Structural change in multipartite entanglement sharing: A random matrix approach

We study the typical entanglement properties of a system comprising two independent qubit environments interacting via a shuttling ancilla. The initial preparation of the environments is modeled using random matrix techniques. The entanglement measure used in our study is then averaged over many histories of randomly prepared environmental states. Under a Heisenberg interaction model, the average entanglement between the ancilla and one of the environments remains constant, regardless of the preparation of the latter and the details of the interaction. We also show that, upon suitable kinematic and dynamical changes in the ancillaenvironment subsystems, the entanglement-sharing structure undergoes abrupt modifications associated with a change in the multipartite entanglement class of the overall system's state. These results are invariant with respect to the randomized initial state of the environments.

Ultra-fast spin dynamics in the 2s2p^2 configuration of C^+ studied by time-dependent R-matrix theory

We employ time-dependent R-matrix theory to study ultra-fast dynamics in the doublet 2s2p(2) configuration of C+ for a total magnetic quantum number M = 1. In contrast to the dynamics observed for M = 0, ultra-fast dynamics for M = 1 is governed by spin dynamics in which the 2s electron acts as a flag rather than a spectator electron. Under the assumption that m(S) = 1/2, m(2s) = 1/2 allows spin dynamics involving the two 2p electrons, whereas m(2s) = -1/2 prevents spin dynamics of the two 2p electrons. For a pump-probe pulse scheme with (h) over bar omega(pump) = 10.9 eV and (h) over bar omega(probe) = 16.3 eV and both pulses six cycles long, little sign of spin dynamics is observed in the total ionization probability. Signs of spin dynamics can be observed, however, in the ejected-electron momentum distributions. We demonstrate that the ejected-electron momentum distributions can be used for unaligned targets to separate the contributions of initial M = 0 and M = 1 levels. This would, in principle, allow unaligned target ions to be used to obtain information on the different dynamics in the 2s2p(2) configuration for the M = 0 and M = 1 levels from a single experime

B-spline methods in R-matrix theory for scattering in two-electron systems

The use of B-spline basis sets in R-matrix theory for scattering processes has been investigated. In the present approach a B-spline basis is used for the description of the inner region, which is matched to the physical outgoing wavefunctions by the R-matrix. Using B-splines, continuum basis functions can be determined easily, while pseudostates can be included naturally. The accuracy for low-energy scattering processes is demonstrated by calculating inelastic scattering cross sections for e colliding on H. Very good agreement with other calculations has been obtained. Further extensions of the codes to quasi two-electron systems and general atoms are discussed as well as the application to (multi) photoionization.

Atomic harmonic generation in time-dependent R-matrix theory

We have developed the capability to determine accurate harmonic spectra for multielectron atoms within time-dependent R-matrix (TDRM) theory. Harmonic spectra can be calculated using the expectation value of the dipole length, velocity, or acceleration operator. We assess the calculation of the harmonic spectrum from He irradiated by 390-nm laser light with intensities up to 4 x 10(14) W cm(-2) using each form, including the influence of the multielectron basis used in the TDRM code. The spectra are consistent between the different forms, although the dipole acceleration calculation breaks down at lower harmonics. The results obtained from TDRM theory are compared with results from the HELIUM code, finding good quantitative agreement between the methods. We find that bases which include pseudostates give the best comparison with the HELIUM code, but models comprising only physical orbitals also produce accurate results.

Time-dependent R-matrix theory applied to two-photon double ionization of He

We introduce a time-dependent R-matrix theory generalized to describe double-ionization processes. The method is used to investigate two-photon double ionization of He by intense XUV laser radiation. We combine a detailed B-spline-based wave-function description in an extended inner region with a single-electron outer region containing channels representing both single ionization and double ionization. A comparison of wave-function densities for different box sizes demonstrates that the flow between the two regions is described with excellent accuracy. The obtained two-photon double-ionization cross sections are in excellent agreement with other cross sections available. Compared to calculations fully contained within a finite inner region, the present calculations can be propagated over the time it takes the slowest electron to reach the boundary.

Harmonic generation by noble-gas atoms in the near-IR regime using ab initio time-dependent R-matrix theory

We demonstrate the capability of ab initio time-dependent R-matrix theory to obtain accurate harmonic generation spectra of noble-gas atoms at near-IR wavelengths between 1200 and 1800 nm and peak intensities up to 1.8 × 10^(14) W/cm^(2). To accommodate the excursion length of the ejected electron, we use an angular-momentum expansion up to Lmax=279. The harmonic spectra show evidence of atomic structure through the presence of a Cooper minimum in harmonic generation for Kr, and of multielectron interaction through the giant resonance for Xe. The theoretical spectra agree well with those obtained experimentally.

