balance energy efficiency with the maintenance of high Quality of Service (QoS). Unlike traditional methods that rely on predefined job arrival distributions, our approach leverages histograms to dynamically characterize job arrival patterns derived from real-world traces, empirical data, or live traffic measurements. We model the data center as a queuing system and introduce an efficient Greedy Window algorithm that is fast, memory-efficient, and capable of adapting to real-time fluctuations in job arrivals. This algorithm computes a suboptimal policy that closely approximates the optimal strategy attainable through stochastic combinatorial optimization techniques such as the Markov Decision Process (MDP). However, MDP-based approaches face challenges related to state space explosion and substantial computational demands in both time and memory. We evaluate our method across a range of data center configurations using real Google traffic traces and compare its performance against several optimization strategies
MYv7: New 3D Monocular Object Detection Improvement for Road and Railway Smart Mobility
No full textInternational audienceThis paper presents MYv7 (Mono-YOLOv7), an adaptation of the YOLOv7 architecture tailored specifically for 3D monocular object detection. Rather than competing with specialized 3D methods, we demonstrate the efficacy of enhancing 3D monocular detection using improved 2D object detection algorithms. We showcase how improvements in 2D algorithms can enhance 3D predictions, presenting MYv7’s twofold advantage over a YOLOv5-based method: increased speed and accuracy. These gains are crucial for efficient operation on embedded systems with limited computational resources. Our results highlight the potential of using advancements in 2D detection methods to significantly improve 3D monocular object recognition, opening new avenues for real-world applications
A genome-wide segmentation approach for the detection of selection footprints
No full textAbstract Motivation In population genetics, the detection of genomic regions under positive selection is essential to understand the genetic basis of locally adaptive trait variation. We propose a principled approach to detect those regions that combines a robust moment based F ST estimator with a segmentation algorithm. Results Our approach allows for pairwise comparisons of populations and does not require any prior knowledge about the size of the regions to be detected. The procedure runs within seconds even for large genome datasets with millions of SNPs, and provides a complete landscape of the F ST distribution over the chromosome. The procedure comes with a grounded estimator of the baseline F ST level, allowing the detection of regions exhibiting high departures from this reference value. The potential of our procedure is illustrated in two applications in animal and human population genetics. We were able to recover in a matter of seconds regions known to be under selection, often with greater precision than what was reported in previous studies. Availability Our approach is implemented in the fst4pg R package available from the CRAN repository. The Sheep dataset is downloadable from the Zenodo repository https://doi.org/10.5281/zenodo.237116 . The 1000 Genome dataset is downloadable from ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/2013050
Bookstores of the World
No full textInternational audienceEven in these days of online commerce, the physical bookstore retains its fascination as a place to gather and discover new art and ideas, often serendipitously; in fact, the bookstore is a refuge from the often unreflective world outside. This splendid tribute to the bookstores of the world begins in France—which has the most bookstores per capita of any country—and continues throughout Europe, the Americas, the Near East, Asia, and Africa. Along the way we encounter legendary emporia such as City Lights in San Francisco and Foyle’s in London, as well as new innovators such as the architecturally breathtaking Zhongshuge in Hangzhou. The insightful text explores the factors that have shaped bookselling in each region, from the price-fixing laws of France and Germany to the cultural and geographic diversity of the United States
Librairies dans le monde
No full textInternational audienceDe la mythique City Lights Bookstore de San Francisco à la boutique-librairie du Caire, en passant par les 50 kilomètres de rayonnages de Foyles à Londres, l'incontournable Shakespeare and Co à Paris, la librairie néogothique Lello à Porto ou la labyrinthique Xidan à Pékin, ce livre ambitieux montre la variété des librairies dans le monde. Considérés comme "essentiels" par le gouvernement français lors de la fermeture de 2021, ces espaces de vente et de promotion du livre, riches d'une histoire millénaire (la librairie telle que nous la connaissons est attestée à Athènes aux Ve et VIe siècles avant notre ère), ont évolué au fil du temps, s'adaptant aux mutations de l'édition, aux fluctuations politiques, aux impératifs commerciaux et aux nouvelles formes de sociabilité.En présentant près de 250 librairies - historiques et contemporaines -, les auteurs invitent le lecteur à un tour du monde qui reflète la vitalité de ce haut lieu culturel
