Aceleración de algoritmos para imágenes hiperespectrales con OpenCL

En el presente trabajo se propone dar solución a uno de los problemas principales surgido en el campo del análisis de imágenes hiperespectrales. En las últimas décadas este campo está siendo muy activo, por lo que es de vital importancia tratar su problema principal: mezcla espectral. Muchos algoritmos han tratado de solucionar este problema, pero que a través de este trabajo se propone una cadena nueva de desmezclado en paralelo, para ser acelerados bajo el paradigma de programación paralela de OpenCl. Este paradigma nos aporta el modelo de programación unificada para acelerar algoritmos en sistemas heterogéneos. Podemos dividir el proceso de desmezclado espectral en tres etapas. La primera tiene la tarea de encontrar el número de píxeles puros, llamaremos endmembers a los píxeles formados por una única firma espectral, utilizaremos el algoritmo conocido como Geometry-based Estimation of number of endmembers, GENE. La segunda etapa se encarga de identificar los píxel endmembers y extraerlos junto con todas sus bandas espectrales, para esta etapa se utilizará el algoritmo conocido por Simplex Growing Algorithm, SGA. En la última etapa se crean los mapas de abundancia para cada uno de los endmembers encontrados, de esta etapa será encargado el algoritmo conocido por, Sum-to-one Constrained Linear Spectral Unmixing, SCLSU. Las plataformas utilizadas en este proyecto han sido tres: CPU, Intel Xeon E5-2695 v3, GPU, NVidia GeForce GTX 980, Acelerador, Intel Xeon Phi 31S1P. La idea de este proyecto se basa en realizar un análisis exhaustivo de los resultados obtenidos en las diferentes plataformas, con el fin de evaluar cuál se ajusta mejor a nuestras necesidades.

On increasing the effective aperture of antennas by data processing /

Includes bibliographical references (p. 58-59)

Calculus, with analytic geometry.

Includes index.

Neural network control of a gas turbine

In this paper we explore the practical use of neural networks for controlling complex non-linear systems. The system used to demonstrate this approach is a simulation of a gas turbine engine typical of those used to power commercial aircraft. The novelty of the work lies in the requirement for multiple controllers which are used to maintain system variables in safe operating regions as well as governing the engine thrust.

Rapid heuristic projection on simplicial cones

A very fast heuristic iterative method of projection on simplicial cones is presented. It consists in solving two linear systems at each step of the iteration. The extensive experiments indicate that the method furnishes the exact solution in more then 99.7 percent of the cases. The average number of steps is 5.67 (we have not found any examples which required more than 13 steps) and the relative number of steps with respect to the dimension decreases dramatically. Roughly speaking, for high enough dimensions the absolute number of steps is independent of the dimension.

Temporal finite elements in the dynamic analysis of structures

This thesis demonstrates that the use of finite elements need not be confined to space alone, but that they may also be used in the time domain, It is shown that finite element methods may be used successfully to obtain the response of systems to applied forces, including, for example, the accelerations in a tall structure subjected to an earthquake shock. It is further demonstrated that at least one of these methods may be considered to be a practical alternative to more usual methods of solution. A detailed investigation of the accuracy and stability of finite element solutions is included, and methods of applications to both single- and multi-degree of freedom systems are described. Solutions using two different temporal finite elements are compared with those obtained by conventional methods, and a comparison of computation times for the different methods is given. The application of finite element methods to distributed systems is described, using both separate discretizations in space and time, and a combined space-time discretization. The inclusion of both viscous and hysteretic damping is shown to add little to the difficulty of the solution. Temporal finite elements are also seen to be of considerable interest when applied to non-linear systems, both when the system parameters are time-dependent and also when they are functions of displacement. Solutions are given for many different examples, and the computer programs used for the finite element methods are included in an Appendix.

Regeneration limit of classical Shannon capacity

Since Shannon derived the seminal formula for the capacity of the additive linear white Gaussian noise channel, it has commonly been interpreted as the ultimate limit of error-free information transmission rate. However, the capacity above the corresponding linear channel limit can be achieved when noise is suppressed using nonlinear elements; that is, the regenerative function not available in linear systems. Regeneration is a fundamental concept that extends from biology to optical communications. All-optical regeneration of coherent signal has attracted particular attention. Surprisingly, the quantitative impact of regeneration on the Shannon capacity has remained unstudied. Here we propose a new method of designing regenerative transmission systems with capacity that is higher than the corresponding linear channel, and illustrate it by proposing application of the Fourier transform for efficient regeneration of multilevel multidimensional signals. The regenerative Shannon limit -the upper bound of regeneration efficiency -is derived. © 2014 Macmillan Publishers Limited. All rights reserved.

Examples Illustrating some Aspects of the Weak Deligne-Simpson Problem

Research partially supported by INTAS grant 97-1644

Motzkin Decomposable Sets

Theodore Motzkin proved, in 1936, that any polyhedral convex set can be expressed as the (Minkowski) sum of a polytope and a polyhedral convex cone. We have provided several characterizations of the larger class of closed convex sets, Motzkin decomposable, in finite dimensional Euclidean spaces which are the sum of a compact convex set with a closed convex cone. These characterizations involve different types of representations of closed convex sets as the support functions, dual cones and linear systems whose relationships are also analyzed. The obtaining of information about a given closed convex set F and the parametric linear optimization problem with feasible set F from each of its different representations, including the Motzkin decomposition, is also discussed. Another result establishes that a closed convex set is Motzkin decomposable if and only if the set of extreme points of its intersection with the linear subspace orthogonal to its lineality is bounded. We characterize the class of the extended functions whose epigraphs are Motzkin decomposable sets showing, in particular, that these functions attain their global minima when they are bounded from below. Calculus of Motzkin decomposable sets and functions is provided.

