Student Abstracts: Computer Science

3D Animation of Pancreatic Molecules. BRITTANY PETERS (Rochester Institute of Technology, Rochester, NY, 14623) DR. PAUL CRAIG (Brookhaven National Laboratory, Upton, NY, 11973)

The pancreas is the supplier of a variety of digestive enzymes and hormones, most notably insulin. In the past a virtual animation of the pancreas has been created, showing a tour of its internal structure. To continue this animation the goal is to introduce molecular detail into this virtual model of the pancreas. We are using Maya’s 3D modeling and animation capabilities to address the challenging task of providing a greater understanding of the pancreas for students, teachers and doctors. Pancreatic proteins contain hundreds or thousands of atoms; the Protein Data Bank (www.rcsb.org) contains many pancreatic proteins structures with x,y,z coordinates of each of its atoms to form a molecule that can be visualized in 3D. The challenge for molecular artists who use Maya is to model these molecules accurately because their structure is too complex to create. One easy was to provide these molecules is to download the PDB files from the Protein Data Bank and visualize them in Maya, but Maya is unable to recognize them. However, the files can be manipulated and changed to achieve the objective. The PDB file first needs to be converted to a simple text document so that the 3D program can read its written information, then it can be drawn through a Maya tool called a mel script. This script is a code that allows Maya to read the PDB file, import it, and model it. From this point the molecular structure can be rendered and animated. This procedure will help to show an up close view of a pancreatic molecule in action. We are developing techniques to display these proteins in a surface format which highlights the overall shape of each molecule, as well as a ball-and-stick format, which emphasizes the individual atoms. One of the advantages of this approach is the molecular artist can import any protein structure that is available in the PDB file format, which will enable viewers to better visualize the physiology of the human pancreas at the molecular level.

3D Stereoscopic Fantastic Voyage of the Heart. NATHANIEL SKINNER (Jamestown Community College, Jamestown, NY, 14702) LEN SLATEST (Brookhaven National Laboratory, Upton, NY, 11973)

For first year Anatomy/Physiology students, the visualization of various organs and body systems can be a bit daunting. This project was designed to combat that problem. Being the first in a long string of organs this model of the heart was designed to give students a working visual knowledge of the heart. The movie starts outside the heart and shows all of the major parts of the heart in their relative positions. The movie then takes the viewer inside the heart. The inside is the part that will be the most helpful for the students, because the best they will ever see if they are lucky are cadavers and drawings. By the end of the movie those viewing this movie will have seen not only the major parts of the movie, but they will also have had a first hand experience with watching the heart beat with the proper timing of the valves. This movie was also put into a stereo version, which will a greatly enhance the viewing experience. The movie was created in the 3D design and rendering program, Maya. Each part of the heart was created as a separate object, which allowed easy manipulation of the parts to get them into the right proportions to each other. The overall movie is actually composed of several smaller movies that were combine with video editing software to give the impression of one voyage through the heart. As stated above the next step is to use some of the same techniques to create various other body parts and organs. The overall long-term goal is to have a flythrough video of every system in the human body.

3-Dimensional Visualization of Molecular Reactions. ANGELICA CORBY (Alfred State College, Alfred, NY, 14802) ROBERT BENNET (Brookhaven National Laboratory, Upton, NY, 11973)

To the undergraduate student certain chemical reactions such as the SN1 and SN2 reactions can be difficult to visualize. By displaying these reactions through a 3-dimensional stereoscopic viewer they become much easier for the student to see. One way to display chemical reactions in 3-D is to take the chemical markup language (CML) file, created from the initial z-matrix used by MOPAC to compute the reactions’ Cartesian coordinates, and convert this file into an extensible 3-Dimensional (X3D) file which can be viewed through a 3-D stereoscopic viewer. After researching the format of the CML and X3D files it is possible to write a Visual BASIC program to rearrange the text of the CML file to the format of the X3D file. The program called “CML-X3D Convert” can take a CML file and convert it into an animated X3D file. This program is intended for educational use. With these X3D files an instructor can more accurately show the 3-dimensional structure of a chemical reaction to a student.

A Simplified Approach to Stereoscopic Imaging of Tomographic Reconstructions in Transmission Electron Microscopy. LAURA DEPOULI, MICHAEL LAMB, SALIM RAHIMI, ROMAN ZRAZHEVSKIY(Nassau Community College, Garden City, NY, 11530) ROBERT BENNETT (Brookhaven National Laboratory, Upton, NY, 11973)

In biological applications, transmission electron microscopy (TEM) enables an investigator to view a specimen at the molecular level, offering a resolving power of ~0.2nm. The two-dimensional (2-D) micrographs captured using the TEM are grayscale images in which contrast is typically achieved using heavy metal stains that limit electron transmission with respect to specimen density. In order to ascertain an adequate understanding of the structure and function of biological systems, it is important to view these structures in three dimensions (3-D), as they appear in life. Currently, there is a myriad of protocols that use various software applications for creating tomographic volumes. Moreover, the lack of a uniform approach often makes the process difficult to replicate for those with limited resources. A simple, consistent method for generating tomographic volumes of TEM micrographs to be viewed stereoscopically was investigated. Sections ranging from ~60-100nm in thickness were collected on formvar-coated copper grids. Samples were viewed using a Philips EM300 TEM with a side-mounted goniometer stage, and images were captured with an integrated Gatan ES500W CCD camera. Common to many laboratories working with limited budgets, various challenges are faced during data acquisition and processing. Physical limitations of the goniometer have restricted datasets to a 45 degree maximum rotation in either direction about the horizontal axis. A tilt image series was digitally captured from -45 to +45 degrees in 1 or 2 degree increments and reconstructed into tomograms and rendered using Stanford University’s EM3D software. Finally, stereoscopic visualizations of the rendered three-dimensional volumes were generated through the creation of movies that were viewed using the software application StereoMovie Maker.

A Stereoscopic Model of a Theoretical Einstein-Rosen Bridge, i.e. Wormhole, as part of an Expedition to a Black Hole at the Center of a Supernova Remnant. TAISIA CARDO (Suffolk County Community College, Selden, NY, 11794) DAVE STAMPF (Brookhaven National Laboratory, Upton, NY, 11973)

A representation of an Einstein-Rosen bridge in three dimensional space has the potential to provide the scientific community with a new educational learning tool. Maya™, a computer program designed to create three dimensional (3-D) renderings, was used to design Non-Uniform Rational B-Spline (NURBS) objects to construct a scaled version of a wormhole. Steady research shows that wormholes are created from two connected singularities, points with infinite densities, from within black holes. The tunnel is theoretically smaller than an atom and stays open for merely a fraction of a second. Kip Thorne suggested that it would be possible to travel through time if one could complete the course through a wormhole intact. Considering such theories, a scaled tunnel-like wormhole was constructed. Using the capabilities of Maya™ all breaks, tears, and/or misalignments throughout the structure were disencumbered. The resulting wormhole has arched curves with no visible attached pieces. This 3-D representation has the potential to aid the inquisitive mind of students and scientists alike in uncovering the mysteries the universe has to offer.

Adaptively Improving Long Distance Network Transfers with Logistics. DAVID LABISSONIERE (East Tennessee State University, Johnson City, TN, 37614) KENNETH ROCHE (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

The congestion control mechanisms in the Transmission Control Protocol (TCP) severely limit the bandwidth achieved by long distance data transfers. However, it is regularly necessary to move data over great distances across the Internet. The throughput of such transfers can be improved by applying the logistical technique of breaking a single long distance transfer into multiple shorter transfers. This technique can result in significantly improved throughput while still working with the TCP congestion controls and not attempting to circumvent them. We have developed an algorithm that uses this technique to traverse the network path that approximates the best available bandwidth while the data is in flight. The algorithm couples graph techniques with real-time latency and bandwidth measurements to adaptively respond to network dynamics. The algorithm shows improvements in speed and flexibility over standard data transfer methods such as FTP. Future work includes algorithmic improvements to better predict and model network bandwidths as well as a demonstration of the algorithm as the data transfer mechanism for a production-scale grid computation.

ADDING FORWARD AND REVERSE MODE DIFFERENTIATION TOOLS TO NEOS. RACHEL SISTERSON (University of Illinois, Champaign-Urbana, IL, 61820) PAUL HOVLAND (Argonne National Laboratory, Argonne, IL, 60439)

NEOS (Network-Enabled Optimization Server) has optimization solvers that provide function and/or gradient evaluations. This is offered freely to the public and achieved with the help of automatic differentiation (AD) tools. ADIC (Automatic Differentiation in C) and Tapenade are two types of AD tools. NEOS has the ability to have more solvers and AD tools added. A NEOS solver finds an x that minimizes f(x) for a user-defined function and the AD tool finds the derivatives of the user-defined function that are used by the solver algorithms for optimization. Tapenade and a newer version of ADIC were to be made available on NEOS through the BLMVM (Bound-Constrained Minimization Limited-Memory Variable-Metric Algorithm) solver. In order to add these AD tools, BLMVM's drivers, Makefiles, and XML (EXtensible Markup Language) file had to be modified. Both AD tools have to be added as an option in the XML file and pathways to the AD tools homes, libraries, and environmental variables needed to be defined within the drivers and Makefile. Future steps might be to have the ADIC upgrade and Tapenade available on solvers other than BLMVM on NEOS.

