École Nationale Supérieure de Techniques Avancées

In principle, the temporal beating of superposed high harmonics obtained by focusing a femtosecond laser pulse in a gas jet can produce a train of very short intensity spikes, depending on the relative phases of the harmonics. We present a method to measure such phases through two-photon, two-color photoionization. We found that the harmonics are locked in phase and form a train of 250-attosecond pulses in the time domain. Harmonic generation may be a promising source for attosecond time-resolved measurements.

With the development of deep representation learning, the domain of reinforcement learning (RL) has become a powerful learning framework now capable of learning complex policies in high dimensional environments. This review summarises deep reinforcement learning (DRL) algorithms and provides a taxonomy of automated driving tasks where (D)RL methods have been employed, while addressing key computational challenges in real world deployment of autonomous driving agents. It also delineates adjacent domains such as behavior cloning, imitation learning, inverse reinforcement learning that are related but are not classical RL algorithms. The role of simulators in training agents, methods to validate, test and robustify existing solutions in RL are discussed.

In mechanically ventilated patients with acute circulatory failure related to sepsis, we investigated whether the respiratory changes in arterial pressure could be related to the effects of volume expansion (VE) on cardiac index (CI). Forty patients instrumented with indwelling systemic and pulmonary artery catheters were studied before and after VE. Maximal and minimal values of pulse pressure (Pp(max) and Pp(min)) and systolic pressure (Ps(max) and Ps(min)) were determined over one respiratory cycle. The respiratory changes in pulse pressure (DeltaPp) were calculated as the difference between Pp(max) and Pp(min) divided by the mean of the two values and were expressed as a percentage. The respiratory changes in systolic pressure (DeltaPs) were calculated using a similar formula. The VE-induced increase in CI was >/= 15% in 16 patients (responders) and < 15% in 24 patients (nonresponders). Before VE, DeltaPp (24 +/- 9 versus 7 +/- 3%, p < 0.001) and DeltaPs (15 +/- 5 versus 6 +/- 3%, p < 0.001) were higher in responders than in nonresponders. Receiver operating characteristic (ROC) curves analysis showed that DeltaPp was a more accurate indicator of fluid responsiveness than DeltaPs. Before VE, a DeltaPp value of 13% allowed discrimination between responders and nonresponders with a sensitivity of 94% and a specificity of 96%. VE-induced changes in CI closely correlated with DeltaPp before volume expansion (r(2) = 0. 85, p < 0.001). VE decreased DeltaPp from 14 +/- 10 to 7 +/- 5% (p < 0.001) and VE-induced changes in DeltaPp correlated with VE-induced changes in CI (r(2) = 0.72, p < 0.001). It was concluded that in mechanically ventilated patients with acute circulatory failure related to sepsis, analysis of DeltaPp is a simple method for predicting and assessing the hemodynamic effects of VE, and that DeltaPp is a more reliable indicator of fluid responsiveness than DeltaPs.

Exploratory activities seem to be intrinsically rewarding for children and crucial for their cognitive development. Can a machine be endowed with such an intrinsic motivation system? This is the question we study in this paper, presenting a number of computational systems that try to capture this drive towards novel or curious situations. After discussing related research coming from developmental psychology, neuroscience, developmental robotics, and active learning, this paper presents the mechanism of Intelligent Adaptive Curiosity, an intrinsic motivation system which pushes a robot towards situations in which it maximizes its learning progress. This drive makes the robot focus on situations which are neither too predictable nor too unpredictable, thus permitting autonomous mental development. The complexity of the robot's activities autonomously increases and complex developmental sequences self-organize without being constructed in a supervised manner. Two experiments are presented illustrating the stage-like organization emerging with this mechanism. In one of them, a physical robot is placed on a baby play mat with objects that it can learn to manipulate. Experimental results show that the robot first spends time in situations which are easy to learn, then shifts its attention progressively to situations of increasing difficulty, avoiding situations in which nothing can be learned. Finally, these various results are discussed in relation to more complex forms of behavioral organization and data coming from developmental psychology.

