Research for Computational Methods

Syndicate content
Computing Mean Responses in Nonlinear Models
December 1, 1989
It is common in empirical studies using nonlinear models to estimate the mean response by evaluating the nonlinear response function at the mean value of its argument(s). However, this procedure conceptually is flawed if the response function has significant curvature in the neighborhood of the mean. Ideally, one should evaluate the estimated response function for each of the estimated responses. In general, there will be some nonzero approximation error if one instead simply evaluates the response function at the mean of the independent variable(s). Furthermore, if the variability is significant in the independent variable(s) of interest, the approximation error of using the 'evaluate at the mean' procedure increases. This paper examines the magnitude of the approximation error, and attempts to identify situations in which somewhat more computationally intensive procedures should be used.
Read More | Download Report
Methodology of an Event-Driven Monte Carlo Missile Simulation
November 1, 1989
This paper describes the methodology used by the Center for Naval Analyses to develop a Monte Carlo missile simulation that computes probabilities of target acquisition and distributions of missile arrival times. The simulation accounts for numerous missile and target uncertainties. Coordinated strikes are simulated by modeling multiple missiles from multiple launch points firing on a group of targets. Tactical applications of the simulation required that it be somputationally efficient. This led to an event-driven time advancement scheme.
Read More | Download Report
Small and Great Circle Routines
July 1, 1989
This information manual documents routines in the CNA mapping software library that calculate the points of small and great circles on the earth. The equations upon which the routines are based are described and a user's walk-through of the programs is given.
Read More | Download Report
A Stable Curvilinear Alternative to Linear Equating
October 1, 1987
In any ongoing testing program, new forms of a test are developed and equated to an earlier form. Linear equating is often used when the new form is nearly parallel to the old one, but it can lead to substantial systematic errors. This research contribution proposes and evaluates a new method for test equating. The method combines the stability of linear equating and small bias of equipercentile equating. See also 02 057100.00
Read More | Download Report
Trends in C-School Skill Training
June 1, 1987
This research memorandum assesses the impact of changes in endstrength, force mix, technology, and other factors on specialized skill training load. It focuses on C-school training because that area has experienced the most growth. This work relates training to the number and types of personnel in the inventory and to manpower requirements.
Read More | Download Report
Total Force Enlistment Programs Simulation - Volume I
June 1, 1987
The structure and development of the Total Force Enlistment Program Simulation is described in this Research Memorandum. The simulation analyzes changes in inventories of enlisted active-duty and reserve personnel associated with changes in the number of active-duty accessions and their distribution among enlistment programs. The simulation is based on continuation and affiliation behavior of individuals in six enlistment programs as observed in recent years. Total accessions and their distribution among enlistment programs and ratings are inputs to the simulation. The effects of changing these inputs are simulated by comparing the future inventories associated with two different accession profiles.
Read More | Download Report
Total Force Enlistment Programs Simulation - Volume II
June 1, 1987
This Research Memorandum presents technical appendixes to supplement the Total Force Enlistment Programs Simulation. It is divided into four sections. Appendix A contains a guide to assist users of the simulation methodology. In addition to a detailed example of the use of the simulation, it discusses some of the potential uses not explored in Volume I. Appendix B is a description of the methods used in computing the active-duty inventories and continuation behavior. It also lists some of the important active-duty data elements used in the simulation, as well as the major computer programs used to generate these data. Appendix C presents similar descriptions for the SELRES data. A flowchart of the simulation methodology and annotated versions of the simulation programs appear in Appendix D.
Read More | Download Report
Continuous Dependence on Modeling in the Cauchy Problem for Nonlinear Elliptic Equations
April 1, 1987
The Cauchy problem for various types of second order nonlinear elliptic equations is considered. A substitution in the equation leads to a perturbed equation whose solution of an unperturbed second order linear elliptic equation obtained by formally setting Epsilon to zero. In each case a logarithmic convexity argument is used to show that appropriately constrained solutions of the original equation (assumed to exist) are shown to differ from a solution of the associated linear equation in a manner depending continuously on the parameter Epsilon.
Read More | Download Report
Calculating the Year-to-Year Size and Experience Level of a Changing Group
February 1, 1987
CNA's Total Force Utilization Study examines alternative ways to man the 600-Ship Navy. Some of the important issues involve questions about accessions, retention, and the evolution of groups-such as ratings-in terms of the numbers and experience levels of their members. Consequently, the study developed a model that relates these factors quantitatively. This research memorandum describes the model and illustrates its potential uses with a few examples. The focus is on documentation of methodology, not on specific applications.
Read More | Download Report
Confidence Interval for Parameter n in a Binomial Distribution
January 1, 1987
A simple procedure to approximate a confidence interval for the parameter n in a binomial distribution is presented in this research memorandum. A simulation procedure to verify the coverage of confidence intervals is presented in appendix A. An interactive computer program is included in appendix B. The program is written in the FORTRAN language, which is readily available in most computing environments. Tables with 90 percent and 95 percent confidence coefficients are included in appendix C.
Read More | Download Report