A5052608804- Markov Chains and Monte Carlo Methods
- Stochastic processes and financial applications
- Advanced Control Systems Optimization
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Queuing Theory Analysis
- Simulation Techniques and Applications
- Fault Detection and Control Systems
- Reinforcement Learning in Robotics
- Reliability and Maintenance Optimization
- Optimization and Search Problems
- Landslides and related hazards
- Economic theories and models
- Scheduling and Optimization Algorithms
- Risk and Portfolio Optimization
- Cryospheric studies and observations
- Control Systems and Identification
- Stochastic processes and statistical mechanics
- Astrophysical Phenomena and Observations
- Traffic control and management
- Insurance, Mortality, Demography, Risk Management
- Stability and Control of Uncertain Systems
- Pulsars and Gravitational Waves Research
- Gamma-ray bursts and supernovae
- Statistical Methods and Inference
- Optimization and Variational Analysis
Institut Polytechnique de Bordeaux
2016-2025
Institut de Mathématiques de Bordeaux
2016-2025
Université de Bordeaux
2011-2024
Institut national de recherche en informatique et en automatique
2012-2024
Centre de Recherche Inria Bordeaux - Sud-Ouest
2009-2021
European Space Agency
2018
Centre National de la Recherche Scientifique
2001-2017
McGill University
2012-2014
Swiss Federal Institute for Forest, Snow and Landscape Research
2004-2011
Groupe de Recherche en Économie Théorique et Appliquée
2005-2011
The Nuclear Spectroscopic Telescope Array (NuSTAR) mission, launched on 2012 June 13, is the first focusing high-energy X-ray telescope in orbit. NuSTAR operates band from 3 to 79 keV, extending sensitivity of far beyond ∼10 keV cutoff achieved by all previous satellites. inherently low background associated with concentrating light enables probe hard sky a more than 100-fold improvement over collimated or coded mask instruments that have operated this bandpass. Using its unprecedented...
We report the discovery of 3.76 s pulsations from a new burst source near Sgr A* observed by NuSTAR observatory. The strong signal SGR J1745−29 presents complex pulse profile modulated with pulsed fraction 27% ± 3% in 3–10 keV band. Two observations spaced nine days apart yield spin-down rate =(6.5 1.4) × 10−12. This implies magnetic field B = 1.6 1014 G, power =5 1033 erg s−1, and characteristic age P/2 =9 103 yr for rotating dipole model. However, current may be erratic, especially during...
This is the story of Jose Bove, radical French farmer who led a protest into town Millau and dismantled, to cheering crowds, new McDonalds. Now national hero in France, he has become leading figure global anti-capitalist protests, famed not only for his passion politics but also Roquefort cheese. For Bove struggle against multinational corporate industry, which been involved since 1968, what calls malbuffe, horrible nosh. In France good food have proved an irresistible mixture. examines...
The main goal of this paper is to establish some equivalence results on stability, recurrence, and ergodicity between a piecewise deterministic Markov process (PDMP) $\{X(t)\}$ an embedded discrete-time chain $\{\Theta_{n}\}$ generated by kernel G that can be explicitly characterized in terms the three local characteristics PDMP, leading tractable criterion results. First we important characterizing as sampling PDMP deriving connection probability first return time set for chains resolvent R...
We present the X-ray timing and spectral evolution of Galactic Center magnetar SGR J1745-2900 for first ~4 months post-discovery using data obtained with Nuclear Spectroscopic Telescope Array (NuSTAR)} Swift observatories. Our analysis reveals a large increase in spin-down rate by factor 2.60 +/- 0.07 over our span. further show that change spin was likely coincident bright burst observed 2013 June Swift, if so, there no accompanying discontinuity frequency. find source 3-10 keV flux has...
In this paper, we study discrete-time absorbing Markov Decision Processes (MDP) with measurable state space and Borel action a given initial distribution. For such models, solutions to the characteristic equation that are not occupation measures may exist. Several necessary sufficient conditions provided guarantee any solution is an measure. Under so-called continuity-compactness conditions, first show measure precisely if only it satisfies additional absolute continuity condition. Secondly,...