An Evaluation of Vista Performance on Current Windows Computer Systems. JAMES ARRIAGA (BigBend Community College, Moses Lake, WA, 98823) CHARLIE VERBOOM (Lawrence Berkeley National Laboratory, Berkley, CA, 94720)

With the impending release of Windows Vista in January 2007, the Information Technology Division at Lawrence Berkeley Laboratory is faced with a decision involving the future transition of computers from the Windows XP operating system to the Windows Vista operating system. Research involved a review of benchmarking tools, the selection of a standard set of tests, and the application of those tests to workstation hardware in use at Berkeley Laboratory. To evaluate the impact of installing the new operating system, performance benchmarks were run on 3 desktop systems with both a standard configuration and with selected hardware upgrades. This provided a comparative analysis to determine the effects of Vista with the Authentic, Energetic, Reflective, and Open (AERO) graphical user interface and how it will react on hardware specifications as determined by Microsoft on systems commonly used at Berkeley Laboratory. The new enhanced user interface built into Vista requires more powerful hardware resources than current computers running XP can handle. The comparison of the minimum and recommended requirements for computer hardware with the Vista operating system installed indicated that legacy systems were more likely to be replaced than upgraded. Computers manufactured within the last 2 years that meet the published guidelines should run Vista with or without the AERO interface.

Automatic Detection of CRISPR elements. KYNDALL BROWN & MICHAEL LOWE (Jackson State University, Jackson, MS, 39217) NIKOS KIRPEDES (Lawrence Berkeley National Laboratory, Berkley, CA, 94720)

Due to the growing interest and importance of CRISPR elements in the scientific community, there was a desire for devices that would speedily and efficiently detect CRISPRs. Initially their detection relied on tandem repeat finding tools such as PatScan, Piler, and Reputer. Unfortunately, this process required hours of manual post processing due to the software being inaccurate in distinguishing CRIPSR elements. To solve this problem, the research team at the Lawrence Berkeley National Laboratory (LBNL) created a CRISPR Recognition Tool (CRT), to fast and efficiently detect CRISPR loci. CRT was implemented in Java, an object-oriented programming language. This algorithm used Boyer-Moore searching and skips techniques in locating CRISPRs. The program starts by scanning for a small region of bases that appear within the searchwindow. Assuming that the pattern is part of a CRISPR, it will appear within a range that is relative to the size of the spacer and repeat. Once the range is determined, the program searches for the pattern using Boyer-Moore and the skip method. Once the match is found, and the full repeat length is determined, it is deemed a CRISPR candidate. The program then passes the candidate through three filters with the user specified parameters. Once the candidate passes all requirements it is confirmed as a CRISPR, and the program repeats the process until all CRISPR elements are found. To determine the effectiveness of CRT, its speed and accuracy was tested against PatScan and PilerCR. As a result, the tests proved CRT to be the fastest and most accurate of the three tools. In conclusion, the algorithm efficiently detected CRISPR elements and will be a useful tool in the scientific community.

Automatic Semantic Inference and Extraction of Textual Information from Internet Resources. PHILLIP MARTIN (Clemson University, Clemson, SC, 29634) THOMAS E. POTOK (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

The Internet is one of the largest available public collections of data. Current methods of extracting the Internet’s copious amount of data by hand in a timely manner have proved inadequate. To make the untapped resources of the World Wide Web (WWW) more readily accessible, it is necessary to automate the data extraction of several different formats. Hyper Text Markup Language (HTML) is the most prevalent data format for the World Wide Web. However, HTML is designed for the visual presentation of data on a website and thus includes little, if any, semantic information. The Resource Description Framework (RDF) developed by the World Wide Web Consortium provides a common format that enables a user to represent low-level semantic information and structure. In addition, the Internet contains several other data formats including documents generated using Microsoft Office and Adobe PDF. The challenge is to gather HTML encoded documents and generate basic semantic data using the RDF format and to expand automatic data extraction to include other formats. To address this challenge, an RDF document is automatically generated from the basic syntax derived from an HTML document. This allows the inference of some basic semantic information from any ordinary HTML file. Text extraction is further expanded to include both Microsoft Word documents and Adobe PDFs. The results demonstrate how the methods described above can provide a broader and more robust basis for automatically gathering information from the internet.

Collection and Interpretation of Teragrid Usage Statistics. JEFFREY LEHEW (Milwaukee School of Engineering, Milwaukee, WI, 53201) DANE SKOW (Argonne National Laboratory, Argonne, IL, 60439)

The TeraGrid network, a multi-facility parallel computing resource allows researchers to answer important questions that require large amounts of computing power. The TeraGrid is relatively new, and coherent collections of data on its usage are scarce. The goal of this project is to create tools to view and interpret existing data sources to allow developers to improve the TeraGrid infrastructure. To solve these problems, two different types of receivers were setup. One receiver was setup on a TeraGrid machine to listen for and store UDP (user datagram protocol) packets sent at the completion of activities using Globus software tools such as GridFTP and GRAM (Globus Resource Allocation Manager). In addition, a second packet receiver was started on the Juniper routers that make up the backbone of the TeraGrid network. These receivers sampled the data flow through the network, and once recorded, the data was sorted and output using the Flow-Tools toolkit. The data from these receivers was presented in a graphical format using Gnuplot, a graphing application. The uniqueness of this project is in combining these various tools to create an easily understandable metric of TeraGrid usage from a large set of unorganized data. This study is part of a larger project, and the tools developed at this stage will be integrated into a long-term study of TeraGrid usage. Future work will involve completing testing of these receivers, making some changes in the way graphical data is presented, and presenting this information in the form of a webpage.

COMPONENTIZING NETPIPE AND INTERFACING WITH GAMESS. RICHARD WANG (Carnegie Mellon University, Pittsburgh, PA, 15213) MASHA SOSONKINA (Ames Laboratory, Ames, IA, 50011)

Common Component Architecture (CCA) allows programs written in different languages and having different interfaces to interoperate with minimal effort. As a result, complex applications with enhanced functionalities may be constructed, so that they run more efficiently and readily facilitate scientific progress. NetPIPE and GAMESS, developed at Iowa State University and Ames Laboratory, have been used to demonstrate this outcome when brought together by CCA. GAMESS is a scientific application that can be used to solve a variety of quantum chemistry computations. NetPIPE is a network evaluation program that sends messages back and forth between two a network nodes. A NetPIPE component has been created and connected with GAMESS components using a CCA framework to provide GAMESS with the knowledge of network resources availability. First, NetPIPE analyzes the current network state and checks whether it is in line with the full capacity for the given network. Second, NetPIPE's findings, GAMESS configures its communication protocols. The aim is to show that GAMESS will run more efficiently when tuned to network conditions as analyzed by NetPIPE in the preprocessing stage. In general, experiments underscore the premise of the component design that scientific applications can execute more efficiently and tackle more complex problems when implemented as components and interconnected within a component framework.

Components for a General Purpose FPGA Implementation of Template Matching Normalized Cross-Correlation. TIMOTHY CHAGNON (Wake Technical Community College, Raleigh, NC, 27603) KEN PERRINE (Pacific Northwest National Laboratory, Richland, WA, 99352)

Template matching normalized cross-correlation (NCC) is a fundamental, but computationally intensive method of feature detection used in applications such as computer vision, optical character recognition, and video data extraction. An implementation of this algorithm exploiting the configurable logic, high throughput, and parallel paths available in a Field Programmable Gate Array (FPGA) is desirable for use as a PC co-processor. Compared to software development however, the design of an FPGA configuration is a time consuming and expensive process. This paper outlines parts of an FPGA implementation which were designed using mature algorithms and basic parallelism as a readily accessible tool for fast software prototyping or specific optimization. The NCC design uses the optimized multipliers and BlockRAM on a Xilinx Virtex-II device to accumulate the correlation and normalization factors for a single template placement within the image to be searched. Fixed point non-restoring square-root and division blocks are then applied to yield the normalized cross-correlation value for that template placement. High speed external ZBT SSRAM (zero bus turnaround synchronous static random access memory) and demand mode DMA (direct memory access) are used to transfer image data from the host PC. As-is, the design components can be a used as building blocks for continuing work on implementations of the NCC algorithm which operate over the whole search domain. Further work to expand upon the design with coarse-grained parallelism is suggested for an application to fully utilize the FPGA hardware. The use of algorithms other than NCC, and the use of constant multiplier blocks as templates are also interesting avenues of research that could be pursued.

