Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Generation of meter-scale hydrogen plasmas and efficient, pump-depletion-limited wakefield excitation using 10 GeV electron bunches
No full textInternational audienceHigh repetition rates and efficient energy transfer to the accelerating beam are important for a future linear collider based on the beam-driven plasma wakefield acceleration scheme (PWFA-LC). This paper reports the first results from the Plasma Wakefield Acceleration Collaboration (E300) that are beginning to address both of these issues using the recently commissioned FACET-II facility at SLAC. We have generated meter-scale hydrogen plasmas using time-structured 10 GeV electron bunches from FACET-II, which hold the promise of dramatically increasing the repetition rate of PWFA by rapidly replenishing the gas between each shot compared to the hitherto used lithium plasmas that operate at 1-10 Hz. Furthermore, we have excited wakes in such plasmas that are suitable for high gradient particle acceleration with high drive-bunch to wake energy transfer efficiency -- a first step in achieving a high overall energy transfer efficiency. We have done this by using time-structured electron drive bunches that typically have one or more ultra-high current (>30 kA) femtosecond spike(s) superimposed on a longer (~0.4 ps) lower current (1.5 Torr. We also show that some electrons in the rear of the bunch can gain several GeV energies from the wake. These results are reproduced by particle-in-cell simulations using the QPAD code. At a pressure of ~2 Torr, simulations results and experimental data show that the beam transfers about 60% of its energy to the wake
Titanium-mediated expedient synthesis of complex nitrogen heterocycles
No full textInternational audienc
The SocialAI School: Insights from Developmental Psychology Towards Artificial Socio-Cultural Agents
No full textPreprint, see v1 for a shorter version (accepted at the "Workshop on Theory-of-Mind" at ICML 2023) See project website for demo and code: https://sites.google.com/view/socialai-schoolInternational audienceDevelopmental psychologists have long-established the importance of socio-cognitive abilities in human intelligence. These abilities enable us to enter, participate and benefit from human culture. AI research on social interactive agents mostly concerns the emergence of culture in a multi-agent setting (often without a strong grounding in developmental psychology). We argue that AI research should be informed by psychology and study socio-cognitive abilities enabling to enter a culture too. We discuss the theories of Michael Tomasello and Jerome Bruner to introduce some of their concepts to AI and outline key concepts and socio-cognitive abilities. We present The SocialAI school - a tool including a customizable parameterized uite of procedurally generated environments, which simplifies conducting experiments regarding those concepts. We show examples of such experiments with RL agents and Large Language Models. The main motivation of this work is to engage the AI community around the problem of social intelligence informed by developmental psychology, and to provide a tool to simplify first steps in this direction. Refer to the project website for code and additional information: https://sites.google.com/view/socialai-school
Machine culture
No full textInternational audienceThe ability of humans to create and disseminate culture is often credited as the single most important factor of our success as a species. In this Perspective, we explore the notion of ‘machine culture’, culture mediated or generated by machines. We argue that intelligent machines simultaneously transform the cultural evolutionary processes of variation, transmission and selection. Recommender algorithms are altering social learning dynamics. Chatbots are forming a new mode of cultural transmission, serving as cultural models. Furthermore, intelligent machines are evolving as contributors in generating cultural traits—from game strategies and visual art to scientific results. We provide a conceptual framework for studying the present and anticipated future impact of machines on cultural evolution, and present a research agenda for the study of machine culture
SBMLTOODEJAX: Efficient Simulation and Optimization of Biological Network Models in JAX
No full textInternational audienceAdvances in bioengineering and biomedicine demand a deep understanding of the dynamic behavior of biological systems, ranging from protein pathways to complex cellular processes. Biological networks like gene regulatory networks and protein pathways are key drivers of embryogenesis and physiological processes. Comprehending their diverse behaviors is essential for tackling diseases, including cancer, as well as for engineering novel biological constructs. Despite the availability of extensive mathematical models represented in Systems Biology Markup Language (SBML), researchers face significant challenges in exploring the full spectrum of behaviors and optimizing interventions to efficiently shape those behaviors. Existing tools designed for simulation of biological network models are not tailored to facilitate interventions on network dynamics nor to facilitate automated discovery. Leveraging recent developments in machine learning (ML), this paper introduces SBMLtoODEjax, a lightweight library designed to seamlessly integrate SBML models with ML-supported pipelines, powered by JAX. SBMLtoODEjax facilitates the reuse and customization of SBML-based models, harnessing JAX's capabilities for efficient parallel simulations and optimization, with the aim to accelerate research in biological network analysis
Fourier Features in Reinforcement Learning with Neural Networks
