Models and algorithms for tracking of maneuvering objects using variable rate particle filters

Standard algorithms in tracking and other state-space models assume identical and synchronous sampling rates for the state and measurement processes. However, real trajectories of objects are typically characterized by prolonged smooth sections, with sharp, but infrequent, changes. Thus, a more parsimonious representation of a target trajectory may be obtained by direct modeling of maneuver times in the state process, independently from the observation times. This is achieved by assuming the state arrival times to follow a random process, typically specified as Markovian, so that state points may be allocated along the trajectory according to the degree of variation observed. The resulting variable dimension state inference problem is solved by developing an efficient variable rate particle filtering algorithm to recursively update the posterior distribution of the state sequence as new data becomes available. The methodology is quite general and can be applied across many models where dynamic model uncertainty occurs on-line. Specific models are proposed for the dynamics of a moving object under internal forcing, expressed in terms of the intrinsic dynamics of the object. The performance of the algorithms with these dynamical models is demonstrated on several challenging maneuvering target tracking problems in clutter. © 2006 IEEE.

Heterogeneous reconfigurable system for adaptive particle filters in real-time applications

This paper presents a heterogeneous reconfigurable system for real-time applications applying particle filters. The system consists of an FPGA and a multi-threaded CPU. We propose a method to adapt the number of particles dynamically and utilise the run-time reconfigurability of the FPGA for reduced power and energy consumption. An application is developed which involves simultaneous mobile robot localisation and people tracking. It shows that the proposed adaptive particle filter can reduce up to 99% of computation time. Using run-time reconfiguration, we achieve 34% reduction in idle power and save 26-34% of system energy. Our proposed system is up to 7.39 times faster and 3.65 times more energy efficient than the Intel Xeon X5650 CPU with 12 threads, and 1.3 times faster and 2.13 times more energy efficient than an NVIDIA Tesla C2070 GPU. © 2013 Springer-Verlag.

Interactive multi-objective particle swarm optimisation using decision space interaction

The most common approach to decision making in multi-objective optimisation with metaheuristics is a posteriori preference articulation. Increased model complexity and a gradual increase of optimisation problems with three or more objectives have revived an interest in progressively interactive decision making, where a human decision maker interacts with the algorithm at regular intervals. This paper presents an interactive approach to multi-objective particle swarm optimisation (MOPSO) using a novel technique to preference articulation based on decision space interaction and visual preference articulation. The approach is tested on a 2D aerofoil design case study and comparisons are drawn to non-interactive MOPSO. © 2013 IEEE.

Evaluation of holonomic quantum computation: Adiabatic versus nonadiabatic

Based on the analytical solution to the time-dependent Schrodinger equations, we evaluate the holonomic quantum computation beyond the adiabatic limit. Besides providing rigorous confirmation of the geometrical prediction of holonomies, the present dynamical resolution offers also a practical means to study the nonadiabaticity induced effects for the universal qubit operations.

2-DELTA-FUNCTION GENERALIZED PLASMA POLE MODEL FOR 1ST PRINCIPLE CALCULATIONS OF QUASI-PARTICLE ENERGIES IN MANY-ELECTRON SYSTEMS

To evaluate the dynamical effects of the screened interaction in the calculations of quasiparticle energies in many-electron systems a two-delta-function generalized plasma pole model (GPP) is introduced to simulate the dynamical dielectric function. The usual single delta-function GPP model has the drawback of over simplifications and for the crystals without the center of symmetry is inappropriate to describe the finite frequency behavior for dielectric function matrices. The discrete frequency summation method requires too much computation to achieve converged results since ab initio calculations of dielectric function matrices are to be carried out for many different frequencies. The two-delta GPP model is an optimization of the two approaches. We analyze the two-delta GPP model and propose a method to determine from the first principle calculations the amplitudes and effective frequencies of these delta-functions. Analytical solutions are found for the second order equations for the parameter matrices entering the model. This enables realistic applications of the method to the first principle quasiparticle calculations and makes the calculations truly adjustable parameter free.

Feature Selection based on Rough Sets and Particle Swarm Optimization

X. Wang, J. Yang, X. Teng, W. Xia, and R. Jensen. Feature Selection based on Rough Sets and Particle Swarm Optimization. Pattern Recognition Letters, vol. 28, no. 4, pp. 459-471, 2007.