Content Analysis of Newswire Articles. TENISHA BARNES, MONCEF BELHADJALI, & SAMUEL MITCHUM (Norfolk State University, Norfolk, VA, 23504) ANTONIO SANFILIPPO (Pacific Northwest National Laboratory, Richland, WA, 99352)

Information extraction research aims at developing tools for pattern recognition with a minimum human intervention. In order to develop a semi-supervised system, a classification approach was developed in which the cosine measure was augmented with knowledge from WordNet to provide a better matching criterion to recognize target pattern in a data sets with reference to a small set of seed patterns. The automated system was built using Java within Eclipse for programming, Connexor for parsing, Cicero Lite and LingPipe for name entity recognition. To test the system, a corpus on management succession from the sixth message understanding conference (MUC-6) was used. The results of precision, recall, and F-measure showed that the modified cosine measure provides interesting results, but didn’t perform as well as previous approaches that used WordNet to compute semantic similarity. Further work will reduce this gap and may even surpass previous results. The package offered in this study can be improved to build a system generic enough to be applied as a text mining tool in a variety of domains.

Creating a new stereo viewing software platform. KYLE HUNT & SETH JACOBS (Alfred State College, Alfred, NY, 14802) MIKE MIGUIGEN (Brookhaven National Laboratory, Upton, NY, 11973)

Creating a new stereo viewing software platform. SETH R. JACOBS (SUNY College of Technology at Alfred, Alfred, NY 14802) MIKE MCGUIGAN(Brookhaven National Lab, Upton, NY 14802) Living in a three dimensional world poses a sizable problem when it comes to educating students and the general public on topics that can not accurately be portrayed in a two dimensional medium. Stereo projection systems and the software used in conjunction with them provide a potentially intuitive and useful tool in conveying ideas of this nature. At current, however, there is no adequate software solution to act as a stable base on which to develop and execute a successful lesson plan. Software of this type that has been provided for public use lacks a number of things that users need and appreciate. Some of these issues include: presence of bugs, lack of support, lack of documentation, lack of portability, lack of file compatibility and a lack of user-friendliness. Due to this negligence on the behalf of software developers, these types of learning tools are rarely used and are considered by many to be a novelty. In order for stereo tools like these to be implemented successfully in the educational environment, there must be at least one single, intuitive and robust software solution. The objective of this project is to create a software package that is full featured, open in design, and addresses the shortcomings of current generation stereo viewing programs. In order to accomplish this goal, the Java programming language and the Java3D advanced programming interface will be used, along with documentation provided by Sun Microsystems' home page http://java.sun.com. The object oriented nature of Java will allow for the program's functions to be split up into modular classes which can be edited individually or completely replaced as needed. The program will use a desktop-like graphical user interface with tabbed windows to display models and views of those models. There will be user interface objects to handle manipulation of the models as well as activating and deactivating stereoscopic viewing. The project will be posted to the public Internet along with source code and documentation so that it can be improved upon past the completion of the summer 2006 internship.

Creating Three Dimensional Stereoscopic Animations of Organic Chemical Reactions through PYMOL for Enhanced Visualization of Reaction Mechanisms. MARIA MATEO (Bergen Community College, Paramus, NJ, 7652) ROB BENNETT (Brookhaven National Laboratory, Upton, NY, 11973)

Three dimensional stereoscopic animation of organic chemical reactions is an effective means for illustrating bond associations and conformational changes that would otherwise be difficult to visualize if represented as two dimensional images. Executing chemical reaction modeling requires the use of various computer programs and programming languages. The main foundation of this project is a chemical calculation program called MOPAC (Molecular Orbital Package) which estimates both the best structure and minimum energy of a molecule or a small reaction given an initial presumption on the molecular structure or reaction path. The information is input as a z-matrix which, after being calculated by MOPAC, outputs an array of digital information that is lengthy and complex. However, the Centre for Molecular and Biomolecular Informatics (CMBI) at Radboud University Nijmegen in the Netherlands created a website, called Mol4D, that presents the MOPAC data within a friendlier, GUI-based (graphical user interface) environment. This website provides the output data that are needed to create the stereoscopic version of the animation. PyMOL, a molecular modeling program, is used to animate the chemical reaction in quad-buffered stereo, which is the type of stereographic setting that is compatible with the 3D projector being used for this project. Because PyMOL functions mostly by command scripts, which can be very tedious to learn, a combination of C++ and Python was used to program a plug-in for PyMOL that, when prompted by the user, imports the correct output file from the MOPAC calculations, separates that file into distinct movie frames, creates the overall movie script and runs it through the PyMOL viewer, and transmits the final animation in quad-buffered stereo. The intention behind creating a more user-friendly adaptation of PyMOL and using it in conjunction with the Mol4D website is so that the final outcome can be used as a valuable teaching tool for chemistry, specifically organic chemistry. The 3D stereoscopic projection of the desired reaction can help a student vividly see the discrete conformational changes that occur within an organic reaction, and thereby give the student a better understanding of the mechanism of that reaction.

Creation of Computerized Datasheets for use in Calibrations. RACHEL KELLER (Hartwick College, Oneonta, NY, 13820) PAUL ZAHRA (Brookhaven National Laboratory, Upton, NY, 11973)

At Brookhaven National Lab (BNL) the procedures used in the Radiological Control Calibration and Instrumentation Group (Rad Con C&I Group) to guide the calibration of radiological equipment cover an entire class of instruments. These procedures provide the guidelines by which an instrument is deemed fit for use; however, the usefulness of these procedures is limited by the knowledge of the technician concerning what needs to be done and what options provided are not applicable for the unit. In addition, many paper data collection forms are used throughout the course of a single instrument calibration, making the procedure for data entry into the computer and managing the hard copy collection of results difficult and cumbersome. Using the upper level general procedures already in place as well as information from the technical manuals for the instruments undergoing revision, new, specific instructions and collections forms are being formulated. Sources that were useful in the work so far to create a computerized calibration included the manual for the Manual Met/Cal program, and the upper level procedures already created by the Rad Con C&I Group at BNL. The program Manual Met/Cal aids in the creation of computerized datasheets with accompanying instructions that match the data fields’ contents. These companion pieces not only tell the technicians what to do, but also where on the forms to enter the data they have collected. Since the datasheets can be printed out to be used away from a computer terminal, the data forms will now be in the same format as the computerized version, thereby making the data entry process easier. When the data entry is complete and the sheet is saved, a standardized report form has now been created to be printed out, providing a hard copy detailing the instrument specifics, the information collected, and the status of the instrument. This is a small portion of a larger project to bring the procedures and data collection systems for calibrations up to date.

CRISPR Recognition Tool (CRT): A Tool for Automatic Recognition of CRISPR Elements. CHARLES BLAND (Jackson State University, Jackson, MS, 0) CHARLES BLAND (Lawrence Berkeley National Laboratory, Berkley, CA, 94720)

Repetitive sequences make up a significant portion of the genomes of both prokaryotes and eukaryotes. The human genome, for example, is known to be composed of as much as 50% repeating patterns. These repeats come in various forms/sizes and may be found dispersed throughout a genome, clustered near each other or occurring contiguously. The identification of repeats has proven to be of great consequence as they have been connected to a number of human diseases, including fragile-X mental retardation, Huntington’s disease, myotonic dystrophy and muscular atrophy. Repetitive sequences also have various functional roles such as gene regulation and immune system development. Furthermore, they are a useful tool to scientists for DNA fingerprinting and genome alignment. This research focuses on a recently recognized family of repeats that has attracted a lot of interest recently. Clustered Regular Interspaced Repeat (CRISPR) are composed of short direct repeats ranging in size from 21 – 51 base pairs. CRISPR are unique in that their repeats are interspaced by non-repeating sequences of similar size and are found only in the genome on prokaryotes. Because of the importance of repetitive sequences, it is essential to develop fast and accurate methods for their detection. Several tools are available for identifying various forms of repeats, however, because the focus on CRISPR elements is only recent, no published tools are yet available for their automatic discovery. Their detection currently uses a generic repeat searching tools (Patscan) and requires considerable manual post-processing. In this study, we present a tool for reliable, fast, and automatic detection of CRISPR elements. This software program, CRISPR Recognition Tool (CRT), uses a fast linear search method for their detection. Accuracy and speed of CRT is determined by analyzing its performance on finished microbial genomes available in the IMG version 1.5 database. Additionally, CRT is compared to Patscan and a recently discovered and unpublished CRISPR detection program, PilerCR, an offshoot of the Piler program. We found CRT to be superior to both Patscan and PilerCR in terms of both accuracy and speed.

