Mario Paz

An analysis of a hybrid neural network and pattern recognition technique for predicting short-term increases in the NYSE composite index

We introduce a method for combining template matching, from pattern recognition, and the feed-forward neural network, from artificial intelligence, to forecast stock market activity. We evaluate the effectiveness of the method for forecasting increases in the New York Stock Exchange Composite Index at a 5 trading day horizon. Results indicate that the technique is capable of returning results that are superior to those attained by random choice.

International Building Code IBC-2000

The IBC-2000 is similar to the UBC-97, but it contains some significant differences. An important difference is that the IBC-2000 includes a set of maps to obtain seismic response spectral values that will result in the same level of risk at any given geographic location in the United States. This code also introduces the concept of Seismic Use Group, which is somewhat analogous to the Importance Factor in the UBC-97. In addition, the IBC-2000 classifies every building in a Seismic Design Category which determines the analysis procedure to be used, the maximum allowed height and drift limitations.

Computer determination of the shear center of open and closed sections

Abstract A computer program is developed for the determination of properties of open and closed section thin-walled areas of rectangular segments. A brief recapitulation of properties of plane areas including the determination of the shear center of open and multi-cellular closed sections is presented.

Mathematical observations in structural dynamics

Abstract In dynamic structural analysis, the basic relations between forces and displacements for a beam element subjected to axial, torsional or flexural vibration are obtained either by solving the appropriate equation of motion or by using an approximate method. The exact equation leads to the dynamic stiffness matrix while the approximate method results in the superposition of elastic and inertial forces represented respectively by the stiffness and mass matrices. The common procedure in finding the natural frequencies is to set the determinant of the dynamic stiffness matrix for the system equal to zero. The approximate method leads to an eigenvalue type problem while the exact method results in a transcendental equation of trigonometric and hyperbolic functions. The natural frequencies in a region of interest are found by a systematic search in the determinntal function. The purpose of this paper is to show that the search technique cannot be applied for certain values of the argument at which the determinantal function is not defined. It is proved that the natural frequencies of any isolated member in the system are critical values for the determinantal function. A practical method is given to obviate the difficulty in order to find the natural frequencies from the determinant, including the critical values at which the dynamic stiffness matrix is not defined. Also, as part of this investigation, the mathematical relation is established between the dynamic stiffness matrix derived by the approximate finite element method and the results obtained from the exact Bernoulli-Euler equation for flexural vibration or the wave equation for axial or torsional vibration.

Nonlinear Structural Response

In discussing the dynamic behavior of single-degree-of-freedom systems, we assumed that in the model representing the structure, the restoring force was proportional to the displacement. We also assumed the dissipation of energy through a viscous damping mechanism in which the damping force was proportional to the velocity. In addition, the mass in the model was always considered to be unchanging with time. As a consequence of these assumptions, the equation of motion for such a system resulted in a linear, second order ordinary differential equation with constant coefficients, namely,

Static Condensation and Substructuring

m\ddot{u} + c\dot{u} + ku = F(t)

Progress report: improving the stock price forecasting performance of the bull flag heuristic with genetic algorithms and neural networks

(6.1) In the previous chapters it was illustrated that for particular forcing functions such as harmonic functions, it was relatively simple to solve this equation (6.1) and that a general solution always existed in terms of Duhamel’s integral. Equation (6.1) thus represents the dynamic behavior of many structures modeled as a single-degree-of-freedom system. There are, however, physical situations for which this linear model does not adequately represent the dynamic characteristics of the structure. The analysis in such cases requires the introduction of a model in which the spring force or the damping force may not remain proportional, respectively, to the displacement or to the velocity. Consequently, the resulting equation of motion will no longer be linear and its mathematical solution, in general, will have a much greater complexity, often requiring a numerical procedure for its integration.

Reduction of Dynamic Matrices

Abstract This paper presents a simple yet general procedure to determine influence lines and surfaces for frames, beams, trusses, and plates. The method is based on the application of the Muller-Breslau principle and finite elements. It can be easily implemented with most commercial finite element codes.