We consider least-squares problems where the coefficient matrices A,b are unknown but bounded. We minimize the worst-case residual error using (convex) second-order cone programming, yielding an algorithm with complexity similar to one singular value decomposition of A. The method can be interpreted as a Tikhonov regularization procedure, with the advantage that it provides an exact bound on the robustness of solution and a rigorous way to compute the regularization parameter. When the perturbation has a known (e.g., Toeplitz) structure, the same problem can be solved in polynomial-time using semidefinite programming (SDP). We also consider the case when A,b are rational functions of an unknown-but-bounded perturbation vector. We show how to minimize (via SDP) upper bounds on the optimal worst-case residual. We provide numerical examples, including one from robust identification and one from robust interpolation.

Although nonlinear methods can provide only the amplitude and the phase of an isolated ultrashort pulse, linear techniques can yield such measurements with a much better sensitivity and reliability when a reference pulse is available. We demonstrate two such methods, dual-quadrature spectral interferometry and Fourier-transform spectral interferometry. These techniques are simple to implement, very sensitive, and provide a complete measurement of the complex electric field, E(ω), as a continuous function of frequency.

In this paper we consider semidefinite programs (SDPs) whose data depend on some unknown but bounded perturbation parameters. We seek "robust" solutions to such programs, that is, solutions which minimize the (worst-case) objective while satisfying the constraints for every possible value of parameters within the given bounds. Assuming the data matrices are rational functions of the perturbation parameters, we show how to formulate sufficient conditions for a robust solution to exist as SDPs. When the perturbation is "full," our conditions are necessary and sufficient. In this case, we provide sufficient conditions which guarantee that the robust solution is unique and continuous (Hölder-stable) with respect to the unperturbed problem's data. The approach can thus be used to regularize ill-conditioned SDPs. We illustrate our results with examples taken from linear programming, maximum norm minimization, polynomial interpolation, and integer programming.

Relativistic interaction of short-pulse lasers with underdense plasmas has recently led to the emergence of a novel generation of femtosecond x-ray sources. Based on radiation from electrons accelerated in plasma, these sources have the common properties to be compact and to deliver collimated, incoherent, and femtosecond radiation. In this article, within a unified formalism, the betatron radiation of trapped and accelerated electrons in the so-called bubble regime, the synchrotron radiation of laser-accelerated electrons in usual meter-scale undulators, the nonlinear Thomson scattering from relativistic electrons oscillating in an intense laser field, and the Thomson backscattered radiation of a laser beam by laser-accelerated electrons are reviewed. The underlying physics is presented using ideal models, the relevant parameters are defined, and analytical expressions providing the features of the sources are given. Numerical simulations and a summary of recent experimental results on the different mechanisms are also presented. Each section ends with the foreseen development of each scheme. Finally, one of the most promising applications of laser-plasma accelerators is discussed: the realization of a compact free-electron laser in the x-ray range of the spectrum. In the conclusion, the relevant parameters characterizing each sources are summarized. Considering typical laser-plasma interaction parameters obtained with currently available lasers, examples of the source features are given. The sources are then compared to each other in order to define their field of applications.

We demonstrate the highest intensity - 300 TW laser by developing booster amplifying stage to the 50-TW-Ti:sapphire laser (HERCULES). To our knowledge this is the first multi-100TW-scale laser at 0.1 Hz repetition rate.

We demonstrate that a beam of x-ray radiation can be generated by simply focusing a single high-intensity laser pulse into a gas jet. A millimeter-scale laser-produced plasma creates, accelerates, and wiggles an ultrashort and relativistic electron bunch. As they propagate in the ion channel produced in the wake of the laser pulse, the accelerated electrons undergo betatron oscillations, generating a femtosecond pulse of synchrotron radiation, which has keV energy and lies within a narrow (50 mrad) cone angle.

We study high-order harmonic generation in the general case of absorbing and dispersive atomic gas media. For ultrashort laser pulses, the harmonic conversion efficiency tends to a limit mainly imposed by the harmonic reabsorption in the gas. This limit, independent on the gas density, is the same for both the case of a loosely focused beam or a beam guided in a gas-filled hollow-core fiber. Under optimum conditions, we measured the highest conversion efficiency to date $(4\ifmmode\times\else\texttimes\fi{}{10}^{\ensuremath{-}5})$ for the 15th harmonic generated in xenon using a 40 fs, 1.5 mJ, 800 nm pulse at a 1 kHz repetition rate.