Data Encryption for Windows PC's. BRICE LUCERO (Big Bend community college, Moses Lake, WA, 98837) CHARLIE VERBOOM (Lawrence Berkeley National Laboratory, Berkley, CA, 94720)

In today’s society the use of mobile technology is becoming increasingly popular. The demand for the ability to secure sensitive data in a safe and easy manner on Windows operating systems is growing with the use of mobile technology. This project involved an analysis of present and future methods for data encryption for Windows operating systems: XP and the soon to be released Vista operating system. The project scope included Encryption File System (EFS) in Vista and XP along with Bitlocker, a new feature that will be included in Vista. In addition two third party programs, Safe Disk and True Crypt, were reviewed. Tests were run in order to measure the time each method took to encrypt as well as the impact they had on data transfer performance. Ease of use, data recovery methods and known vulnerabilities were also taken into consideration during the review. It was determined that Windows EFS and Bitlocker were effective, easy to use methods for data encryption and have reliable recovery methods when managed though a domain. Safe Disk is protected by one centralized password and has an interface that was easy to learn but offered no recovery method. It is not a recommended method. True Crypt had limited options available for data recovery and a choice of password or key files for authentication. The interface was slightly more complicated then Safe Disk and the native Windows encryption methods. True Crypt is still a recommended method for data encryption.

DATABASE AND APPLICATION DESIGN FOR INTEGRATING RFID TRACKING METHODS INTO THE COMMERCIAL SUPPLY CHAIN PROCESS FOR RADIOACTIVE MEDICAL ISOTOPES. JAMES HICE (Albion College, Albion, MI, 49224) DAVID RESSEGUIE (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Current methods of tracking use line of sight bar codes which encode a number, the package's carrier tracking number, to identify each package. Each package is tracked from when the package is shipped at a carrier through final delivery. Progress of the package is marked when the bar code is manually scanned at each of the carrier's facilities and when the delivery is made. Tracking packages containing radioactive medical isotopes and other sensitive materials beyond the standard shipping routes and carrier facilities is becoming increasingly important. A test for tracking sensitive packages using Radio Frequency Identification (RFID) tags has been created to allow sensitive package tracking without requiring manual, line of sight scanning. These RFID tags transmit a unique identification number and will be placed inside packages on each radioactive medical isotope container. Shippers register each package using a web application which inserts the information into a database. RFID readers set up at the shippers', carriers' and recipients' facilities read the packages as they pass within range and update the database with the RFID identification number, time, and location. No web application or database existed for these tests, so extensive work was done to design and create the database structure and required application. A relational database design was used to store the package, RFID, and carrier provided tracking data. The web application, which serves as the shippers' interface for inserting and managing information in the database, consists of multiple Java Server Pages. The main considerations in creating the web application's design were ease of use for shippers as well as capturing and managing the necessary data for packages of interest. Future applications will use the database information to display package information, geographic location, and compare tracking methods. Further work can be done to display the web application more dynamically, providing real-time updates from the database. The web application and database can also be used for continued tracking of shipments beyond this test, including general, non-sensitive RFID package tracking.

Design of Plotting Utilities for the Collider-Accelerator Department. DAVID KATZ (Alfred State College, Alfred, NY, 14802) SETH NEMESURE (Brookhaven National Laboratory, Upton, NY, 11973)

The Collider-Accelerator Department at Brookhaven National Laboratory has developed a mature set of C++ utilities for analyzing data. The purpose of this project is to provide a set of similar tools written in Java that allows both developers and users a similar set of functionality. The feature set that has been provided to date includes a set of curve fitting utilities, the ability to cut, paste, load, and save data sets from plots, and a flexible way to add text to any plot. There are two commercial packages being used for the development of the plotting utilities. One is a math package created by Dr. Michael Thomas Flanagan of University College London and the other is called JClass, which is a plotting package created by Quest Software. The curve fitting feature uses both of the packages to allow the user to perform essential curve fits on raw data collected. Gaussian, Gaussian with an offset, Exponential, Exponential with an offset, Polynomial, and Double Exponential are among the fitting functions supported. After the fits were coded there was extensive stress testing to determine how well each fit performs on raw data. In addition to curve fitting the Java package that has been developed provides a means for the user to open saved plots created in the existing C++ software. There is also a feature that allows the user to copy and then paste a plot onto two independent y-axis setups with different scales. Labels can also be added to the plots with a variety of looks and styles which the user can customize to their preference. The Java package has a similar graphical user interface setup as the existing C++ utilities so essentially there is no learning curve for developers and users. This package will be used at the Collider-Accelerator Department to potentially make critical decisions on the raw data collected.

Designing a Mobile Application for Purification Database Systems. PAMILA NELSON (Governor State University, University Park, IL, 60466) SOON-OK PARK (Governors State University, University Parkway, IL, 60466) RAKEYA SMITH (Governors State University, University Park, IL, 60466) DR.ANDRZEJ JOACHIMIAK (Argonne National Laboratory, Argonne, IL, 60439)

The Midwest Center for Structural Genomics at Argonne National Laboratory develops and optimizes integrated methods for determination of protein structures through x-ray crystallography. Scientists at the center traditionally have written down their experimental results on paper, later returning to their desktop computer to enter the data on a Web page. The objective of this project was to design and develop a personal digital assistant (PDA) exhibiting the same functionality as the Web page but providing much greater convenience and accuracy. The PDA application was developed by using Microsoft .NET technology. Specifically, user interfaces were developed with Mobile ASP.NET forms, HTML, and JavaScript. The scientific logics were implemented with C# programming language. Like the Web page, the PDA application uses Oracle 9i database for data storage. With the PDA, however, scientists can scan bar-coded purification instruments and enter the experimental values directly, updating the database in real time and comparing values with prerecorded information about volume, temperature, and concentration ranges for proteins. The PDA thus increases the efficiency of data collection and analysis and minimizes the errors inherent in transposing data from paper to computer. In order to assist the scientists, an online user manual was also developed as part of this project. The PDA application is now available for use at Argonne. This application enables scientists to access and modify the data, in the lab or almost anywhere, while preserving the sophistication of the Web application. Moreover, the PDA provides an important step toward automation of biochemical laboratory and integration with databases.

Developing Domain Specific Ontology to Define Relational Data Flow Process Modeling. DANIELLE EVANS (Big Bend Community College, Moses Lake, WA, 99357) MITCH PELTON (Pacific Northwest National Laboratory, Richland, WA, 99352)

The Framework for Risk Analysis in a Multimedia Environmental System (FRAMES) is a platform which allows a user to develop and simulate an ontology for a specific domain. Modules in FRAMES use a relational data structure to simulate real world processes. Each module maintains a set of ontological dictionaries, called a connection scheme. This allows FRAMES to connect modules by what they consume and produce. FRAMES has two types of dictionaries developed for data flow process modeling module input and boundary condition. Module input dictionary contains module specific information that is usually supplied by the user. The boundary condition dictionaries are the relational (ontological) data that flows between modules in FRAMES. To aide in the development of the ontological boundary condition dictionaries four categories of dictionaries were developed and used for this example ontology, they are ‘selection’, ‘feature’, ‘property’, and ‘series’ dictionaries. Selection dictionaries define the focus of the simulation. Feature dictionaries define the where conditions and are made up of geographical points, polygons and lines. Property dictionaries are the static data for models and consist of the names of variables. Series dictionaries consist of the analog signal output for a simulation. These categories helped in the management of data within a relational data flow process model. Both the dictionaries and variables for this ontology follow a domain specific naming convention. Dictionaries contain the metadata of a domain specific ontology. Ontologies represent the particular meanings of terms as they apply to a specific domain. This paper will attempt to examine an ontology developed for an environmental fate and transport modeling domain in FRAMES and make some generalizations that may aide in the development of other domain specific ontologies.

Development of 3D Stereo Visualization Tool for Teaching Introductory and Organic Chemistry Courses. JUNG MIN RYU (Bergen Community College, Paramus, NJ, 7652) ROBERT E. BENNETT (Brookhaven National Laboratory, Upton, NY, 11973)

Substitution Nucleophilic Bimolecular (S_N2) reactions require a nucleophile to collide with an electropositive atom with an attached leaving group. Visualizing the physical displacement of molecular components in such a reaction is a key to understanding the chemistry involved. Traditional methods have relied on standard print illustrations that can directly represent any two dimensions (X-Y, X-Z or Y-Z), but which must then imply the third dimension (3-D) using a set of standard, but arbitrary symbols. There are now online and open-source software tools that make 3-D stereographic representations of such reactions possible; however, feature sets vary, user interfaces are generally clumsy and interoperability among packages is limited. This project created a tool which integrates the capabilities of a web-based visualization resource (Mol4D) and an open-source visualization product (PyMOL). Our goal has been to produce a user interface that makes the tool appropriate and effective for use in the classroom. The Mol4D (Molecules in Four Dimensions) website provides a graphical user interface that displays 3D structures and animations using a plug-in or a Java applet. One component on the web site, MOPAC (Molecular Orbital PACkage), calculates orbital energies, electron densities and molecular structure numbers, bond lengths and angles. Our group extended the capabilities of MOPAC by converting the data it generates into a format suitable for use with an open-source visualization product, PyMOL. PyMOL, which employs an embedded Python interpreter and generates high-quality molecular graphics images and animations, is not user-friendly since its functions are command scripts. Using C++ and Python, we have created a user-friendly plug-in for PyMOL that accepts converted MOPAC data files from Mol4D, separates it into frames, prompts the user to indicate bond connectivity, generates a movie script, and then outputs the final animation to the PyMOL viewing screen.

