Mathematics, Computer Science and Statistics.

Chair: Ravinder Kumar, Alcorn State University 

Vice-chair: Robert S. Fritzius, Shady Tree Physics

THURSDAY MORNING

Ship Isle

9:15 Divisional Poster Session

THE ASSOCIATION BETWEEN BODY MASS INDEX AND MORTALITY IN HISPANICS

LaTrese Davis*, Juan Gallegos, Beverly Gonzales, and Che Smith, Alcorn State University, Alcorn State, MS 39096; University of Houston-Downtown, TX; University of Illinois at Urbana-Champagne, IL; and Spelman College, GA

Recent research indicate that obesity is an epidemic in the United States; its prevalence continuously increases across all age groups, races, education levels, and lifestyles, causing much concern about the risks of many chronic diseases with which obesity is associated (Mokdad, 1999). It is evident that different ethnic groups have unique body compositions; thus, it is important to investigate all aspects and implications of adiposity for many ethnic groups rather than a select few. Currently, there are countless publications that have compared measures of obesity for blacks and whites, with little to no information pertaining to Hispanics. In our study, we consider data such as BMI, age, and mortality of the 51,945 Hispanic participants in the National Health Interview Survey, conducted from 1985-1994 by the National Center for Health Statistics. Along with simple data filtration and manipulation, we use logistic regression and Cox proportional hazard, which are widely used in epidemiological research. Our findings contradicted the previous findings about the American society. For non-smokers, the thinnest group had the highest mortality rates, whereas smokers had the lowest mortality rates in the obese groups. This analysis examines association between BMI and mortality, along with the combined relationship of age, smoking status, and BMI to mortality and causes of death.

Divisional Talks

9:30 LOOPING EFFECT ON KINDERGARTEN AND FIRST GRADERS

Virginia J. Moore and Sam S. Gordji*, University of Mississippi, University, MS 38677

The purpose of this study was to examine the impact of "looping" on kindergarten and first grade students' achievement in reading. Looping is when groups of students stay with the same instructor at least two years and possibly longer. Looping is practiced in Japan and several parts of USA. In this study subjects (kindergarten students) were from New Albany, Mississippi. Subjects were tested during August when they first entered New Albany School District and in May of the following year. The subjects were administered the Marie Clay Reading Observation Survey prior to and after the study period. Both t-test and Analysis of Variance were performed on both groups individually. That is t-tests were performed only on looping group with only pre-test and post-test for one group, these analyses were performed for both groups. The pre-test and posttest for both groups (looping and non-looping) showed a significant difference. To account for this difference, the Analysis of Covariance was performed. The result indicated that at least for this group, looping did not improve students' test score. Additionally, descriptive statistical information was obtained from the parents and students in the second year of the study when students were near the end of first grade. Results of the surveys revealed that generally all parents felt the looping process was a positive experience for their child and they believed the looping process benefited them emotionally and socially. Students also enjoyed the familiar surrounding (same teacher and classmates). This is the dissertation topic for the senior author.

9:50 EVALUATING BIAS OF THE KAPLAN-MEIER ESTIMATOR FOR THE SMALL-SAMPLE CASE

Karl Calderon, Tollie Thigpen III*, Sarah Williams, and Javier Rogo, University of Arizona, Tucson, AZ; Alcorn State University Alcorn State, MS 39096; University of Colorado, Denver, CO; and Rice University, Houston, TX

A study of the bias of the Kaplan-Meier Estimator was performed in the case of the small-sample. Research was carried out through simulation study in the interest of a controllable environment. This bias was investigated in the dependence of the censoring distribution. It is shown that the bias of the estimator remains similar as the survival and censoring distributions are derived from independent and identical distributions. It is also determined that the amount of bias does vary when stochastic order of the censoring distribution is increased and decreased. This research was done at Rice University in the summer of 2003.