Stochastic Arithmetic Theory and Experiments

Stochastic arithmetic has been developed as a model for exact computing with imprecise data. Stochastic arithmetic provides confidence intervals for the numerical results and can be implemented in any existing numerical software by redefining types of the variables and overloading the operators on them. Here some properties of stochastic arithmetic are further investigated and applied to the computation of inner products and the solution to linear systems. Several numerical experiments are performed showing the efficiency of the proposed approach.

Sensor Location Problem for a Multigraph

MSC 2010: 05C50, 15A03, 15A06, 65K05, 90C08, 90C35

A semi-Lagrangian finite volume method for solving viscoelastic flow problems

Recently, there has been considerable interest in solving viscoelastic problems in 3D particularly with the improvement in modern computing power. In many applications the emphasis has been on economical algorithms which can cope with the extra complexity that the third dimension brings. Storage and computer time are of the essence. The advantage of the finite volume formulation is that a large amount of memory space is not required. Iterative methods rather than direct methods can be used to solve the resulting linear systems efficiently.

Low-power high-performance SAR ADC design with digital calibration techniques

This dissertation presents the design of three high-performance successive-approximation-register (SAR) analog-to-digital converters (ADCs) using distinct digital background calibration techniques under the framework of a generalized code-domain linear equalizer. These digital calibration techniques effectively and efficiently remove the static mismatch errors in the analog-to-digital (A/D) conversion. They enable aggressive scaling of the capacitive digital-to-analog converter (DAC), which also serves as sampling capacitor, to the kT/C limit. As a result, outstanding conversion linearity, high signal-to-noise ratio (SNR), high conversion speed, robustness, superb energy efficiency, and minimal chip-area are accomplished simultaneously. The first design is a 12-bit 22.5/45-MS/s SAR ADC in 0.13-μm CMOS process. It employs a perturbation-based calibration based on the superposition property of linear systems to digitally correct the capacitor mismatch error in the weighted DAC. With 3.0-mW power dissipation at a 1.2-V power supply and a 22.5-MS/s sample rate, it achieves a 71.1-dB signal-to-noise-plus-distortion ratio (SNDR), and a 94.6-dB spurious free dynamic range (SFDR). At Nyquist frequency, the conversion figure of merit (FoM) is 50.8 fJ/conversion step, the best FoM up to date (2010) for 12-bit ADCs. The SAR ADC core occupies 0.06 mm2, while the estimated area the calibration circuits is 0.03 mm2. The second proposed digital calibration technique is a bit-wise-correlation-based digital calibration. It utilizes the statistical independence of an injected pseudo-random signal and the input signal to correct the DAC mismatch in SAR ADCs. This idea is experimentally verified in a 12-bit 37-MS/s SAR ADC fabricated in 65-nm CMOS implemented by Pingli Huang. This prototype chip achieves a 70.23-dB peak SNDR and an 81.02-dB peak SFDR, while occupying 0.12-mm2 silicon area and dissipating 9.14 mW from a 1.2-V supply with the synthesized digital calibration circuits included. The third work is an 8-bit, 600-MS/s, 10-way time-interleaved SAR ADC array fabricated in 0.13-μm CMOS process. This work employs an adaptive digital equalization approach to calibrate both intra-channel nonlinearities and inter-channel mismatch errors. The prototype chip achieves 47.4-dB SNDR, 63.6-dB SFDR, less than 0.30-LSB differential nonlinearity (DNL), and less than 0.23-LSB integral nonlinearity (INL). The ADC array occupies an active area of 1.35 mm2 and dissipates 30.3 mW, including synthesized digital calibration circuits and an on-chip dual-loop delay-locked loop (DLL) for clock generation and synchronization.

A parallel numerical solver using hierarchically tiled arrays

Solving linear systems is an important problem for scientific computing. Exploiting parallelism is essential for solving complex systems, and this traditionally involves writing parallel algorithms on top of a library such as MPI. The SPIKE family of algorithms is one well-known example of a parallel solver for linear systems. The Hierarchically Tiled Array data type extends traditional data-parallel array operations with explicit tiling and allows programmers to directly manipulate tiles. The tiles of the HTA data type map naturally to the block nature of many numeric computations, including the SPIKE family of algorithms. The higher level of abstraction of the HTA enables the same program to be portable across different platforms. Current implementations target both shared-memory and distributed-memory models. In this thesis we present a proof-of-concept for portable linear solvers. We implement two algorithms from the SPIKE family using the HTA library. We show that our implementations of SPIKE exploit the abstractions provided by the HTA to produce a compact, clean code that can run on both shared-memory and distributed-memory models without modification. We discuss how we map the algorithms to HTA programs as well as examine their performance. We compare the performance of our HTA codes to comparable codes written in MPI as well as current state-of-the-art linear algebra routines.

MDP periodically time-varying convolutional codes

In this paper we use some classical ideas from linear systems theory to analyse convolutional codes. In particular, we exploit input-state-output representations of periodic linear systems to study periodically time-varying convolutional codes. In this preliminary work we focus on the column distance of these codes and derive explicit necessary and sufficient conditions for an (n, 2, 1) periodically time-varying convolutional code to have Maximum Distance Profile (MDP).