Development of a Web-Based Report Tool for User Queries Made of the Integrated Relational Model of Installed Systems (IRMIS) Control System Database at the Advanced Photon Source Accelerator. DAWN CLEMONS (Purdue University Calumet, Hammons, IN, 46321) DEBBY QUOCK (Argonne National Laboratory, Argonne, IL, 60439)

The Advanced Photon Source (APS) accelerator is run by a complex and specialized control system. IRMIS, a descriptive control system relational database application, allows accelerator operations to be more easily monitored and maintained. Several IRMIS viewers (user interfaces for querying the database) present information in ways that are comprehensible and useful to those who support the accelerator. However, lacking in the viewers was the mobility of their results separate from the web interface; thus, a reporting tool was necessitated. The main achievement of this project is to provide a means for easily saving queried results. Results can be saved as text files, comma-separated (CSV) files, or viewed as printer-friendly HTML pages. Methods of software architecture were employed, as well as code qualities useful to it, such as adaptability, readability, and appropriate documentation. Because the report tool was to be adapted to an established database project, the tool required an interface design that merged with one already existing, and the code needed to be integrated into hierarchically structured viewer source code. Resources included the languages PHP 4, HTML, and JavaScript; the Zend scripting engine; and the Apache web server. The IRMIS report tool aids with APS control system troubleshooting and diagnostics. It also is a step towards providing a unified PHP viewer framework. Because the tool adds to the descriptiveness of IRMIS, it also lends itself to a desired future prescriptiveness of the database.

Enhancing Javapeño. AARON FLECKENSTEIN (Western Michigan University, Kalamazoo, MI, 49093) DR. BRADLEY T. REARDEN (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Javapeño is a Java-based data visualization package that is part of Oak Ridge National Laboratory’s SCALE software suite. Javapeño provides a graphical interface that allows the user to visualize data in customizable plots. Javapeño was created by Dr. Bradley Rearden in 2001 as a port from plotting software called SenPlot. Originally supporting only sensitivity data from the TSUNAMI modules of SCALE, Javapeño has been improved to support additional data types. Currently, Javapeño enables visualization of data generated from SCALE modules: TSUNAMI, SMORES, KMART, and XSDRN-PM, as well as a generic 2D data type. Sponsors of SCALE development have requested that Javapeño be expanded to support visualizing neutron and gamma cross-section data in the AMPX format and cross-section-covariance data in the COVERX format. For this project these new data types required adding three dimensional (3D) plotting capabilities to Javapeño. To maintain cross-platform compatibility Javapeño’s 3D plotting engine uses the OpenGL version of Java 3D, an open source Java extension. Javapeño’s 3D engine is an extensively modified version of FreeHEP’s Lego Plot application. After developing the 3D engine, support was added for the COVERX data type. There were conflicts with the existing methods implemented in Javapeño and the way Java handles Java 3D objects. Java 3D objects are treated as heavy-weight objects, while the rest of Javapeño is light-weight. In Java, heavy-weight objects are always placed on top of light-weight objects, even when they are not active. Thus, unless modified, the new 3D plotting capability would prevent using the existing 2D plotting capabilities of Javapeño. To address this conflict, a new interface and window manager was created. Planned future enhancements include support for AMPX data files, saving plots, and configuration files.

Information System Comprised of Oncological Data Pertaining to Methylation in Ovarian Carcinoma. WALTER LEWIS (Cheyney University of Pennsylvania, Cheyney, PA, 19319) SEAN MCCORKLE (Brookhaven National Laboratory, Upton, NY, 11973)

Over the past decade DNA sequencing technology has vastly improved and become more affordable. Consequently DNA and protein sequencing databases have grown exponentially, and the annotation of the human genome has increased immensely. Currently existing oncological databases are sparse forcing the researcher to manually search through scores of biomedical journals and related websites. The project attempts to address this issue by creating a smaller scale information system, one that will perfectly suit the needs of cancer-methylation researchers. The system consists of three major components: data collection, the Postgresql server, and the Java user interface. The Postgresql server was first established, then data was gathered and filtered. Once the data was stored in the Postgresql server the user interface was created. This interface acts as a bridge between the biologist and the Postgresql database, allowing non-programmers to perform many of the same Postgresql search commands without knowing the language. The result is a net-based database that will aid in the research involving the methylation of CpG islands in promoter regions of genes particular to ovarian carcinoma. Via a number of sources favored by the oncological community, the database embodies a plethora of relevant data such as the gene identifiers, chromosomal position, CpG island methylation, relevant oncological information, and hyperlinks to abstracts from related oncological studies. Data gathered in the tables can be correlated with other preexisting tables by simple queries performed by the user. While the stored data is particular enough to be very helpful to this project, it is not limited to just ovarian oncology. The database has already provided useful information for an unrelated P53 radiation damage study, proving the system can be easily augmented to assist other oncological research.

Integrating mpiP and Eclipse. JOSEPH LAMBERT (Bucknell University, Lewisburg, PA, 17837) JEFFREY VETTER (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Internal IT Communication through Use of Web Applications. CHRISTOPHER DUNCAN (Brigham Young University, Provo, UT, 84606) CARL FENNON (Idaho National Laboratory, Idaho Falls, ID, 83415)

In large organizations, IT issues that affect many users need effective communication to ensure the quickest, most efficient, and reasonable response from both end users and IT staff. The IT infrastructure at the Idaho National Laboratory, or INL, comprises both an Operations Center to handle calls and a Field Services arm to make service calls to its thousands of clients. Currently, INL IT communicates information regarding issues that affect many users via email or phone. To unify response from IT and other staff (and to eliminate emails), it became important to explore how the IT organization could more efficiently communicate about these issues by using web or other applications, the Operations Center’s several plasma screens, and Field Services’ Blackberry devices. Using free PHP scripts to connect XHTML pages to the laboratory’s SQL Server 2000 databases, a web application (of approximately 7000 lines of code) was created to store and retrieve information on current IT issues. To take advantage of the Operation Center’s Plasma screens, a similar web application was used to fill an entire plasma screen in addition to two smaller java based applications that display less of the same information as the original web application. Pending approval from the INL Cyber Security Department, a setup for blackberries was created with access to a SSL secure page using RSS technology to transport the same information as on the main web application in a simplified, secure, and "real-time" format. If all prove to be fairly error-free and stable, then the system would provide an efficient way of communicating issues for INL IT. If a server, network, or other system goes down, the technician that works the problem can efficiently and conveniently let everyone at the laboratory know about the problem and take appropriate action. IT and other staff at the laboratory can all do their jobs with greater confidence and less stress because of access to this forum for technology issues that affect multiple individuals at the laboratory.

Linearity Analysis. LUIS RAMOS (La Sierra University, Riverside, CA, 92505) PAUL HOVLAND (Argonne National Laboratory, Argonne, IL, 60439)

Linearity analysis determines which variables depend on which other variables and whether the dependence is linear or nonlinear. One of the many applications of this analysis is determining whether a loop involves only linear loop-carried dependences and therefore the adjoint of the loop may be reversed and fused with the computation of the original function. This paper specifies the data-flow equations that compute linearity analysis. In addition, the paper describes using linearity analysis with array dependence analysis to determine whether a loop-carried dependence is linear or nonlinear.

Logical Networks for Gene Regulatory Interaction Modeling. ERIC LANCE & CHRISTOPHER LEWIS (New Mexico State University, Las Cruces, NM, 88011) ELISSA CHESLER (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Modeling the dynamics of gene expression levels can lead to a quantitative understanding of mechanisms in cellular processes, such as signal transduction and disease pathways. Previous studies of gene regulatory networks (GRN) have established Boolean networks as good dynamic models for GRNs. The Boolean network model regards every gene in a GRN as a binary variable whose value is a Boolean function of its regulators. In order to apply this model, continuous gene expression levels must be quantized to binary states; however, existing quantization methods discretize one gene at a time and are highly likely to miss complex interactions among genes. Furthermore, binary quantization may be insufficient to capture all interactions. To remedy these issues a novel quantization method able to capture complex interactions and a generalization of Boolean networks to multi-ary logical networks are required. The novel quantization approach creates a nonuniform grid in gene-space to encapsulate clusters of data points. Together, the grid lines form hypercubes, each representing a different state of the GRN. By using these grid lines as quantization thresholds, a logical network is constructed by finding the most statistically significant combination of regulators for every gene. Statistical significance is determined using the chi-squared test with a null hypothesis of independence between regulators and regulated genes. Quantization of several test data sets related to time course of brain gene expression following alcohol exposure have shown grid quantization to be more effective than marginal K-means clustering in capturing the true distribution and that grid quantization is effective at performing feature selection. From this quantization a statistically significant 16 node GRN was inferred. Preliminary verification of this network shows that the genes directly influenced by alcohol in the logical network are involved in pathways related to alcohol injection. The biological significance of these results is encouraging, and further refinement of these methods is merited as these methods are not specific to GRNs but applicable to any complex network. Future applications could include dynamic modeling of cerebellar development, cell response to exposure agents, and many other biological phenomena under investigation at ORNL.