Solving the advection-dispersion equation using the discrete puff particle method

A flexible, mass-conservative numerical technique for solving the advection-dispersion equation for miscible contaminant transport is presented. The method combines features of puff transport models from air pollution studies with features from the random walk particle method used in water resources studies, providing a deterministic time-marching algorithm which is independent of the grid Peclet number and scales from one to higher dimensions simply. The concentration field is discretised into a number of particles, each of which is treated as a point release which advects and disperses over the time interval. The dispersed puff is itself discretised into a spatial distribution of particles whose masses can be pre-calculated. Concentration within the simulation domain is then calculated from the mass distribution as an average over some small volume. Comparison with analytical solutions for a one-dimensional fixed-duration concentration pulse and for two-dimensional transport in an axisymmetric flow field indicate that the algorithm performs well. For a given level of accuracy the new method has lower computation times than the random walk particle method.

Two-stage RBF network construction based on particle swarm optimization

The conventional radial basis function (RBF) network optimization methods, such as orthogonal least squares or the two-stage selection, can produce a sparse network with satisfactory generalization capability. However, the RBF width, as a nonlinear parameter in the network, is not easy to determine. In the aforementioned methods, the width is always pre-determined, either by trial-and-error, or generated randomly. Furthermore, all hidden nodes share the same RBF width. This will inevitably reduce the network performance, and more RBF centres may then be needed to meet a desired modelling specification. In this paper we investigate a new two-stage construction algorithm for RBF networks. It utilizes the particle swarm optimization method to search for the optimal RBF centres and their associated widths. Although the new method needs more computation than conventional approaches, it can greatly reduce the model size and improve model generalization performance. The effectiveness of the proposed technique is confirmed by two numerical simulation examples.

COMPUTATION OF SYNTHETIC SPECTRA FROM SIMULATIONS OF RELATIVISTIC SHOCKS

Particle-in-cell (PIC) simulations of relativistic shocks are in principle capable of predicting the spectra of photons that are radiated incoherently by the accelerated particles. The most direct method evaluates the spectrum using the fields given by the Lienard-Wiechart potentials. However, for relativistic particles this procedure is computationally expensive. Here we present an alternative method that uses the concept of the photon formation length. The algorithm is suitable for evaluating spectra both from particles moving in a specific realization of a turbulent electromagnetic field or from trajectories given as a finite, discrete time series by a PIC simulation. The main advantage of the method is that it identifies the intrinsic spectral features and filters out those that are artifacts of the limited time resolution and finite duration of input trajectories.

Fabrication and characterization of silicon position sensitive particle detectors

Position sensitive particle detectors are needed in high energy physics research. This thesis describes the development of fabrication processes and characterization techniques of silicon microstrip detectors used in the work for searching elementary particles in the European center for nuclear research, CERN. The detectors give an electrical signal along the particles trajectory after a collision in the particle accelerator. The trajectories give information about the nature of the particle in the struggle to reveal the structure of the matter and the universe. Detectors made of semiconductors have a better position resolution than conventional wire chamber detectors. Silicon semiconductor is overwhelmingly used as a detector material because of its cheapness and standard usage in integrated circuit industry. After a short spread sheet analysis of the basic building block of radiation detectors, the pn junction, the operation of a silicon radiation detector is discussed in general. The microstrip detector is then introduced and the detailed structure of a double-sided ac-coupled strip detector revealed. The fabrication aspects of strip detectors are discussedstarting from the process development and general principles ending up to the description of the double-sided ac-coupled strip detector process. Recombination and generation lifetime measurements in radiation detectors are discussed shortly. The results of electrical tests, ie. measuring the leakage currents and bias resistors, are displayed. The beam test setups and the results, the signal to noise ratio and the position accuracy, are then described. It was found out in earlier research that a heavy irradiation changes the properties of radiation detectors dramatically. A scanning electron microscope method was developed to measure the electric potential and field inside irradiated detectorsto see how a high radiation fluence changes them. The method and the most important results are discussed shortly.

