An In-place Algorithm for Tensor Sum Combinations

Exel Caraballo
Advisor: Dr. Dorothy Bollman
Department of Mathematics
University of Puerto Rico at Mayagüez

Abstract

Tensor sum combinations play an important role in the design and modification of algorithm s for computing the vector of indices for a large family of permutations that include the so-called “digit-index” permutations. Although very fast methods for computing tensor sum com binations are known, they are not readily adaptable to implementations in an imperative language such a s FORTRAN or C because of the difficulties of efficiently managing a variable number of intermediate vectors. In this work we propose a solution to an important special case of this problem.

A Study on Interval Lagrange Interpolation: Numerical Exp eriments

Angelina Cardenas and Chenyi Hu
Department of Computer and Mathematical Sciences
University of Houston-Downtown
Houston, TX 77002

Abstract

Numerical Interpolation and approximation are powerful tools to solve real world applicati on problems. To take data measurement errors into considerations, we applied interval computations to Lagrange interpolation formula.

In this paper, numerical experiments on Lagrange interpolation with interval nodes and rig orous interval arithmetic are reported. An index on average over estimation due to interval arithme tic is defined. Some preliminary conclusions are presented as well.

Damage Affecting the Geometry of a Structure Modeled by Perturba tion Techniques

John Carrasquillo
Luis A. Godoy
Civil Infrastructure Research Center
and Civil Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000
John@rmce02.upr.clu.edu

Abstract

Damage in the form of shape deviation is considered in thin-walled structures. It is assum ed that the material remains elastic, and the displacements are linear with the load. A perturbation analysis is carried out, in which the displacements are expanded in terms of the maximum amplitude of the shape distortion. The formulation is written using finite element notation. An application to circular plates under in-plane loads is presented.

Experimental Development of Digital Systems using VHDL Technology

Edward David, Yamil Chaar, Pepin Torres
Advisor: Dr. Ramón E. Vásquez Espinosa
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000
s392625@sark.upr.clu.edu

Abstract

Among the several hardware design languages, the most widely used for fast prototyping is Very-Large Scale Integration Hardware Design Language (VHDL). VHDL is an industry standard and such a powerful tool would make a perfect addition to the electrical and compute r engineering curricula. As a starting point for the creation of a VHDL laboratory in this camp us, several digital systems were designed, among them an ASCII-to-Morse decoder, the interface of a mouse, and a parking-lot gate controller. These systems were completely drawn, compiled, simulated, and implemented on hardware using field programmable gate arrays.

Comparison of the Implementation of Heapsort in Wwk and C++

Javier A. Dastas Méndez, UIPR Arecibo Campus
Advisor: Alfredo Cruz, PhD, UIPR Arecibo Campus
Interamerican University of Puerto Rico
Arecibo Campus, Arecibo, Puerto Rico 00612
are7@ns.inter.edu

Abstract

The Awk programming language is an alternative for the simplified data analysis programmin g process. All-purpose analysis programs can be written faster than in a common programming language. The purpose of this paper is to distinguish the pros and cons running the HEAPSORT algorithm in Awk script programming versus machine-language compiled programming such as C++.

On the Implementation of Gabor Transforms in Multicomputer Netw orks

Vidya Manian and Ramón E. Vásquez
Computational Signal Processing Laboratory
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez, PR 00681-5000
manian@exodo.upr.clu.edu

Abstract

This paper presents the implementations of Gabor transform methods in parallel MIMD multicomputer networks. The discrete Gabor transform algorithms are based on two techniques. The first method computes the coefficients using the discrete Zak transform which can be implemented using Fast Fourier Transforms. The second method computes the Gabor coefficients based on an optimization criterion, which is the minimization of the difference vector between the original signal and the signal that is to be reconstructed from the coefficients. Some results of the parallel implementation of the algorithms in hypercube transputer networks is presented. The algorithms can be implemented in any MIMD multicomputer network.

GPS Based Remote Tracking Cellular System (GB/RTCS)

José R. Medina, Nelson Torres, Alfred Tarafa, Raymond Sánchez
Advisor: Prof. Eng. Othoniel Rodríguez, P.E.
Capstone Design
Electrical Engineering Department
Polytechnic University of Puerto Rico
Hato Rey, Puerto Rico 00919-2017
ee09662, ee12669, ee09887, ee12767@einstein.pupr.edu

Abstract

A Global Positioning System (GPS) provides for 24-hour position, time and velocity informa tion to a properly equipped user. GPS is founded by and controlled by the U.S. Department of Defense (DOD). Though they designed this system for Military use, now it is available for Civilian use. This paper describes the progress achieved so far toward development of a security locator system using this technology. The system will be capable of tracking and fin ding any moving objects to which it is attached.