Medical Data Analysis Using Visualization and Statistical Tools. LILA GHEMRI (Texas Southern University, Houston, TX, 77004) JONATHAN YOUNG (Pacific Northwest National Laboratory, Richland, WA, 99352)

Analyzing Intensive Care Unit (ICU) and Emergency Room (ER) Unit Records in Order to Potentially Reduce Medical Errors and Improve Healthcare Facility Performance. LAKEISHA D. MELTON (Texas Southern University, Houston, TX 77004) JONATHAN YOUNG (Pacific Northwest National Laboratory, Richland, WA 99352). By analyzing Intensive Care Unit (ICU) incident reports and Emergency Room (ER) Unit medical records using data analysis, data mining, and cluster analysis, a statistical representation and a data visualization of the analyzed data can support the discovery of complex and unanticipated relationships extant in the data. The primary focus of this project is to identify potential weaknesses and systematic problems throughout the datasets of patient medical records and potential unanticipated process behavior in order to potentially sustain improvements in the reduction of patient medical errors and the enhancement of healthcare facility performance. A tool called IN-SPIRE TM, is used to explore each type of data (incident reports and medical records) individually and thoroughly to find the complex relationships in the records. The analysis tools were also used to analyze clusters of records, categorize and group those records in order to potentially produce data visualizations and statistical representations of the analyzed data.

Mobile Data Collection Applications: A Proof of Concept. JONATHAN CHANG (University of California, Berkeley, Berkeley, CA, 94720) TIM LOWE (Lawrence Livermore National Laboratory, Livermore, CA, 94550)

This project’s goal is to provide a proof of concept for mobile data collection applications, and identify the best ways such applications could be implemented and used. Such an application should decrease the time and resources users now need to devote to redundant data processes, and provide an easy of locating and retrieving data at a later time. The two types of available mobile devices, Personal Digital Assistants and Tablet Personal Computers, each have their particular strengths that suggest themselves for certain types of applications. As such, parallel data collection applications have been developed, with a common web application for uploading information to the database. While these aspects have been developed and proven, it still remains to refine these applications, develop the tables to hold their data, and field-test with users for their feedback.

Modeling a Stereoscopic Satellite as Part of a New Way to View the Fundamental Teachings of Astronomy. NITAI FINKELSTEIN (Suffolk County Community College, Selden, NY, 11794) DAVE STAMPF (Brookhaven National Laboratory, Upton, NY, 11973)

Maya™, a computer software package that creates 3-Dimensional renderings of objects that a user designs, appears to be a useful tool in expressing the theories and laws utilized in Astronomy. The goal of this project was to accurately represent four astronomical scenarios using Maya’s™ animations as a teaching tool. These animations were created to simulate what is believed to be occurring in astronomical phenomena. For the animations, a man made satellite was designed to travel from Earth to a super nova remnant (SNR), where it used its various instruments to study the phenomena. While traveling through the SNR, the satellite got caught by the gravitational forces of a black hole, throwing it into a wormhole. This is a theoretical tear in space-time connecting two black holes. The satellite then exited through the other end of the wormhole where it observed an explosion of a star, also known as a supernova. When studying Astronomy, it may sometimes be difficult to grasp a certain concept or fact. Teaching with the aid of 3-Dimensional (3D) animations may help students better understand what they are studying. When viewing an image of the SNR in 3D, there is depth in that image, making it easier to pinpoint and study specific parts of the SNR, something that could not be done with an ordinary 2-Dimensional image. With the modeling of the satellite using Maya™, its inner workings and how it was affected by the spatial phenomena it had encountered were able to be viewed with more ease than before. When all four scenes finished rendering, they were presented using 3-D Stereo Movie Maker. This software is used to translate videos into 3-D simulations. A special screen was needed to be used along with Stereoscopic glasses to properly view this effect. While modeling in Maya™, it was found that the animations created could be useful tools for all levels of education and scientific research. They may also be utilized as a way to interest young Americans in science and technology, furthering our hopes that our children will once again be leaders in the scientific community.

Monitoring Temperature and Fan Speed Using Ganglia and Winbond Chips. CAITIE MCCAFFREY (Cornell University, Ithaca, NY, 14850) YEMI ADESANYA (Stanford Linear Accelerator Center, Stanford, CA, 94025)

Effective monitoring is essential to keep a large group of machines, like the ones at Stanford Linear Accelerator Center (SLAC), up and running. SLAC currently uses Ganglia Monitoring System to observe about 2000 machines, analyzing metrics like CPU usage and I/O rate. However, metrics essential to machine hardware health, such as temperature and fan speed, are not being monitored. Many machines have a Winbond w83782d chip which monitors three temperatures, two of which come from dual CPUs, and returns the information when the sensor command is invoked. Ganglia also provides a feature, gmetric, that allows the users to monitor their own metrics and incorporate them into the monitoring system. The programming language Perl is chosen to implement a script that invokes the sensors command, extracts the temperature and fan speed information, and calls gmetric with the appropriate arguments. Two machines were used to test the script; the two CPUs on each machine run at about 65º Celsius, which is well within the operating temperature range (The maximum safe temperature range is 77º -82º Celsius for the Pentium III processors being used). Installing the script on all machines with a Winbond w83782d chip allows the SLAC Scientific Computing and Computing Services group to better evaluate current cooling methods.

Multi-threaded Communication in ZeptoOS I/O Daemon (Zoid) on the IBM BlueGene/L Supercomputer. JASON KOTENKO (Edinboro University of Pennsylvania, Edinboro, PA, 16444) KAMIL ISKRA (Argonne National Laboratory, Argonne, IL, 60439)

ZeptoOS is an open source operating system for the IBM BlueGene/L supercomputer. The BlueGene/L has two types of nodes, I/O and compute nodes. I/O nodes handle requests from compute nodes to read and write data. The communication between compute and I/O nodes is currently handled by a daemon named Zoid, which replaces a proprietary daemon developed by IBM. Zoid currently gives unequal access to the multiple compute nodes associated with each I/O node. This is a problem for parallel applications which rely on synchronized data transmission between compute nodes. A message queue has been implemented to enforce synchronization of internal communications. An optimization currently being developed is to use multi-threading in order to hide latency in I/O requests. Multi-threading allows a running process to divide itself into multiple light-weight processes which share memory space. By allowing multiple I/O operations to be handled at the same time, the processor time previously lost to latency can be used to execute pending requests.

New Algorithms for the Network Protocol Independent Performance Evaluator. CALEB KLAPP (Westminster College, Fulton, MO, 65251) TROY BENJEGERDES (Ames Laboratory, Ames, IA, 50011)

The Network Protocol Independent Performance Evaluator, (NetPIPE) has been in development for over 5 years. As the website http://www.scl.ameslab.gov/netpipe/ describes it, “NetPIPE is a protocol independent performance tool that visually represents the network performance under a variety of conditions. It performs simple ping-pong tests, bouncing messages of increasing size between two processes, whether across a network or within an SMP system. Message sizes are chosen at regular intervals, and with slight perturbations, to provide a complete test of the communication system.” The original NetPIPE program was designed with a single-pass algorithm using exponentially increasing message sizes that exits when a time limit to transfer a message was reached. In order to improve on this design, a multi-stage algorithm was implemented. This new program uses a pan and scan method to accurately derive a representative equation for bandwidth at all message sizes. Some preliminary tests have been performed on well behaved networks where the accuracy of the program's calculations can be assessed. Comparing the experimental data with the curve generated by the program has provided strong evidence that the program produces valid performance information as well as statistical analyses of the network performance data which was not available with previous implementations.

Providing Additional Context with MapDisplay. LINDA SATO (University of Oregon, Eugene, OR, 97403) DAVID MILLARD (Pacific Northwest National Laboratory, Richland, WA, 99352)

While an important feature of visualization and analytical tools is to give users the ability to manipulate the context in which they view their data, this functionality can normally be found to some extent in all but the most basic software tools. At the same time, the importance of having multiple representations of the data, however, can be underestimated by the designer, and its usefulness, overlooked by the user. The focus of this project was to design and implement a prototype of a map-display component for an existing software application to provide users with an alternate view of their data. To simplify the project, an existing map-based display service for the actual display was incorporated into the design. The main resource for this project was the web-based version of Operational Status Board (WebOSB). WebOSB is an application that allows users to maintain a database of information and control how that information is categorized and displayed via a spreadsheet-like interface. One existing feature of WebOSB is the ability to export a snapshot, i.e. an immediate record of the rows of a table; this was implemented as an XML document. Thus, the main goal was to create a transformation process that would transform the XML document into a format the map-display service could read; the challenge was to give users some choices in how the data was displayed, providing a very basic means of filtering their data. The solution involved creating an XML configuration file containing values defining a mapping set by the user, and using Java to parse the file and create the target display document. Currently, the software we are using for the final display is Google Earth. The addition of the MapDisplay component to WebOSB makes it possible for users to interact with their data not only in a different context, but one which is user-defined, and it is the combination of the two views that provides WebOSB users with a more complete representation of their data.