10:10 INFORMATION REENGINEERING FOR LIBRARY ACQUISITION: A CASE STUDY

Jessie Arnold, Bobbie Fells, Eric Speas, and Lixin Yu*, Alcorn State University, Alcorn State, MS 39360

The university library has a bottleneck in its acquisition process checking a long list of target books against the library catalog on a one-to-one basis to avoid duplication in purchase. This is time-consuming work even though the library has been computerized for years. This project explores how to automate this process and finish the job in a minute. The project starts with examining the data flow of the acquisition process. The procedure is redesigned taking into consideration the capability of the information technology. Programs are developed to make the duplication checking a batch process. Some other features are implemented to reduce the time-consuming routine work of the staff. The experience of the project can be applied to a wide variety of applications. It demonstrates how information technology is used to improve not only the efficiency but also the effectiveness. It also shows how the information technology department can collaborate with other departments in an organization.

10:30 Break

10:40 COMPUTER LAB MANAGEMENT: A COMPARATIVE STUDY OF METHODS TO KEEP LAB COMPUTERS RUNNING WITH PEAK PERFORMANCE

Joseph Mwangi and Lixin Yu*, Alcorn State University, Alcorn State, MS 39360

One of the problems with computer labs is users' changing system configuration. System files may be deleted, virus may be introduced, and application configurations may be reset. The consequence is a group of working hardware that stop serving the users, or a group of users who face a different user interface on each computer. The following maintenance methods are tested to keep the lab computers running at their peak performance: 1. Set the system profile to restrict user activities at the operating system level; 2. Use system repair software to correct problems and optimize the system; 3. Create image of the system on a CD-ROM and restore the system periodically; 4. Create image of the system, store the image on the network server, and restore the system periodically. Measurements of evaluation are set in consideration of the conflict between users' convenience of using the system and the integrity of the system, and the conflict between the ease of maintaining the system and the performance of the system. This study is not aiming at recommending the best method of keep the lab computers working. In stead, it presents a cost benefit analysis of the various ways of maintenance and helps readers to select the most suitable method for them.

11:10 PERFORMANCE ANALYSIS OF LINEAR ALGEBRA AND NUMERICAL AERODYNAMIC SIMULATION BENCHMARKS ON THE MCSR BEOWULF LINUX CLUSTER

Tyler Simon* and Tanner Pirim, University of Mississippi, University, MS 38677

The Mississippi Center for Supercomputing Research, MCSR, has recently upgraded their Beowulf research cluster. This paper intends to outline the particular method for running linear algebra and numerical aerodynamic simulation benchmarks, as well as monitor the analysis of the results. The benchmarks have been performed on a single Beowulf Linux cluster composed of 219 heterogeneous processors over 10/100 Fast and Gigabit Ethernet message passing ring networks. The results are presented in the form of millions of floating point operations per second over an interval of time and execution time for diverse configurations of computational nodes. The results achieved on the benchmarks for two network speeds have been introduced and compared. Thus, the paper addresses the importance of running such tests as a method for obtaining reliable processing capability for multi-node supercomputing applications. Thorough analysis of the benchmark results has been an exceptional guide in helping the MCSR address and reconfigure cluster bottlenecks, thereby optimizing overall system performance.

11:30 3-DIMENSIONAL SCENE RECONSTRUCTION AND ANIMATION USING VRML/X3D

Stephanie Myrick* and Chad A. Steed, Naval Research Laboratory, Stennis Space Center, MS 39529

Research results are typically reported using 2-dimensional methods that include tables, figures, and charts. With the availability of 3-dimensional (3D) visualization applications, such as the Virtual Reality Modeling Language (VRML) and Extensible 3D (X3D) graphics, alternative methods of information presentation and display are becoming more common. Moreover, these 3D applications can be displayed with viewer software on conventional Internet web-browsers (e.g., Netscape and MS Explorer) with many viewers free and available for download on the Internet. These applications may be used effectively in oral presentations, and for separate viewing on the Internet. The Naval Research Laboratory (NRL) has created visualizations with VRML and X3D to expand and enhance its reporting; this paper describes applications that were developed to visualize naval amphibious assault vehicle (AAV) navigation performances during field demonstrations. Each visualization depicts a beach and ocean scenario and uses actual track data, recorded as a series of latitude and longitude points, to animate a 3D AAV model as it navigates through a planned course. The 3-D AAV model can be viewed separately to convey the physical and visual constraints of the vehicle driver and manipulated (e.g., rotated or moveable parts adjusted) and studied for familiarization purposes. This paper will address the methods used for scene reconstruction and AAV animation in these 3D visualizations.