Une nouvelle méthode smoothed particle hydrodynamics : simulation des interfaces immergées et de la dynamique Brownienne des molécules avec des interactions hydrodynamiques

Dans cette thèse, nous présentons une nouvelle méthode smoothed particle hydrodynamics (SPH) pour la résolution des équations de Navier-Stokes incompressibles, même en présence des forces singulières. Les termes de sources singulières sont traités d'une manière similaire à celle que l'on retrouve dans la méthode Immersed Boundary (IB) de Peskin (2002) ou de la méthode régularisée de Stokeslets (Cortez, 2001). Dans notre schéma numérique, nous mettons en oeuvre une méthode de projection sans pression de second ordre inspirée de Kim et Moin (1985). Ce schéma évite complètement les difficultés qui peuvent être rencontrées avec la prescription des conditions aux frontières de Neumann sur la pression. Nous présentons deux variantes de cette approche: l'une, Lagrangienne, qui est communément utilisée et l'autre, Eulerienne, car nous considérons simplement que les particules SPH sont des points de quadrature où les propriétés du fluide sont calculées, donc, ces points peuvent être laissés fixes dans le temps. Notre méthode SPH est d'abord testée à la résolution du problème de Poiseuille bidimensionnel entre deux plaques infinies et nous effectuons une analyse détaillée de l'erreur des calculs. Pour ce problème, les résultats sont similaires autant lorsque les particules SPH sont libres de se déplacer que lorsqu'elles sont fixes. Nous traitons, par ailleurs, du problème de la dynamique d'une membrane immergée dans un fluide visqueux et incompressible avec notre méthode SPH. La membrane est représentée par une spline cubique le long de laquelle la tension présente dans la membrane est calculée et transmise au fluide environnant. Les équations de Navier-Stokes, avec une force singulière issue de la membrane sont ensuite résolues pour déterminer la vitesse du fluide dans lequel est immergée la membrane. La vitesse du fluide, ainsi obtenue, est interpolée sur l'interface, afin de déterminer son déplacement. Nous discutons des avantages à maintenir les particules SPH fixes au lieu de les laisser libres de se déplacer. Nous appliquons ensuite notre méthode SPH à la simulation des écoulements confinés des solutions de polymères non dilués avec une interaction hydrodynamique et des forces d'exclusion de volume. Le point de départ de l'algorithme est le système couplé des équations de Langevin pour les polymères et le solvant (CLEPS) (voir par exemple Oono et Freed (1981) et Öttinger et Rabin (1989)) décrivant, dans le cas présent, les dynamiques microscopiques d'une solution de polymère en écoulement avec une représentation bille-ressort des macromolécules. Des tests numériques de certains écoulements dans des canaux bidimensionnels révèlent que l'utilisation de la méthode de projection d'ordre deux couplée à des points de quadrature SPH fixes conduit à un ordre de convergence de la vitesse qui est de deux et à une convergence d'ordre sensiblement égale à deux pour la pression, pourvu que la solution soit suffisamment lisse. Dans le cas des calculs à grandes échelles pour les altères et pour les chaînes de bille-ressort, un choix approprié du nombre de particules SPH en fonction du nombre des billes N permet, en l'absence des forces d'exclusion de volume, de montrer que le coût de notre algorithme est d'ordre O(N). Enfin, nous amorçons des calculs tridimensionnels avec notre modèle SPH. Dans cette optique, nous résolvons le problème de l'écoulement de Poiseuille tridimensionnel entre deux plaques parallèles infinies et le problème de l'écoulement de Poiseuille dans une conduite rectangulaire infiniment longue. De plus, nous simulons en dimension trois des écoulements confinés entre deux plaques infinies des solutions de polymères non diluées avec une interaction hydrodynamique et des forces d'exclusion de volume.

Nonlinear system identification using particle swarm optimisation tuned radial basis function models

A novel particle swarm optimisation (PSO) tuned radial basis function (RBF) network model is proposed for identification of non-linear systems. At each stage of orthogonal forward regression (OFR) model construction process, PSO is adopted to tune one RBF unit's centre vector and diagonal covariance matrix by minimising the leave-one-out (LOO) mean square error (MSE). This PSO aided OFR automatically determines how many tunable RBF nodes are sufficient for modelling. Compared with the-state-of-the-art local regularisation assisted orthogonal least squares algorithm based on the LOO MSE criterion for constructing fixed-node RBF network models, the PSO tuned RBF model construction produces more parsimonious RBF models with better generalisation performance and is often more efficient in model construction. The effectiveness of the proposed PSO aided OFR algorithm for constructing tunable node RBF models is demonstrated using three real data sets.