Abstract. The paper analyses and compares infrasonic seismic data from snow avalanches monitored at the Vallée de la Sionne test site in Switzerland 2009 to 2010. Using a combination of infrasound sensors, it is possible not only detect avalanche but also distinguish between different flow regimes analyse duration, average speed (for sections path) size. Different sensitiveness sensors shown. Furthermore, high amplitudes observed signal for one were modelled assuming that suspension layer...
A novel application of the interacting multiple models (IMM) algorithm in which passive infrared sensors are fused for tracking a target maneuvering three dimensions is discussed. More accurate motion proposed to improve performance. When general used describe periods, it shown that IMM behaviour not satisfactory, innovations associated with different do discriminate between corresponding regimes. The turning Markov chain transition matrix, i.e., priori information, then crucial obtaining...
We consider a Markov decision process (MDP) with constraints under the total expected discounted cost optimality criterion. are interested in proposing approximation methods of optimal value this constrained MDP. To end, starting from linear programming (LP) formulation MDP (on an infinite-dimensional space measures), we propose finite state LP problem. This is achieved by suitably approximating probability measure underlying random transitions dynamics system. Explicit convergence orders...
The stochastic model considered is a linear jump diffusion process X for which the coefficients and processes depend on Markov chain Z with finite state space. First, we study optimal filtering control problem these systems non-Gaussian initial conditions, given noisy observations of perfect measurements Z. We derive new sufficient condition ensures existence uniqueness solution nonlinear differential equations satisfied by output filter. quadratic show that separation principle holds. Next,...
Using the reference probability method and change of measure in discrete time, state estimator problem is considered for linear systems observed Gaussian noise when coefficients are functions a noisily observed, finite-state Markov chain. The methods new, finite-dimensional filters obtained. However, number statistics increases time. A numerical comparison this filter with interactive multiple model algorithm introduced by Blom Bar-Shalom (1988) given.
If the reliability community remains interested in dynamic theory, it is not really convinced by ability of already available approaches to treating current problems from within operational domain, even if methodological quality these undeniable. This paper keeping with two papers presented earlier conferences. Its aim show potentialities a method that combines high modelling capacity piecewise-deterministic processes great computing power inherent Monte Carlo simulation. has been applied...
We consider a discrete-time Markov decision process with Borel state and action spaces, possibly unbounded cost function. assume that the transition kernel is absolutely continuous respect to some probability measure . By replacing this its empirical distribution for sample of size n, we obtain finite space control problem, which used provide an approximation optimal value policy original model. impose Lipschitz continuity properties on model associated density functions. accuracy by means...
In this paper, we study a form of stability for general family nondiffusion Markov processes known in the literature as piecewise-deterministic process (PDMP). By here mean existence an invariant probability measure PDMP. It is shown that such equivalent to $\sigma$-finite kernel G linked resolvent operator U PDMP, satisfying boundedness condition or, equivalently, Radon--Nikodým derivative. Here generalize existing results [O. Costa, J. Appl. Prob., 27, (1990), pp. 60--73; M. Davis, Models...
We report new spectral and temporal observations of the magnetar 1E 1841-045 in Kes 73 supernova remnant obtained with Nuclear Spectroscopic Telescope Array (NuSTAR). Combined Swift archival XMM-Newton Chandra observations, phase-averaged spectrum is well characterized by a blackbody plus double power-law model, agreement previous, multi-mission X-ray results. However, we are unable to reproduce results reported using Suzaku observations. The pulsed fraction source found increase photon...
Mark H.A. Davis introduced the Piecewise-Deterministic Markov Process (PDMP) class of stochastic hybrid models in an article 1984. Today it is used to model a variety complex systems fields engineering, economics, management sciences, biology, Internet traffic, networks and many more. Yet, despite this, there very little way literature devoted development numerical methods for PDMDs solve problems practical importance, or computational control PDMPs. This book therefore presents collection...