Khoros Classification Algorithms applied to SAR images

Edwin J. Medina Rivera
Advisor: Dr. Ramón E. Vásquez Espinosa
Laboratory for Applied Remote Sensing and Image Processing
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000
wainder@exodo.upr.clu.edu

Abstract

Image preprocessing and analysis were applied to SAR images to improve and automate the recognition and classification of the different regions. Using KHOROS image application sofware, the process was performed in a data flow visual language. In this work, Weighted Minimum Distance, and Localized Receptive Field unsupervised classification algorithms implemented in KHOROS were applied to the SAR image to analyze the performance of them.

Methodology to Estimate Truck Factors for Flexible Pavements in Puerto Rico

Abraham Medina Vega
Advisors: Dr. Benjamín Colucci and Dr. Nazario D. Ramírez
Civil Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000

Abstract

A methodology to estimate Truck Factors (Tf) associated with different classification of t rucks in five homogeneous climatic zones for the primary highway system of Puerto Rico is presented .

Development of Computer Simulation for PNAME Controllers

Luis V. Meléndez González
Advisor: Dr. Gerson Beauchamp
Process Control and Instrumentation Laboratory
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico, 00681-5000

Abstract

The Proportional - Integral - Derivative (PNAME) controller has been used over the years for controlling many processes. The purpose of this work is to develop a simple yet accurate simulation of the action of a PNAME controller. This simulation can then be used for simulating a control process under PNAME control. The simulation for the controller will be implemented usin g Matlab's Simulink simulation software. The system will perform a simulation of the OMEGA CN9000A series PNAME Temperature Controller. The actual transfer function of the controller wil l be determined experimentally using the software LabVIEW and Tablecurve.

Kronecker’s Product and Efficient DIP Cycle Representatio ns

Angel Rivera
Advisor: Dr. Jaime Seguel
Advanced Computation Laboratory
Department of Mathematics
University of Puerto Rico at Mayagüez
Mayagüez, PR 00681
angel@rummat1.upr.clu.edu

Abstract

The major mathematical features leading to fast algorithms for computing the digit-index permutation cycle representations are outlined and illustrated through examples.

Some Experiments Measuring Error and Degree of Sparsity in Tre sholded Linear Solvers

Martin Sánchez
Advisor: Dr. Jaime Seguel
Advanced Computation Laboratory
Department of Mathematics
Univerisity of Puerto Rico at Mayagüez
Mayagüez, PR 00681
m-sanch@rummat1.upr.clu.edu

Abstract

A new trend in fast linear solver design uses image compression techniques to reduce the amount nonzero entries in the matrix representation of the system. This technique involves a change of basis which “concentrates” the non-null values of the matrix and a subseq uent elimination of all non-null entries whose absolute value is below a pre-established treshold. In this paper, experiments for the assessment of errors induced by tresholding versus sparsity g ains are carried out with a especially designed MatLab toolbox.

A Comparison of Sobel and Wavelet transform for Edge Detection

Rajeev Singh
Advisor: Dr. Ramón E. Vásquez
Laboratory For Applied Remote Sensing and Image Processing
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000
rajeev@exodo.upr.clu.edu

Abstract

This paper presents a comparison of traditional gradient based edge operator Sobel with ed ge detection using wavelets. Gradient based edge detectors have been found to be very sensitive to noise and on most of the occasion either they miss some of the edges or report false edges. T his gives need to find more innovative ways of detecting edges which forms a vital part of image analysis. Some experimental results using Daubechies wavelet filter are being presented.

On Tree-structured Wavelet Tansform for Texture analysis

Reena Singh
Advisor: Dr. Ramón E. Vásquez
Laboratory For Remote Sensing and Image Processing
Electrical and Computer Engineering Department
University of Puerto Rico, Mayagüez Campus
Mayagüez, Puerto Rico 00681-5000
reena@rmece01.upr.clu.edu

Abstract

Traditional statistical methods for texture analysis focus too much on the local coupling between image pixels. Recent time/frequency analytical tools such as the wavelet transform can effici ently characterize the coupling of different scales in textures, and help to overcome this difficul ty. However, pyramid-structured wavelet transform, which has a finer resolution in the lower frequency channels, does not work properly for textured images, since textures are quasi-periodic signals whose dominant frequencies are located in the middle frequency regions . This paper presents description and implementation of a tree-structured wavelet transform approach.

This page is maintained by CECORD. E-mail comments to webers@exodo.upr.clu.edu http:// Last modified on

CRC

Campus Home Page