Particle swarm optimization aided orthogonal forward regression for unified data modeling

We propose a unified data modeling approach that is equally applicable to supervised regression and classification applications, as well as to unsupervised probability density function estimation. A particle swarm optimization (PSO) aided orthogonal forward regression (OFR) algorithm based on leave-one-out (LOO) criteria is developed to construct parsimonious radial basis function (RBF) networks with tunable nodes. Each stage of the construction process determines the center vector and diagonal covariance matrix of one RBF node by minimizing the LOO statistics. For regression applications, the LOO criterion is chosen to be the LOO mean square error, while the LOO misclassification rate is adopted in two-class classification applications. By adopting the Parzen window estimate as the desired response, the unsupervised density estimation problem is transformed into a constrained regression problem. This PSO aided OFR algorithm for tunable-node RBF networks is capable of constructing very parsimonious RBF models that generalize well, and our analysis and experimental results demonstrate that the algorithm is computationally even simpler than the efficient regularization assisted orthogonal least square algorithm based on LOO criteria for selecting fixed-node RBF models. Another significant advantage of the proposed learning procedure is that it does not have learning hyperparameters that have to be tuned using costly cross validation. The effectiveness of the proposed PSO aided OFR construction procedure is illustrated using several examples taken from regression and classification, as well as density estimation applications.

Transport in the low-latitude tropopause zone diagnosed using particle trajectories

Recent literature has described a “transition zone” between the average top of deep convection in the Tropics and the stratosphere. Here transport across this zone is investigated using an offline trajectory model. Particles were advected by the resolved winds from the European Centre for Medium-Range Weather Forecasts reanalyses. For each boreal winter clusters of particles were released in the upper troposphere over the four main regions of tropical deep convection (Indonesia, central Pacific, South America, and Africa). Most particles remain in the troposphere, descending on average for every cluster. The horizontal components of 5-day trajectories are strongly influenced by the El Niño–Southern Oscillation (ENSO), but the Lagrangian average descent does not have a clear ENSO signature. Tropopause crossing locations are first identified by recording events when trajectories from the same release regions cross the World Meteorological Organization lapse rate tropopause. Most crossing events occur 5–15 days after release, and 30-day trajectories are sufficiently long to estimate crossing number densities. In a further two experiments slight excursions across the lapse rate tropopause are differentiated from the drift deeper into the stratosphere by defining the “tropopause zone” as a layer bounded by the average potential temperature of the lapse rate tropopause and the profile temperature minimum. Transport upward across this zone is studied using forward trajectories released from the lower bound and back trajectories arriving at the upper bound. Histograms of particle potential temperature (θ) show marked differences between the transition zone, where there is a slow spread in θ values about a peak that shifts slowly upward, and the troposphere below 350 K. There forward trajectories experience slow radiative cooling interspersed with bursts of convective heating resulting in a well-mixed distribution. In contrast θ histograms for back trajectories arriving in the stratosphere have two distinct peaks just above 300 and 350 K, indicating the sharp change from rapid convective heating in the well-mixed troposphere to slow ascent in the transition zone. Although trajectories slowly cross the tropopause zone throughout the Tropics, all three experiments show that most trajectories reaching the stratosphere from the lower troposphere within 30 days do so over the west Pacific warm pool. This preferred location moves about 30°–50° farther east in an El Niño year (1982/83) and about 30° farther west in a La Niña year (1988/89). These results could have important implications for upper-troposphere–lower-stratosphere pollution and chemistry studies.

Tightened Lieb-Oxford Bound for Systems of Fixed Particle Number

The Lieb-Oxford bound is a constraint upon approximate exchange-correlation functionals. We explore a nonempirical tightening of that bound in both universal and electron number-dependent form. The test functional is PBE. Regarding both atomization energies (slightly worsened) and bond lengths (slightly improved), we find the PBE functional to be remarkably insensitive to the value of the Lieb-Oxford bound. This both rationalizes the use of the original Lieb-Oxford constant in PBE and suggests that enhancement factors more sensitive to sharpened constraints await discovery.