Double ionization in R-matrix theory using a two-electron outer region

We have developed a two-electron outer region for use within R-matrix theory to describe double ionisation processes. The capability of this method is demonstrated for single-photon double ionisation of He in the photon energy region between 80 eV to 180 eV. The cross sections are in agreement with established data. The extended RMT method also provides information on higher-order processes, as demonstrated by the identification of signatures for sequential double ionisation processes involving an intermediate He⁺ state with n=2.

Applications of random set theory in econometrics

In recent years, the econometrics literature has shown a growing interest in the study of partially identified models, in which the object of economic and statistical interest is a set rather than a point. The characterization of this set and the development of consistent estimators and inference procedures for it with desirable properties are the main goals of partial identification analysis. This review introduces the fundamental tools of the theory of random sets, which brings together elements of topology, convex geometry, and probability theory to develop a coherent mathematical framework to analyze random elements whose realizations are sets. It then elucidates how these tools have been fruitfully applied in econometrics to reach the goals of partial identification analysis.

Plane-wave scattering-matrix theory of antennas and antenna-antenna interactions /

Mode of access: Internet.

Ergodic capacity and outage performance of amplify-and-forward protocols

This thesis analyses the performance bounds of amplify-and-forward relay channels which are becoming increasingly popular in wireless communication applications. The statistics of cascaded Nakagami-m fading model which is a major obstacle in evaluating the outage of wireless networks is analysed using Mellin transform. Furthermore, the upper and the lower bounds for the ergodic capacity of the slotted amplify-and-forward relay channel, for finite and infinite number of relays are derived using random matrix theory. The results obtained will enable wireless network designers to optimize the network resources, benefiting the consumers.

Aspects géométriques et intégrables des modèles de matrices aléatoires

Travail réalisé en cotutelle avec l'université Paris-Diderot et le Commissariat à l'Energie Atomique sous la direction de John Harnad et Bertrand Eynard.

Advances in scaling deep learning algorithms

Les algorithmes d'apprentissage profond forment un nouvel ensemble de méthodes puissantes pour l'apprentissage automatique. L'idée est de combiner des couches de facteurs latents en hierarchies. Cela requiert souvent un coût computationel plus elevé et augmente aussi le nombre de paramètres du modèle. Ainsi, l'utilisation de ces méthodes sur des problèmes à plus grande échelle demande de réduire leur coût et aussi d'améliorer leur régularisation et leur optimization. Cette thèse adresse cette question sur ces trois perspectives. Nous étudions tout d'abord le problème de réduire le coût de certains algorithmes profonds. Nous proposons deux méthodes pour entrainer des machines de Boltzmann restreintes et des auto-encodeurs débruitants sur des distributions sparses à haute dimension. Ceci est important pour l'application de ces algorithmes pour le traitement de langues naturelles. Ces deux méthodes (Dauphin et al., 2011; Dauphin and Bengio, 2013) utilisent l'échantillonage par importance pour échantilloner l'objectif de ces modèles. Nous observons que cela réduit significativement le temps d'entrainement. L'accéleration atteint 2 ordres de magnitude sur plusieurs bancs d'essai. Deuxièmement, nous introduisont un puissant régularisateur pour les méthodes profondes. Les résultats expérimentaux démontrent qu'un bon régularisateur est crucial pour obtenir de bonnes performances avec des gros réseaux (Hinton et al., 2012). Dans Rifai et al. (2011), nous proposons un nouveau régularisateur qui combine l'apprentissage non-supervisé et la propagation de tangente (Simard et al., 1992). Cette méthode exploite des principes géometriques et permit au moment de la publication d'atteindre des résultats à l'état de l'art. Finalement, nous considérons le problème d'optimiser des surfaces non-convexes à haute dimensionalité comme celle des réseaux de neurones. Tradionellement, l'abondance de minimum locaux était considéré comme la principale difficulté dans ces problèmes. Dans Dauphin et al. (2014a) nous argumentons à partir de résultats en statistique physique, de la théorie des matrices aléatoires, de la théorie des réseaux de neurones et à partir de résultats expérimentaux qu'une difficulté plus profonde provient de la prolifération de points-selle. Dans ce papier nous proposons aussi une nouvelle méthode pour l'optimisation non-convexe.