Terry Feagin

Permanent URI for this collectionhttps://hdl.handle.net/10657.1/2287


Dr. Terry Feaginis is a Professor of Computer Science at University of Houston-Clear Lake. His research interests are in the areas of Numerical Methods, Parallel Algorithms, Artificial Intelligence, Machine Learning, Scientific Visualization, Fault Management, Research Methods.


Recent Submissions

Now showing 1 - 10 of 10
  • Item
    A Tutorial on CLIPsTOOL, a Graphical Interface to CLIPS
    (Third CLIPS Conference Proceedings, 1994-09-14) Feagin, Terry
  • Item
    GOLDS, a Blackboard System for Fault Diagnosis
    (Third CLIPS Conference Proceedings, 1994-09-14) Feagin, Terry
  • Item
    A Tenth-Order Runge-Kutta Method with Error Estimate
    (Proceedings of the IAENG Conf. on Scientific Computing, 2007) Feagin, Terry
    A tenth-order explicit Runge-Kutta method with embedded results of order eight is exhibited. The difference between the results of orders eight and ten can be used to estimate the local truncation error and thus to vary the stepsize. Numerical experiments demonstrate that the method compares favorably with other high-order embedded methods.
  • Item
    Edge effects in lacunarity analysis
    (Ecological Modelling, 2007) Feagin, Terry
    Lacunarity results can be skewed by edge effects and this may have negative implications for research projects in ecological pattern analysis and modeling. The problem occurs because the standard gliding-box algorithm over-samples the center of a map and under-samples along its edges. This effect is particularly strong when the scale of inquiry is large relative to the extent of the map, as fewer box mass estimates are utilized to form the distribution from which lacunarity is calculated. We devised a new algorithm where we allowed the gliding-box to overlap beyond the edge of the map and wrap back around to the opposing side, thereby solving both problems. In this study, we compare the standard lacunarity algorithm with this new periodic boundary algorithm (a method often used in cellular automata modeling) to quantify the differences between the two approaches and to determine when the standard algorithm may suffer from deleterious effects. We performed our analysis upon several neutral landscapes to evaluate the importance of pattern as well. We found that the standard algorithm skews results when the pattern is strongly heterogeneous or aggregated, especially when the classes are not evenly distributed around the center of the map or when percent cover pi is low. The advantages and disadvantages of both algorithms, as well as other potential remedies such as up-weighting samples along the edges, are discussed.
  • Item
    An Explicit Runge-Kutta Method of Order Fourteen
    (Numerical Algorithms, 2009) Feagin, Terry
  • Item
    High-Order m-symmetric Runge-Kutta Methods
    (Proceeding of the 23rd Biennial Conference on Numerical Analysis, 2009-06) Feagin, Terry
  • Item
    An Implicit Runge-Kutta Method for Perturbed Ordinary Differential Equations
    (Proceedings of Neural, Parallel & Scientific Computations, 2010) Feagin, Terry
    A novel approach is developed for computing the numerical solution of perturbed ordinary differential equations using implicit Runge-Kutta methods. At each step of the integration of the differential equations, the method uses an enhanced initial approximation for each stage and then solves the implicit equations using a modified Picard iteration. The new approach offers greater efficiency for computing numerical solutions to perturbed initial value problems.
  • Item
    High-order Explicit Runge-Kutta Methods Using M-Symmetry
    (Neural, Parallel & Scientific Computations, 2012-12) Feagin, Terry
    The Runge-Kutta equations of condition are reformulated. The concept of m-symmetry is defined. It is shown that any m-symmetric method is of order m. The equations of condition for a twelfth-order explicit Runge-Kutta method with twenty-five stages are solved using m-symmetry. The method contains an embedded tenth-order method that can be used to estimate the local truncation errors and thus to vary the stepsize. Numerical experiments demonstrate that the method compares favorably with other high-order methods, especially for those problems requiring highly accurate solutions.
  • Item
    The efficient solution of Kepler’s equation using a quartic approximation and rational functions
    (Neural, Parallel & Scientific Computations, 2016-12) Feagin, Terry