Quality Assurance of Brookhaven National Laboratory's Accelerator Controls Devision 's Communications Protocols. EDWARD D'AZZO-CAISSER (St. Joseph's College, Patchogue, NY, 11768) DAN OTTAVIO (Brookhaven National Laboratory, Upton, NY, 11973)

The Accelerator Controls Division (ACD) manages the accelerators at Brookhaven National Laboratory (BNL). These machines, controlled by front end computers (FECs), require real-time interaction and precision to ensure the quality of retrieved information. These FECs can only be accessed remotely, using communication software (CS) which transfer information from programs to the FECs themselves. Since this software proved to be a major source of errors, we posit that targeting inconsistent behavior in program execution will reveal a variety of faults inside the CS. Traditionally, all communications testing was accomplished by visual observations and small test applications. However, using Java 2 Platform Standard Edition 5.0, we created a regression test, or suite of programs, to check the quality of ACD's Java protocols automatically. Since an exhaustive check of these protocols would result in an unfeasible amount of permutations and thus an inordinate amount of time, our regression test only interacted with higher software layers, looking for consistency and expected behavior of the CS. This is accomplished by three separate test functions: 1) a basic test to perform communication commands to a working FEC; 2) a test which communicates to a broken FEC; 3) a test which communicates to multiple FECs. The information gathered in these tests were used to locate faults within the comunication software. The testing suite was finished as planned; we created the tests we intended and we uncovered 10 faults in the CS thus far. The former system overlooked these program faults. Our new test offers a systematic way to test for inconsistencies and expedite the debugging process. We do not foresee any changes to the sweeps already in existence, and though the basic architecture will remain in tact, more automation could be implemented. The application of our regression test can help to improve the functionality of ACD's current communications software and any uprades made to it. The regression test can also be built upon to ensure a greater versatility, ensuring its usefulness over time.

RADIATION SOURCE TRACKING SYSTEM: USING VISUAL C++ .NET AND MEASUREMENT STUDIO. DAVID ENGBLOM (Western Michigan University, Kalamazoo, MI, 49008) RICHARD VILIM (Argonne National Laboratory, Argonne, IL, 60439)

Detection and tracking of radioactive sources is important to scientific study and national security. The increasing threat of terrorist activities in the world has fueled the need to develop new technologies to track dangerous materials. Current technologies are bulky and fixed. This project produced a portable and accurate Radiation Source Tracking system called RadTrac. The system consists of Sodium-Iodide (gamma-ray) detectors, USB hubs to connect the detectors, a laptop and a software program written in C and C++ using the Microsoft Foundation Class (MFC). Microsoft’s Visual Studio programming suite was used to develop the software. The program communicates with up to six detectors using manufacturer-provided software controls. The RadTrac program is able to track a radioactive source in real-time. The program has improved radiation source detection to within one foot and RadTrac’s ability to save detector readings into text files allows for post-processing and analysis of experimental data. Further error checking and an enhanced user interface are among the recommended improvements for future releases. Additionally, future versions of the program will include support for different detector types and guaranteed support for more than six detectors.

Scientific Application of 3D Modeling and Animation Utilizing Maya to Recreate the Formation of a Supernova Remnant. MICHAEL CASSANO 3RD (Suffolk County Community College, Riverhead, NY, 11954) DAVE STAMPF (Brookhaven National Laboratory, Upton, NY, 11973)

3-Dimesional (3D) modeling and animation software is an effective educational tool that can be used to visualize the unexplored universe. Maya™ was used to create 3D visualizations of spatial objects. When attempting to learn about an object that can’t be directly observed, take all of the known data and create a visual reconstruction to aid in future studies concerning that object. When new information is acquired, simply change the attributes in Maya™ to render an updated image. A 2.20ghz workstation utilizing 1gb of 400mhz memory running Maya™ 5.0, was used to create a three dimensional visualization of a Supernova remnant (SNR). Acquisition of data was necessary to depict the relative distance that the shocked ejecta are repulsed from the core as well as the different spectrum of light the expanse shows. High and low energy x-rays define the overall expanse of the SNR. X-ray spectra are needed to show the spatially resolved spectroscopy of the ejecta. SNR1987a gave the relative symmetrical pattern that SNR exhibit. Manipulation of Maya’s interface allows implementation of all data to be put into a Cartesian space to recreate a dying star. A 3D animated fly-through has demonstrated the chaos of the expulsed material that forms the SNR. A 2D render has been completed to illustrate a comparison to our telescopic images of a SNR. Additionally, a 3D animated model of an SNR has demonstrated the capabilities of Maya™ to recreate this interstellar medium (ISM) for educational purposes. To expand on this experiment we could calculate the radial expanse, and visually depict the temperatures of the ejecta. Possible future studies could examine in a 3D coordinate environment, the gravitational fields that a black hole of a SNR exhibits or the effects the magnetic field has on the surrounding area of the dying star. Further examination could lead to an understanding of the affects a SNR has on surrounding star systems. This software could expedite and enhance not only the effectiveness of theoretical and experimental ideas and concepts, but also enhance learning and the effectiveness of teaching tremendously.

SHAPESGUI: SHAPES-GUI: A GRAPHICAL FRONT END TO THE SHAPES SIMULATION TOOL. JOHN DIMOND (University of Illinois, Champaign-Urbana, IL, 61820) BOYANA NORRIS (Argonne National Laboratory, Argonne, IL, 60439)

SHAPES is a program that simulates the interaction of electromagnetic waves with twodimensional planar nano sized objects of varying composition and shape. The input for the program comes in the form of a humanreadable text file that contains all the parameters necessary to run a simulation. These parameters include particle geometries and positions, material properties, wave sources, the size and coarseness of the sampling grid, and what data should be output from the simulation. The readability of the input file makes the job of setting up and running a basic simulation relatively straightforward, but it is extremely cumbersome to input a complex simulation setup by hand. This project addressed this problem by developing an intuitive graphical program which creates the necessary input file. The SHAPES-GUI program allows the user to define the simulation domain, place objects on a grid, then resize and move them freely, as in a typical drawing application. It also provides simple dialogs that let the user set additional parameters of the simulation in a straightforward manner. This application will be expanded to create input files for the upcoming threedimensional version of SHAPES.

Simulation Environment for Digital Feedwater Control System. YURI MALITSKY (Cornell University, Ithaca, NY, 14850) TSONG-LUN CHU (Brookhaven National Laboratory, Upton, NY, 11973)

The Digital Feedwater Control System (DFWCS) of a nuclear power plant is responsible for monitoring and controlling the flow of feedwater in the secondary coolant loops of the pressurized water reactors. This control system is also responsible for automatically responding and alerting people when something goes wrong. Because of this crucial function to the safe operation of the power plant, it is necessary to evaluate the fault tolerance of DFWCS by testing situations that can disrupt the stability of the plant and whether they will be detected and dealt with appropriately. This project created a prototype of the simulation environment that allows the testing of the system with a virtual feedwater system (FWS) model. The system consists of two independent processes connected via network sockets. The primary process runs the original control algorithms and the second implements the input and output interface of the feedwater devices. To promote portability and modularity, the system was implemented with Java, linking to the original control algorithms through the Java Native Interface (JNI). The resulting prototype of the virtual model allows controls the input values to help identify critical situations and later the development of corresponding algorithms for analyzing the safety of the system.

SNS Beam Characterization Using Wire Scanner Analysis. FEDRICK REYNOLDS & BAKARI UNDEWOOD (Tennessee State University, Nashville, TN, 37209) TED WILLIAMS (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

The Spallation Neutron Source (SNS) is the most powerful pulsed source of neutrons in the world. The SNS consists of a Linear Accelerator (Linac) that accelerates negatively charged hydrogen ions (H-) throughout the various regions of the Linac to approximately 88% of the speed of light. These regions include the Medium Energy Beam Transport (MEBT), Drift Tube Linac (DTL), Coupled Cavity Linac (CCL), Super Conducting Linac (SCL), High Energy Beam Transport (HEBT), Accumulator Ring, Ring to Target Beam Transport (RTBT), and finally the Mercury (Hg) target. After the H- beam is accelerated, it is transported to the Accumulator Ring where all of the electrons in the beam are stripped off, thereby converting the beam to only protons. Once released from the Accumulator Ring, the beam then hits the Hg target and knocks out numerous neutrons from the Hg nucleus, which are used for research. Diagnostic equipment throughout the Linac provides data that are used to monitor, correct, and characterize the trajectory and shape of the beam. Some of these devices are wire scanners, which are used to measure the beam’s position. Furthermore, the wire scanner data are used to set the magnetic field strengths of the quadrupole magnets; these are used to focus the beam. This is particularly important to the beam-to-target transfer process. Without correct steering, the full benefit of the beam on the target could not be achieved. Using a variety of programs, including Wireanalysis and Graphical Analysis; Gaussian and bi-Gaussian fits are calculated to measure the beam sizes. The measurements of the magnets’ parameters are included in the data taken. The beam sizes and the magnets’ parameters are fed into the model, and the output generated gives a plot of the sizes of the beam at various positions in the MEBT. The data is also compared on the same graph. The disparity between the two indicates how the beam changes in the MEBT. Adjustments were made to better fit the data. For example, while using wire analysis, the program was only allowed to fit the data that represented a Gaussian profile. This changed some values, but ultimately resulted in a better fit of the data. Although the beam will never be characterized without any error, there is the possibility of characterization with minimal error. Therefore, it is imperative that this research continue.