No full textInternational audienceIn classic Reinforcement Learning (RL), encoding the inputs with a Fourier feature mapping is a standard way to facilitate generalization and add prior domain knowledge. In Deep RL, such input encodings are less common since they could, in principle, be learned by the network and may therefore seem less beneficial. In this paper, we present experiments on Multilayer Perceptrons (MLP) that indicate that even in Deep RL, Fourier features can lead to significant performance gains in both rewards and sample efficiency. Furthermore, we observe that they increase the robustness with respect to hyperparameters, lead to smoother policies, and benefit the training process by reducing learning interference, encouraging sparsity, and increasing the expressiveness of the learned features. However, a major bottleneck with conventional Fourier features is that the number of features increases exponentially with the state dimension. As a remedy, we propose a simple, light version that only has a linear number of features yet empirically provides similar benefits. Our experiments cover both shallow/deep, discrete/continuous, and on/off-policy RL settings
Power Allocation Techniques for Non-orthogonal Multiple Access Based MIMO Visible Light Communication Systems
No full textInternational audienceVisible light communication (VLC) is a promising solution for indoor wireless connectivity but its capacity is hindered owing to the light-emitting diodes (LEDs) bandwidth limitation. In this paper, a non-orthogonal multiple access (NOMA) scheme is applied with a multiple-input multiple-output (MIMO) technique for enhancing the achievable rate of VLC systems. As the NOMA performance is extremely affected by the power allocation technique, we evaluate, in an indoor 2x2 MIMO-VLC system based on the NOMA scheme, the performance of three efficient and low-complexity power allocation schemes: fixed power allocation (FPA), gain ratio power allocation (GRPA), normalized gain difference power allocation (NGDPA). According to the results we obtained, the system's best achievable rate was achieved using the FPA with a power allocation factor of 0.9. Moreover, the superiority of FPA over NGDPA and GRPA is clearly highlighted through our simulation results in a two-user case scenario with a sum rate enhancement of 3.7% and 18.7%, respectively
Trainable dynamical estimation of above-surface wind speed using underwater passive acoustics
No full textInternational audienceCovering more than 70% of Earth surface, oceans play a key role in climate regulation, are the main medium of world commercial trade and are a source of renewable energy, to cite few aspects. Despite its importance, ocean surface state reconstruction poses some challenges, due to its non-linear behavior and the heterogeneity of the spatio-temporal scales involved. State-of-the-art techniques for forecast and prediction involve numerical weather models, such as data assimilation approaches. Besides, remote sensing techniques deliver finer-grained information about the surface state. Among others, underwater passive acoustics uses the underwater soundscape to infer the above-surface atmospheric state. In this work, with a particular focus on the surface wind speed reconstruction, we propose a framework that bridges data assimilation and machine learning schemes, to exploit both the prior physical knowledge and the capability of machine learning modelling to take advantage of large data bases. Extensive numerical experiments show that this hybrid framework can outperform the state-of-the-art data-driven models with a relative gain up to 16% in terms of root mean squared error. Experiments also involve tests on multi-modal data, namely underwater passive acoustics and wind speed reanalyses, giving promising results
Channeling acceleration in crystals and nanostructures and studies of solid plasmas: new opportunities
No full textInternational audiencePlasma wakefield acceleration (PWFA) has shown illustriousprogress and resulted in an impressive demonstration of tens of GeVparticle acceleration in meter-long single structures. To reach evenhigher energies in the 1 TeV to 10 TeV range, a promising schemeis channeling acceleration in solid-density plasmas within crystalsor nanostructures.The E336 experiment studies the beam-nanotarget interactionwith the highly compressed electron bunches available at theFACET-II accelerator. These studies furthermore involve an in-depthresearch on dynamics of beam-plasma instabilities in ultra-denseplasma, its development and suppression in structured media likecarbon nanotubes and crystals, and its potential use to transverselymodulate the electron bunch
Nouméa: a new multi-mission calibration and validation site for past and future altimetry missions?
No full textInternational audienceToday, monitoring the evolution of sea level in coastal areas is of importance, since almost 11 % of the world's population lives in low-lying areas. Reducing uncertainties in sea level estimates requires a better understanding of both altimetry measurements and local sea level dynamics. In New Caledonia, the Nouméa lagoon is an example of this challenge, as altimetry, coastal tide gauge, and vertical land motions from global navigation satellite systems (GNSSs) do not provide consistent information. The GEOCEAN-NC 2019 field campaign addresses this issue with deployments of in situ instruments in the lagoon (GNSS buoy, pressure gauge, etc.), with a particular focus on the crossover of one Jason-series track and two Sentinel-3A missions tracks. In this study, we propose a method to virtually transfer the Nouméa tide gauge at the altimetry crossover point, using in situ data from the field campaign. Following the philosophy of calibration and validation (Cal/Val) studies, we derive absolute altimeter bias time series over the entire Jason and Sentinel-3A periods. Overall, our estimated altimeter mean biases are slightly larger by 1–2 cm compared to Corsica and Bass Strait results, with inter-mission biases in line with those of Bass Strait site. Uncertainties still remain regarding the determination of our vertical datum, only constrained by the three days of the GNSS buoy deployment. With our method, we are able to re-analyse about 20 years of altimetry observations and derive a linear trend of −0.2 ± 0.1 mm yr−1 over the bias time series. Compared to previous studies, we do not find any significant uplift in the area, which is more consistent with the observations of inland permanent GNSS stations. These results support the idea of developing Cal/Val activities in the lagoon, which is already the subject of several experiments for the scientific calibration phase of the SWOT wide-swath altimetry mission