Plasmas are an attractive medium for the next generation of particle accelerators because they can support electric fields greater than several hundred gigavolts per meter. These accelerating fields are generated by relativistic plasma waves-space-charge oscillations-that can be excited when a high-intensity laser propagates through a plasma. Large currents of background electrons can then be trapped and subsequently accelerated by these relativistic waves. In the forced laser wake field regime, where the laser pulse length is of the order of the plasma wavelength, we show that a gain in maximum electron energy of up to 200 megaelectronvolts can be achieved, along with an improvement in the quality of the ultrashort electron beam.

Monolayer MoS2 is a direct-gap two-dimensional semiconductor that exhibits strong electron-hole interactions, leading to the formation of stable excitons and trions. Here we report the existence of efficient exciton-exciton annihilation, a four-body interaction, in this material. Exciton-exciton annihilation was identified experimentally in ultrafast transient absorption measurements through the emergence of a decay channel varying quadratically with exciton density. The rate of exciton-exciton annihilation was determined to be (4.3 ± 1.1) × 10(-2) cm(2)/s at room temperature.

The laminarity of high-current multi-MeV proton beams produced by irradiating thin metallic foils with ultraintense lasers has been measured. For proton energies >10 MeV, the transverse and longitudinal emittance are, respectively, <0.004 mm mrad and <10(-4) eV s, i.e., at least 100-fold and may be as much as 10(4)-fold better than conventional accelerator beams. The fast acceleration being electrostatic from an initially cold surface, only collisions with the accelerating fast electrons appear to limit the beam laminarity. The ion beam source size is measured to be <15 microm (FWHM) for proton energies >10 MeV.

We show that a variety of antenna array pattern synthesis problems can be expressed as convex optimization problems, which can be (numerically) solved with great efficiency by recently developed interior-point methods. The synthesis problems involve arrays with arbitrary geometry and element directivity, constraints on far- and near-field patterns over narrow or broad frequency bandwidth, and some important robustness constraints. We show several numerical simulations for the particular problem of constraining the beampattern level of a simple array for adaptive and broadband arrays.

A very large high-energy shift of exciton resonances in GaAs multiple-quantum-well structures is observed during irradiation of the sample with femtosecond nonresonant radiation. This effect is discussed in terms of excitons "dressed" by photons.

We attribute a strong forward directed THz emission from femtosecond laser filaments in air to a transition-Cherenkov emission from the plasma space charge moving behind the ionization front at light velocity. Distant targets can be easily irradiated by this new source of THz radiation.

In robotic applications of visual simultaneous localization and mapping techniques, loop-closure detection and global localization are two issues that require the capacity to recognize a previously visited place from current camera measurements. We present an online method that makes it possible to detect when an image comes from an already perceived scene using local shape and color information. Our approach extends the bag-of-words method used in image classification to incremental conditions and relies on Bayesian filtering to estimate loop-closure probability. We demonstrate the efficiency of our solution by real-time loop-closure detection under strong perceptual aliasing conditions in both indoor and outdoor image sequences taken with a handheld camera.

The localization and solvation of excess electrons in pure water have been resolved at the femtosecond time scale. Before it becomes solvated, the electron thermalizes and reaches in 110 fs a localized state absorbing in the infrared. This transient species with lifetime 240 fs has been postulated to exist but has not been observed previously in liquid water.

Bulk damage induced by fs IR laser pulses in silica is investigated both experimentally and numerically. In a strong focusing geometry, a first damage zone is followed by a narrow track with submicron width, indicating a filamentary propagation. The shape and size of the damage tracks are shown to correspond to the zone where the electron density created by optical field ionization and avalanche is close to ${10}^{20}\text{ }\text{ }{\mathrm{c}\mathrm{m}}^{\ensuremath{-}3}$. The relative role of avalanche and photoionization is studied. The plasma density produced in the wake of the pulse is shown to saturate around $2--4\ifmmode\times\else\texttimes\fi{}{10}^{20}\text{ }\text{ }{\mathrm{c}\mathrm{m}}^{\ensuremath{-}3}$.