THURSDAY AFTERNOON

Ship Isle

1:00 A NEW METHODOLOGY FOR CLUSTERING GENOMICS AND HEALTHCARE DATA

Haibo Wang (1)*, Bahram Alidaee (1), and Gary Kochenberger (2), (1) University of Mississippi University, MS 38677, and (2) University of Colorado, Denver, CO 80237

Clustering data is intended to divide objects into groups so that objects within groups have a high degree of similarity. Without the constraints of capacities, many such clustering problems can be modeled as clique partitioning problems (CPP). The associated CCP model is solved by first re-casting it into the form of an unconstrained quadratic binary program (UQP) which is then solved by a tabu search heuristic. Many Genomics and Healthcare data are high-dimensional nominal-scaled and difficult to be clustered. We apply our model and methodology to these data and produce clustering results with high quality. We also propose two new measurement indexes to the high-dimensional nominal-scaled data in this study. This methodology can be applied to other application areas. Preliminary computational experience on the MCSR supercomputers will be presented.

1:20 BIT ERROR RECOVERY IN FACSIMILE TRANSMISSION

Hyunju Kim, Jackson State University, Jackson State University, Jackson, MS 39217

This research proposes bit error recovery methods for Extended 2 Dimensional MMR coded bitstream, which is used in most current facsimile machines. When an error occurs in an MMR coded bitstream, the bitstream cannot be decoded correctly after the error point so that the information after the point is lost. To prevent losing all valid information after the point, we developed a bit error recovery system that detects bit errors, determines the error region, and applies bit inversion and re-decoding to correct the errors. We also investigated the relationships between bit errors and their detection points and developed an algorithm that utilizes the error detection points to reduce computation overhead of the system. Testing results show that around 95% of bit errors are corrected with the system.

1:40 Divisional Business Meeting

2:00 MISSISSIPPI CENTER FOR SUPERCOMPUTING RESEARCH (MCSR) USER ADVISORY GROUP MEETING, POSTER SESSION, AND SPECIAL SUBSESSION ON SUPERCOMPUTING

David G. Roach, University of Mississippi, University, MS 38677

The Mississippi Center for Supercomputing Research was established in 1987 by the Mississippi Legislature and the Institutions of Higher Learning (IHL) in order to provide high performance supercomputing (HPC) support for research and instruction at all state universities. The Mississippi Supercomputer User Advisory Committee (MSUAG) was established by the IHL Research Consortium to provide user input and advice to MCSR management and technical staff on policies and procedures for the Center's operations. It includes member representatives from all IHL institutions. The Advisory Group will meet at this MAS conference. Mr. David G. Roach, Director of the MCSR, will conduct the meeting. The agenda includes an update on MCSR HPC facilities and services, introduction of new MCSR staff members, and site reports and ongoing research updates by MSUAG representatives. A poster session, showcasing research projects that utilize MCSR facilities and services, will follow the Advisory Group Meeting. A Special HPC Subsession of the Mathematics, Computer Science, and Statistics Division, sponsored by the MCSR, will also be held to serve as a forum on supercomputing in which faculty and graduate student researchers will have the opportunity to describe their research projects that involve HPC, Internet2, Grid Computing, Visualization, Network Security, Computer Systems Administration, and the use of MCSR resources. IHL faculty and graduate students, with an interest in HPC and/or MCSR facilities and services, are also invited to attend and participate.

FRIDAY MORNING

Ship Isle

9:30 Workshop on Biostatistical Analysis

Organized by Elgenaid Hamadain, Co-chaired by Elgenaid Hamadain and Todd Nick

Presenters: Elgenaid Hamadain, Todd Nick, and Kanchan Manaktala