Gain and loss of gene function shaped the nickel hyperaccumulation trait in Noccaea caerulescens
No full textInternational audienceNickel hyperaccumulation is an extreme adaptation to ultramafic soils observed in more than 500 plant species. However, our understanding of the molecular mechanisms underlying the evolution of this trait remains limited. To shed light on these mechanisms, we have generated a high-quality genome assembly of the metal hyperaccumulator Noccaea caerulescens. We then used this genome as reference to conduct comparative intraspecific and interspecific transcriptomic analyses using various accessions of N. caerulescens and the non-accumulating relative Microthlaspi perfoliatum, to identify genes associated with nickel hyperaccumulation. Our results suggest a correlation between nickel hyperaccumulation and a decrease in the expression of genes involved in defense responses and the regulation of membrane trafficking. Surprisingly, these analyses did not reveal a significant enrichment of genes involved in the regulation of metal homeostasis. However, we found that the expression levels of selected metal transporters, namely NcHMA3, NcHMA4 and NcIREG2, is consistently elevated in N. caerulescens accessions hyperaccumulating nickel. Furthermore, our analyses identified frameshift mutations in NcIRT1 associated with the loss of nickel hyperaccumulation in a few accessions. We further showed that the expression of a functional NcIRT1 in roots of the La Calamine accession increases nickel accumulation in shoots. Our results demonstrate that NcIRT1 participate in nickel hyperaccumulation in N. caerulescens. They also suggest that nickel hyperaccumulation is an ancient trait in N. caerulescens that has evolved from the high and constitutive expression of few metal transporters including NcIREG2 and that the trait was subsequently lost in a few accessions due to mutations in NcIRT1
Paramètres d'effets ou de synthétiseurs audios: estimer des distributions plutôt que des valeurs déterministes
No full textInternational audienceAudio effects and sound synthesizers are widely used processors in popular music. Their parameters control the quality of the output sound. Multiple combinations of parameters can lead to the same sound. While recent approaches have been proposed to estimate these parameters given only the output sound, those are deterministic, i.e. they only estimate a single solution among the many possible parameter configurations. In this work, we propose to model the parameters as probability distributions instead of deterministic values. To learn the distributions, we optimize two objectives: (1) we minimize the reconstruction error between the ground truth output sound and the one generated using the estimated parameters, as is it usually done, but also (2) we maximize the parameter diversity, using entropy. We evaluate our approach through two numerical audio experiments to show its effectiveness. These results show how our approach effectively outputs multiple combinations of parameters to match one sound
Memory and oblivion:how the past obliterated the 1929 crisis in France
No full textInternational audienceOral presentation duringe session "Financial Crises: Analyses, Perceptions, Memories in a Long-run Historical Perspective (Part 1/2)" (id 66211
Co-Design of Functional Interval Observer-Based Control for Uncertain Linear Parameter Varying Switched Systems
No full textInternational audienceThis paper presents a novel method for the co-design of observers and controllers for switched linear parameter-varying (LPV) systems subject to unknown but bounded uncertainties, disturbances, and faults. First, a polytopic functional interval observer (FIO) is employed to estimate the lower and upper bounds of the system states. Next, a proportional-integral (PI) observer is designed to estimate fault signals accurately. Building on these estimates, a fault-tolerant control (FTC) strategy is developed to ensure the stability of the closed-loop system and maintain reference model tracking in the presence of faults. The sufficient conditions for the existence of observers and controllers are derived using Linear Matrix Inequalities (LMIs), leveraging multiple Lyapunov functions and ensuring input-to-state stability (ISS) under the average dwell time (ADT) approach. Finally, the effectiveness of the proposed approach is validated through a simulation applied to vehicle lateral dynamics estimation and control