The Tricks of Trigonometry. JENNIFER OBRAY (Brigham Young University- Idaho, Rexburg, ID, 83440) DON DUDENHOEFFER (Idaho National Laboratory, Idaho Falls, ID, 83415)

The Critical Infrastructure Protection project is tasked with the protection of physical and cyber security assets which are important to our nation. They are so important that the destruction or impairment of these assets would weaken our national security, the security of our national economic policies, and/or national public health and safety. For this portion of the project an information portal was created on the Intranet at the Idaho National Laboratory which contains subjects on Critical Infrastructure Protection. The portal is a means to access and add content that pertains to this important and growing field. Since recent disasters such as the attack of September 11, 2001 and Hurricane Katrina, along with other threats, we have realized the weight of this protection program and we have been working to make the Critical Infrastructure Protection project stronger. The information portal that I have created gives people an opportunity to find the information they need pertaining to different subjects involving the protection programs. The information provided in the portal was given to me by my mentor, Don Dudenhoeffer, who had previously decided what to include in the Intranet portal. I then took that information, organized it, and added it to the portal in the proper format. For example, there are models of Hurricane Katrina evacuation plans to provide information for people who may be working to improve the evacuation time or process. This useful portal is not only a place to find information but it is also a place to add other information. There is an upload form under each of the different subjects that allows anyone with useful information relating to the subjects to be able to upload it onto the page so that other people may use it in their research. This project will increase people’s ability to quickly find the information they need in order to make decisions about national security and protection programs, thusly protecting us and the physical and cyber assets of this nation.

Three Dimensional Hermite Interpolation for Boundary Integral Analysis. THEODORE SAVICH (Earlham College, Richmond, IN, 47374) LEONARD J. GRAY (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Partial differential equations, e.g. the Laplace equation, can sometimes be transformed into integral equations on the boundary of the domain. These boundary integral equations can then be solved numerically by employing simple approximations of the boundary and boundary functions. Generally, linear (first order) polynomials have been sufficient, yielding continuous interpolations of the geometry and the boundary functions. However, for a variety of applications in heat transfer, electrostatics, and elasticity, there is interest in achieving a smoother interpolation, and this has lead to the investigation of a cubic Hermite approximation. In particular, a differentiable Hermite interpolation may provide improved accuracy for moving boundary simulations wherein one would like a smooth representation of the evolving surface. Differentiability of the surface interpolation is achieved by incorporating the known normal vectors to the surface at the nodes. A differentiable interpolation for the primary function requires the function derivatives at the nodes; these are obtained by solving integral equations for the gradient of the primary function simultaneously with the original boundary integral equation for the primary function. The use of cubic elements complicates the analysis, as now the boundary integral kernels involve the square root of sixth degree rather than quadratic polynomials. As a consequence, the analytic integrations employed in the linear analysis are not immediately possible. The integrals must now be split into a non-singular piece which can be evaluated numerically and a singular piece which must be integrated analytically. The splitting for the Hermite interpolation has been developed, and test calculations on simple problems have determined that the three-dimensional Hermite algorithm is a viable approach for determining the potential and flux in the Laplace equation. The implementation of the gradient equations is presently underway.

Upgrading the Video Crystal Centering System at the National Synchrotron Light Source. KEVIN RYAN (State University of New York at Albany, Albany, NY, 12222) DIETER SCHNEIDER (Brookhaven National Laboratory, Upton, NY, 11973)

The National Synchrotron Light Source allows hundreds of scientists to examine the structure of biological crystals using x-ray beams and sophisticated detectors. Data collection can take hours to complete while the researcher has to adjust and recenter the crystal to ensure proper collection. Currently, an analog camera attached to the microscope lens sends video to an Axis Joint Photographic Experts Group (JPEG) engine, which inefficiently compresses the images and broadcasts them over the network. This project will install a Sony digital camera coupled to a video server, to capture high resolution images, convert them to Portable Network Graphics (PNG) format and stream them over the network. To control the camera a C program was written using the open source libdc1394 application programming interface (API) to capture the images and set camera features. From there the images will be losslessly compressed using the libpng API which was customized to output the data to a web server. This multi threaded program needs to coordinate the capture and compression precisely to create seamless image streaming from the server with only a half second delay. The advantages of using the PNG compression algorithm is the lossless compression and the existing web code for the Axis server can be reused with only slight modifications. Allowing the client to access the images through a web browser makes this project completely platform independent, eliminates the need to install unique software on the clients computer and lets us update/change the web applications functionality with ease. With this upgrade the sharper images can be more efficiently processed to provide accurate automatic crystal centering to be implemented. Completing this project will enable the researchers to collect their data more efficiently from remote locations, and the automatic centering will allow them to concentrate their attention on the important task of analyzing the data.

Using Neural Network Procedures to Classify Spatial Data from the Columbia River Basin. ANDREW PARELIUS (Piedmont Virginia Community College, Piedmont, VA, 22902) LANCE VAIL (Pacific Northwest National Laboratory, Richland, WA, 99352)

The Columbia River is a primary resource in the Pacific Northwest, providing water for irrigation, power generation, and biodiversity. Improved models of the catchment dynamics in the region will lead to the more efficient utilization of this resource. Creating more accurate models requires using a larger quantity of information which causes bottlenecks when using traditional information processing techniques. Using Digital Neural Networks (DNN’ s), a procedure to classify subregions of the Columbia Basin based on layers of accumulated data such as terrain aspects, climate, meteorology, snow pack etc., allows for these large amounts of information to be integrated into a model. The data is first accumulated into a server which allows the use of scripts to retrieve the appropriate resource for classifying a given subregion. Self-Organizing Feature Maps (SOFM’s) are an appropriate implementation for this classification procedure. Given the appropriate training rules, SOFM’s can generalize on their input vectors to identify to which class a region belongs. It is expected, that by fine tuning the DNN, a model will produce results with a high degree of accuracy which can be tested against actual field measurements. These results may be used by various organizations in the region to better manage the waters of the Columbia River Basin.

Utilizing Google EarthTM to Achieve Accurate and Efficient Data Distribution Analysis. JANE MEYERS (East Tennessee State University, Johnson City, TN, 37614) TOMMY NELSON (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

Although computing languages such as PL/SQL and SQL facilitate database querying, they do not always provide clear and/or intuitive results. Thereby to achieve efficient data analysis, most developers find it necessary to rely on additional expensive software. In an attempt to tackle this problem, database queries were manipulated to integrate with free software. The Google Earth™ software provides a free, accurate, and user-friendly interface for observing spatial distribution and performing GIS (geographical information systems) analysis. Google Earth’s™ file format, KML, requires a simple syntax comparable to that of standard HTML. It is easy to generate KML files based on query results, the difficulty lies in dynamically creating a KML file and storing it in a manageable format. To generate Google Earth™ files “on the fly”, a PL/SQL procedure is run upon query submittal that parses the results and inserts the appropriate KML tags. The resulting file is stored in a database table and provides users with a way to manage and view past and present files. These Google Earth™ files can then be saved or opened automatically to provide efficient and scalable views of the data’s spatial distribution. To enhance the developed functionality, custom Google Earth™ placemarks are created to represent each of the different data classifications. This feature allows the user to observe overall data distribution as well as the distributions based upon specific database fields. Future work with these integrated database tools will extend the functionality to incorporate data attributes, in addition to geographical location. These enhancements will allow for a larger field of comparison.

Web Design and Development. KENNETH SHELTON (Pellissippi State Technology Community College, Knoxville, TN, 37933) JAN ANDERSON (Oak Ridge National Laboratory, Oak Ridge, TN, 37831)

The nature of this project involves the creation and maintenance of the internal and external web-sites for the Nuclear Science & Technology Division (NSTD). The purpose of these sites is to fulfill the information exchange needs of the NSTD. The information exchange occurs via two methods at NSTD, the traditional internet website or by the utilization of the Microsoft tool Sharepoint® which also utilizes the internet. The technological tools used to facilitate this project included Dreamweaver 8 to generate the HTML code, F

Computer Science Abstracts: