7#$"qB"p$$$$$28888 Bx* 2* 7$JB access Spring 1995 vol. 9 no. 1 National Center for Supercomputing Applications o University of Illinois at Urbana-Champaign Cover: Helping to make companies more profitable is just one aspect of NCSA's new center-wide Commercial Applications Program, which joins NCSA staff and resources with business people, NCSA's industrial partners, and UIUC academic units including the College of Commerce and Business Administration. Cover produced on a Macintosh IIci using Adobe Photoshop, Aldus FreeHand, and a UMAX UC630 scanner. (Concept and illustration, John Havlik; concept, Fran Bond) Table of Contents departments 2 ncsa contacts 3 editor's note 26 center activities 32 book review 4 Let's get down to business: NCSA's Commercial Applications Program research 6 Shaping tomorrow's Internet 8 Antibodies and Antigens 11 Unraveling cosmological mysteries 14 Mapping the body's electrical fields 16 Network development--NCSA's link to the future new technology 18 World Wide Web servers at NCSA 20 NCSA Mosaic update education 22 Crazy about you, Kid: John Ziebarth makes plans for NCSA's Education and Outreach program 24 NSF grant networks Illinois schools industrial program 25 NCSA holds partner user meeting ERRATA, access, Fall 1994: Radha Nandkumar, Carol Burwell, and Vicki Halberstadt contributed to Kenneth Chang's article, "Cornucopia of CRAY Y-MP science," page 9. Spring 1995 access (ISSN 1064-9409) is published by the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign (UIUC) with support from the National Science Foundation, the Advanced Research Projects Agency, other federal agencies, corporate partners, the University of Illinois, and the State of Illinois. Permission to reprint any item in access is freely given, provided that the author and access are acknowledged. Copyright (c) 1995 Board of Trustees of the University of Illinois. Editor: Fran Bond fbond@ncsa.uiuc.edu Managing Editor: Melissa LaBorg Johnson melissaj@ncsa.uiuc.edu New Technologies Editor: Ginny Hudak-David Copy Editors: Ginny Hudak-David Paulette Sancken Designer: Linda Jackson Printing by Flying Color Graphics Inc. This publication is printed on Evergreen Matte and Evergreen Matte Cover National Center for Supercomputing Applications 605 East Springfield Avenue Champaign, IL 61820-5518 phone: (217) 244-0072 FAX: (217) 244-1987 ncsa contacts GENERAL INFORMATION AND PROGRAMS Allocations Radha Nandkumar (217) 244-0650 allocations@ncsa.uiuc.edu Vicki Halberstadt (217) 244-5709 allocations@ncsa.uiuc.edu Applications Group/Faculty Program Melanie Loots (217) 244-2921 mloots@ncsa.uiuc.edu Chemistry User Group Balaji Veeraraghavan (217) 333-2754 balajiv@ncsa.uiuc.edu Communications Maxine Brown (217) 244-0072 maxine@ncsa.uiuc.edu Community Outreach Coordinator Scott Lathrop (217) 244-1099 scott@ncsa.uiuc.edu Computing and Communications Charles Catlett (217) 333-1163 catlett@ncsa.uiuc.edu Consulting--HPC and Mosaic (217) 244-1144 8:30 a.m.-5:00 p.m. Central Time consult@ncsa.uiuc.edu Education and Outreach Program John Ziebarth (217) 244-1961 ziebarth@ncsa.uiuc.edu Industrial Program John Stevenson (217) 244-0474 Media Relations John Melchi (217) 244-3049 jmelchi@ncsa.uiuc.edu Media Technology Resources Tony Baylis (217) 244-1990 tbaylis@ncsa.uiuc.edu media@ncsa.uiuc.edu (services) NCSA Operations Sue Lewis (217) 244-0708 slewis@ncsa.uiuc.edu (217) 244-0710 (services/help) NCSA Receptionist (217) 244-0072 FAX (217) 244-1987 NCSA Security Officer Michael Smith (217) 244-7714 msmith@ncsa.uiuc.edu Networking network@ncsa.uiuc.edu Orders for Publications, NCSA Software, and Multimedia Debbie Shirley (217) 244-4130 orders@ncsa.uiuc.edu Publications Group Melissa Johnson (217) 244-0645 melissaj@ncsa.uiuc.edu Software Development Group Jae Allen (information) (217) 244-3473 jallen@ncsa.uiuc.edu Training Program Alan Craig (information) (217) 244-1988 acraig@ncsa.uiuc.edu Deanna Spivey (registration) (217) 244-1996 deannas@ncsa.uiuc.edu User Services Lex Lane (217) 244-0642 lex@ncsa.uiuc.edu Visitors Program Jean Soliday (217) 244-1972 jsoliday@ncsa.uiuc.edu NOTE: All electronic mail addresses are via Internet. See inside back cover for details on anonymous FTP, Gopher, and World Wide Web. editor's note "And Springth the wude nu"--from Cuckoo Song, c. 1250 As the words of a centuries-old song remind us, the world (is) new in spring. In keeping with the season, this issue of access introduces some new ideas and people affiliated with NCSA. Using high-performance computing to solve real-world problems for the business community is the thrust of NCSA's recently formed Commercial Applications Program--a collaborative undertaking with several UIUC colleges and NCSA's industrial partners (see page 4). An innovative federally funded project to organize large databases of digital information on the Internet is described in "Shaping tomorrow's Internet" (page 6). Research utilizing some of NCSA's newer machines is covered in "Antibodies and antigens" (page 8), "Unraveling cosmological mysteries" (page 11), and "Mapping the body's electrical fields" (page 14). Read "Network development--NCSA's link to the future" to find out how NCSA stays at the leading edge in networking technology (page 16). NCSA's role in pioneering the technology of scalable information servers is outlined by some of its developers in "World Wide Web servers at NCSA" (page 18). "Crazy about you, Kid: John Ziebarth makes plans for NCSA's Education and Outreach Program" (page 22) introduces NCSA's newest associate director as he conceptualizes efforts to advance computational technology in classrooms and the community. A report of NCSA's first Industrial User Meeting is on page 25. Rounding out this issue is the coverage of a variety of center activities, beginning on page 26, and a review of two recent books about the Web and NCSA Mosaic on page 32. --Fran Bond, Editor LET'S GET DOWN TO BUSINESS: NCSA'S COMMERCIAL APPLICATIONS PROGRAM BY SARA LATTA High-performance computing is not just for engineers and scientists anymore. It is also for the corporate executive who wants to detect unusual customer buying patterns, like the discount retailer who discovered that people tended to buy diapers and beer on Friday after work. Place the two items next to each other, and voila!, sales shoot up. It's for the CEO of a major credit card company who wishes to market a product to a targeted group of consumers without sending expensive mailings to every credit card holder in the country. The results? Cut mailing costs by a factor of twenty and get a sixfold improvement in response rate. These and other commercial applications are often impossibly slow or just plain impossible to carry out with mainframe computers. Enter the NCSA Commercial Applications Program, a new centerwide venture in cooperation with the UIUC's College of Commerce and Business Administration, Departments of Computer Science and Mechanical and Industrial Engineering (of the College of Engineering), and NCSA's industrial partners. "NCSA intends to build a major effort in scalable infoserving," says NCSA Director Larry Smarr, "just as we've done with scalable computing for science and engineering. Our staff is excited about the opportunity to explore the applications of both data serving and data mining." It is a program whose time has come, according to Joe Blackmon, the program manager of NCSA's Industrial Partner Program. "I really believe that the application of high-performance computing to this area will have a much bigger impact than it has had on engineering--not that we are going to slack up on our engineering emphasis." Kem Ahlers, Caterpillar's manager for university relations at NCSA, concurs. "We [Caterpillar] see a lot of opportunities to do business simulations--what makes companies work, what makes them profitable--in much the same way that we have done product simulations here." PUTTING BUSINESS TO THE TEST The Commercial Applications Program includes an integration testbed, designed to allow industrial partners to prototype new systems before taking them to their own corporations. At the heart of the testbed are the new Cray Research Superserver 6400, the Thinking Machines CM-5, the CONVEX Exemplar, and the SGI POWER CHALLENGE Array [see access, Fall 1994, pages 4-8]. NCSA staff will examine the performance of emerging commercial information software, coupled with practical problems supplied by the industrial partners. By using real-world problems supplied by the industrial partners, the results will be representative of actual, not theoretical, performance. One of the key issues in solving commercial problems is data storage and management. NCSA's Database Group has installed three major pieces of database management software on the testbed (see sidebar, page 5). Partners who bring database problems to the testbed will be able to benchmark their applications software. Assume that a financial corporation like J.P. Morgan--one of NCSA's 11 industrial partners--wants to collect a large amount of data and use that data to predict bond prices or exchange rates. Data management software is essential of course, but the testbed will also contain the predictive tools necessary to make such projections. These predictive tools, based on intelligent algorithms (see sidebar), recognize patterns in billions of pieces of data and use these patterns to make predictions about the data. BRINGING IN FACULTY AND STUDENTS The Commercial Applications Program is not a one-way street for industrial partners only. In fact, it is more like a freeway with lanes for faculty and student research, curriculum development, vendor participation, and potential commercial spin-offs as well. Faculty from the College of Commerce and Business Administration may form research collaborations with NCSA staff. They may choose to work with the industrial partners on a particular research problem, participate in seminars or workshops, or incorporate HPCC commercial applications into their curriculum. Undergraduate or graduate students from the college who want to learn more about business and high-performance computing might do an internship at NCSA. "There's a clear need from our point of view to link with technology applications that are at the cutting edge," says Howard Thomas, dean of the College of Commerce and Business Administration. "There's an analytic tradition in this school [the Business School] that marries well with the tradition of engineering and science departments on campus. The corporations are a reflection of where our students will be employed. We should be trying to work with them not just to develop initiatives that are at the cutting edge of applications, but to translate those initiatives back into the curriculum. "I'm a great believer in relationships between science and business," Thomas continues, "and I think the comparative advantage of a business school in Illinois is to link with the traditional strengths of a land grant university." One of the missions of a land grant university, Thomas explains, is to promote economic and structural developments in the state. "It strikes me that the link between ourselves [the college], NCSA, the engineering college, and the industrial partners is a re- interpretation of the land grant mission for the twenty-first century. It creates the architecture and the framework in which people can invent and develop new projects. It creates a whole new set of opportunities." And, adds Thomas, the potential for commercially viable products. Cray Research Inc., Thinking Machines Corp., Silicon Graphics Inc., and Convex Computing Corp. have committed teams of experts in database and business applications to lend expertise to the Commercial Applications Program. EDUCATION AND WORKSHOPS NCSA will host a series of commercial applications workshops for interested industrial partners, faculty from the computer science and mechanical and industrial engineering departments and the College of Commerce, national researchers, and vendors. "Workshops of this kind," says Melanie Loots, associate director for NCSA's Applications Program, "are crucial in bringing together industrial users with real-world problems; academic researchers in business, finance, and computer science who provide a theoretical framework; vendor experts on the available hardware and software; and NCSA programming staff with expertise in developing parallel applications." The first workshop, "Commercial Applications and Parallel Processing," will be on April 24-25 in the Beckman Institute. NCSA IS A DOOR TO UIUC High-performance computing is not just for scientists and engineers anymore. It is for scientists and engineers and the business people who work with them to develop new computer applications. "NCSA is our door into the university," says Caterpillar's Ahlers. "We bring intellectually stimulating problems to the academic world; and in return, we get access to the technology they have developed." s Sara Latta is a freelance science writer. SCALABLE SOFTWARE SOLUTIONS Here is some of the software that will be available on the commercial applications testbed. Database management o ORACLE parallel query and parallel server o SYBASE o INFORMIX Intelligent algorithms o Classification and regression trees o Neural network systems o Simulated evolution and genetic systems Melanie Loots (Photo by Thompson-McClellan) Howard Thomas (Photo by Thompson-McClellan) RESEARCH SHAPING TOMORROW'S INTERNET BY CHRISTOPHER ADASIEWICZ Interspace--that's what NCSA Research Scientist Bruce Schatz calls the Internet of tomorrow. Instead of connecting computer servers only, as today's networks do, Interspace will connect information. Interspace is both a metaphor and a concrete plan for a more organized, more useful Internet. In October 1994, NCSA and UIUC researchers began a four-year federally funded project to develop a small-scale pilot Interspace, a database of the full text of selected engineering and science books and periodicals. Once the system is up and running, students and faculty at UIUC and other Big Ten universities will be able to access the electronic collection via a special version of NCSA Mosaic, which NCSA's Software Development Group is designing for the project. BRINGING ORDER TO INTERNET Schatz and his team are part of a nationwide effort called the Digital Library Initiative (DLI), so named because it seeks to bring order to the Internet just as libraries and librarians have brought order to books and periodicals. With the aid of a card catalog and shelves organized by the Dewey Decimal System, one can locate a specific book from among the UIUC Library's 15 million volumes in a matter of minutes. The same cannot be said about data on the Internet. The Internet and applications like Gopher and NCSA Mosaic "are easy access to information in terms of browsing, but to get to a specific piece of data quickly is a different story," says Joseph Hardin, NCSA associate director for software development. "There's nothing magical NCSA Mosaic alone can do to organize the Internet," Hardin says, which is why researchers are focusing on ways to improve the Internet itself. DLI aims to set protocols for publishing and organizing Internet information and to develop advanced linking and searching techniques. Connecting information, not just servers, will mean users can, in effect, "ask the Internet a question," Hardin says, "and get a useful answer." Schatz is the principal investigator in the four-year Illinois DLI project, supported by grants from the National Science Foundation, NASA, and the Department of Defense's Advanced Research Projects Agency. In October 1994, these institutions awarded more than $24 million to six universities for DLI work. THE NCSA-UIUC COLLABORATION Locally DLI research is a joint effort among NCSA, the UIUC Department of Computer Science, the University Library, and the UIUC Graduate School of Library and Information Science, where Schatz is an associate professor. The Illinois project is two parts. The first involves developing a small-scale testbed of the Interspace concept. The second is conducting research into the technology and user sociology of a digital library--incorporating some of the findings into the existing testbed and making recommendations for future directions. Researchers at UIUC's new Grainger Engineering Library Information Center, led by Director William Mischo, already have started gathering materials for an online collection of 10,000 engineering and science journal articles, complete with pictures, graphics, and citations. Several publishers will provide documents in SGML (Standard Generalized Markup Language) format that can be directly downloaded to the testbed server. The publishers include the Institute of Electrical and Electronics Engineers (IEEE) Computer Society, American Institute of Aeronautics and Astronautics (AIAA), American Physical Society (APS), American Society of Civil Engineers (ASCE), American Society of Agricultural Engineering (ASAE), Institute of Physics (IOP), and John Wiley and Sons. Other publishers are in negotiation. Collaborating with publishers for the testbed documents not only frees the researchers from having to create data for the testbed, Schatz says, but also jump-starts industry on standardizing and organizing Internet information and sets a precedent for other publishers to follow. "Eventually, industry will have the lead in providing information electronically, so it only makes sense to involve them from the beginning," Schatz says. The documents will be combined with existing sources of information, including major periodical indexes in science and engineering and the BIMA Grand Challenge database in radio astronomy [see access, Spring 1994, page 10]. These resources will be transparently linked to the testbed. EXAMINING OTHER ISSUES During the four years of the project, thousands of University of Illinois students and faculty at all campuses, students and faculty at other Big Ten schools, and NSF MetaCenter users are expected to access the testbed. Hence, the testbed affords researchers opportunities to study in real time the technical and sociological issues related to organizing and searching Internet information. Larry Jackson, NCSA Mosaic technical manager, will lead a team of software developers to create an enhanced version of the application that will enable users to access the testbed's SGML documents. Jackson's group is exploring how to design an interface that meets the needs of sight-impaired users. By setting rules for how the interface and the information it retrieves look on the screen, the researchers can run NCSA Mosaic alongside applications that translate video into audio. NCSA software developers also are looking into ways NCSA Mosaic can track the operations its users perform. This technology will enable the digital library interface to collect data on how people utilize the system. The project's sociology group, led by Library and Information Science Assistant Professor Ann Bishop (principal investigator) and Associate Professor of Sociology Leigh Star, is using surveys and focus groups to assess the needs of the digital library user community and make recommen-dations to the technical developers. Already the group has interviewed potential DLI users-- undergraduates, graduate students, and faculty--and drawn early conclusions. "One of the most important features people want," Bishop says, "is the ability to see information on a screen as if they were pages in a book. People tell us they remember information by where they saw it." Bishop says users want the ability to customize search protocols, display characteristics, and the interface itself. "And they want easy access to charts and figures. Faculty and students say they read these first," she says. OTHER INSTITUTIONS RESEARCHING Five other universities are pioneering DLI projects: Carnegie Mellon University, the University of California at Berkeley, the University of California at Santa Barbara, the University of Michigan, and Stanford University. Scientists at these institutions are investigating issues such as how to catalogue and search video and multimedia documents. These universities' projects seek to organize different kinds of data--environmental information, earth and space data, geographical images, and maps-- as opposed to engineering journal articles. The DLI project is a collaborative one among these schools. Researchers from each project will meet periodically to share their developments; the first such meeting is scheduled for the UIUC campus this spring. Using NCSA Mosaic, the groups are keeping in touch with each other by posting project notes and papers on the World Wide Web. Each DLI project has its own home page, all accessible from the URL on the opposite page. OTHER NCSA INVOLVEMENT In addition, other NCSA researchers are working on DLI-related projects. NCSA's Mike Folk, a researcher in the Software Development Group, is developing digital library technologies that will make it easier for the public to access astronomical and earth science data. Also working on the three-year project, funded with a grant from NASA, are NCSA Research Scientists Richard Crutcher and Robert Wilhelmson (Applications), Research Programmer Nancy Yeager (Computing and Communications), and Schatz. Folk's work with science data, as opposed to journal text, is representative of the challenges researchers face if they truly want to extend the Interspace concept to the Internet--the need to deal with many different kinds of information. s Christopher Adasiewicz, a senior in Journalism at UIUC, is a student intern in the NCSA Director's Office. Bruce Schatz (Photo by Thompson-McClellan) DLI projects: http://www.grainger.uiuc.edu/dli Antibodies and Antigens BY HOLLY KORAB There's an old saying about curing the common cold that Biophysicist Shankar Subramaniam likes to recall when explaining why he models antibodies and antigens: "With medicine, it takes a week to cure a cold; without it, seven days." During the early stages of infection, says Subramaniam, "your body is really trying to learn and grope and adapt to the foreign particle that is coming in. It's a way of maturing to a stage where the immune system is able to tackle the infection." In the past several decades, biological and medical research has deciphered many of the biochemical reactions by which the human immune system recognizes a particle as foreign and launches an attack. Where things get fuzzy, however, is in deciphering the basis underlying this recognition. This knowledge is essential to accelerating relief and to targeting medicines more accurately. To fine tune the process of biological recognition, NCSA Research Scientist Subramaniam is leading a team of computational biologists and computer scientists at NCSA who are modeling how one of the body's key defensive mechanisms, the antibodies, recognize and bind with one of the largest of microbial invaders, the antigens. Recently, the group successfully computed the complete nonlinear Poisson-Boltzmann equation and combined the electrostatic steering forces obtained from this equation with the diffusive motion of Brownian dynamics. The Poisson-Boltzmann equation is a fundamental equation in electrostatics once considered so difficult to solve as to be intractable. Previous efforts had solved the linear version of the equation, which does not account for the electrostatic affects of the ionic medium on the antigen's diffusion in the blood stream. The added realism of the nonlinear equation made it possible for Subramaniam and his colleagues to model the monoclonal antibody HyHEL-5 binding to the hen-egg lysozome at a rate that replicated experimental results--a first for antibody-antigen modelers. THE IMMUNE WAR Every time you eat, every time you breathe, foreign particles enter your body. Your immune system is called upon to destroy an estimated billion different types of potential invaders, each of which can bind only to a lymphocyte uniquely configured to complement the invader's chemical composition. Amazingly, your immune system is usually successful. The key to the system's success--as well as the reason for its seemingly slow pace--is its capacity to be flexible yet specific. Rather than having a billion different lymphocytes patrolling your blood vessels, the immune system sends out a few lookouts who, upon detecting an invader, rapidly sort through its genetic library to construct the suitable T cells or B cells to replicate and eventually subdue the invader. T cells are lymphocytes that originate primarily in the thymus and destroy fungi, viruses, and parasites by engulfing and digesting them. B cells are formed in the bone marrow and give rise to antibodies, which confer resistance against most bacteria. Antibodies are Y-shaped chains of polypeptides that are identical for a small section of amino acids at the forked end of the Y. The attractive and repulsive interactions between the tip of the fork of the antibody and the reactive portion of the antigen determines whether the two will bond weakly or strongly. The binding affinity between an antibody and antigen is the basis of molecular recognition. IDENTIFYING THE ENEMY Subramaniam's group is identifying antibody's "association motifs"--the binding affinity associated with a genetic arrangement--by measuring the rates of reaction for the monoclonal antibody HyHEL-5 binding with the hen-egg lysozome. After establishing its benchmark rate, they make slight alterations in the genetic composition of the antibody and measure its effect on the rate at which this "mutant" binds with the antigen. In this way they can detect and record the subtle changes in reaction rates caused by changes in the protein's genetic arrangement. Their model depicts an antibody surrounded by an ionic medium. Antigens were placed around the edges of a box at 500,000 different points and, one at a time, allowed to diffuse randomly just as they would in blood. If an antigen bonded strongly (in the correct orientation) with the antibody during one of these trajectories, Subramaniam considered the reaction over. If it bounced back or wandered off, the encounter was considered unsuccessful. By averaging these trajectories, he generated a rate of reaction for the antibody-antigen encounter. Because it would be impossible to model the entire antibody-antigen complex, only the 4,000 atoms involved in bonding were modeled. To model the two forces that propel antigens toward antibodies--the electrostatic forces represented in the full nonlinear Poisson- Boltzmann equation and Brownian motion--Subramaniam teamed up with Faisal Saied, assistant professor in the UIUC Department of Computer Science, graduate student Michael Holst, and postdoctoral associate Richard Kozack. The new method that arose from their collaboration employed a multigrid technique, a methodology that more efficiently solves equations on grids by resolving them at different levels of detail. The large memory required for this equation is amenable to a solution on NCSA's CONVEX C3880, or C3. The linear Poisson-Boltzmann equation had been solved before using a single-mesh approach, but not with a nonlinear technique. The existing methods were so slow as to discourage people from doing so. "Our method is so fast that it translates into a new capability," says Saied, who developed the algorithm for running the equation on NCSA's machines. Once the electrostatic forces were obtained, Kozack and Subramaniam used Brownian dynamics to model the diffusional motion of the antigen to the antibody. "Brownian motion is what we call an embarrassingly parallel equation," says Subramaniam in explaining why they chose to model this force responsible for diffusion on NCSA's Thinking Machines CM-5 and SGI POWER CHALLENGE. Brownian motion is the seemingly random motion of small particles as they are buffeted about by collisions with other molecules in a solution. Each of the 500,000 trajectories were basically calculations of Brownian motion. By combining the nonlinear Poisson-Boltzmann equation with Brownian motion, Subramaniam says they now have quantitative rates of reaction they can use to look at absolute values of reaction rates. RATIONAL DESIGN Obtaining quantitative reaction rates is the first step along a path to rational design and engineering, says Richard Willson, an associate professor of chemical engineering and biochemical and biophysical sciences at the University of Houston, who was one of two biologists whose experiments verified Subramaniam's model. Rational design, he says, "will constitute genuine protein engineering rather than where we stand now, which is often so unpredictable that is has been referred to as 'protein terrorism'." Understanding how a protein's structure influences its actions is essential for developing immunotoxins, biosensors, and in industrial applications such as bacterial organisms genetically altered to degrade oil spills. Who knows? It may even help biologists beat Mother Nature in curing the common cold. NOTE: Papers based on this research have appeared in the following journals. R. E. Kozack, M. J. D'Mello, and S. Subramaniam. "Computer modeling of electrostatic steering and orientational effects in antibody-antigen association." Biophys. J. (in press). M. Holst, R. E. Kozack, F. Saied, and S. Subramaniam. 1994. "Treatment of electrostatic effects in proteins: Multi-grid-based Newton iterative method for solution of the full nonlinear Poisson-Boltzmann equation." Proteins: Structure, Function, and Genetics 18(3): 231-245. M. Holst, R. E. Kozack, F. Saied, S. Subramaniam. 1994. "Protein electrostatics: Rapid multigrid-based Newton algorithm for solution of the full nonlinear Poisson-Boltzmann equation." J. Biomol. Struct. Dyn. 11(6): 1437-1445. S. P. Slagle, R. E. Kozack, and S. Subramaniam. 1994. "Role of electrostatics in antibody-antigen association: anti-hen egg lysozyme/lysozyme complex (HyHEL-5HEL)." J. Biomol. Struct. Dyn. 12:439-456. R. E. Kozack and S. Subramaniam. 1993. "Brownian dynamics simulations of molecular recognition in an antibody-antigen system." Protein Science 2:915-926. s Holly Korab is a science writer in the Publications Group. The complex between the antibody fragment HyHEL-5 and hen-egg lysozyme. The key amino acid residues involved in complexation are shown (large spheres). The negatively charged amino acids are red; the positively charged ones, blue. The small spheres highlight other charged residues in the antibody fragment and hen-egg lysozyme. (Courtesy Shankar Subramaniam) Shankar Subramaniam (Photo by Thompson-McClellan) Unraveling cosmological mysteries BY KENNETH CHANG From how gravity tugs at stars and galaxies, scientists know there's far more to the universe than what they see. Astronomers have puzzled for more than a decade over the nature of this dark matter and how much of it there is. Even what little of the universe that can be seen offers yet more mysteries to cosmologists. MODELING DARK MATTER--COLD AND HOT To try to understand the universe, ghosts and all, researchers led by Joel Primack, physicist at University of California, Santa Cruz, and Anatoly Klypin, astronomer at New Mexico State University, Las Cruces, have taken advantage of the multigigabytes of memory on NCSA's CONVEX C3880. They have created a universe-within-a-computer that looks much like the real thing. "We're working on the edge, and we're not sure any of the assumptions about dark matter are right, but experimenters at Los Alamos have observed neutrino oscillations indicating that neutrinos have mass in the range required by our models. It's beginning to look promising," Primack says. "And if it does work, it will be a triumph that cosmological observations and supercomputer calculations actually predicted the neutrino mass." That recent surveys by the Hubble Space Telescope failed to find dark matter does not concern Primack. "That has nothing to say about the quantity of dark matter," Primack says. "It just says it's not in the form of dim stars." For theoretical reasons, many astrophysicists believe the average density of matter in the universe is likely to equal a value called the critical density. A denser cosmos eventually stops expanding and then collapses into the "Big Crunch," while a sparser universe disperses into nothingness. The critical density is the dividing line--the universe still expands forever, but the rate of expansion is continually slowing down. The problem is when astronomers look up in the skies, many see a sparse universe. Adding up visible matter with estimates of the dark matter based on indirect measurements, some come up with an answer only one-third, or less, of the critical density. Primack and his colleagues, however, do not see a crisis. In fact, through their computer simulations, they think they may understand enough of the unseen part of the universe to be able to explain the part that is seen. "Every single measurement people have cited to show that the universe has less than critical density, we have reproduced in our simulations," Primack says. Primack's theory divides dark matter like a water faucet: hot and cold. Hot dark matter whizzed at close to the speed of light in the early universe; cold dark matter sloshed about more sluggishly. Galaxies coalesced from slight lumps in the cold dark matter. A decade ago, several researchers, including Primack, put forth a model with just cold dark matter. It worked well, but predicted too many galaxies. To fix the earlier model, hot dark matter was added--roughly one part hot for each three parts cold. The high velocities of the hot dark matter would have smoothed out the lumps somewhat and trimmed galaxy production to some degree. Data from recent experiments could shed light on the hot dark matter. The big bang created a torrent of ghostly particles called neutrinos--in a gallon there are about a million of these primordial specks--and there are indications that two of the three types of neutrinos have a smidgen of mass. With these values plugged in, the cold plus hot dark matter model correctly produces such numbers as the average speed of galaxies and the density of galaxy clusters. Computer simulations by NCSA Research Scientist Mike Norman and NCSA Research Assistant Greg Bryan have shown the x-ray properties of galaxy clusters in the cold plus hot dark matter model agree with observation as well [see access, Summer 1994, pages 3 and 27]. Given its success so far, Primack is cautiously hopeful. The torrent of data now coming in from ground- and space-based telescopes will rigorously test the predictions of the competing cosmological models. "If any theory can survive," Primack says, "maybe it's even true." SIMULATING GALACTIC CHASMS Astronomers mapping out the night sky have found the visible universe a richly textured fabric: long threads of galaxies woven into wide, thin sheets with immense, shadowy, mysterious chasms of apparent nothingness lying in between. Those observations would seem to be a wonderful validation of a theory put forth decades ago by Russian astrophysicists that predicted such intricate formations--except research-ers have long since discarded that theory. It does not correctly describe how smaller objects, namely galaxies, came into being. Conversely, the contemporary picture for the evolution of the universe, which does work well in explaining galaxies, does not explain the voids, threads, and sheets. "There's an apparent contradiction here," says Adrian Melott, professor of physics and astronomy at the University of Kansas, Lawrence. Based on thousands of hours of simulations, first on NCSA's CRAY-2 computer and now the CONVEX C3880, Melott offers a resolution to the paradox. In addition to clarifying the qualitative picture of the universe's history, the work of Melott and his collaborators provides an important tool that allows researchers running computer simulations to quickly test and refine specific theories. Melott's picture of the universe melds together the "pancake" theory proposed by Russian theorist Yakov Zel'dovich in the 1970s with the "hierarchical clustering" model assembled by American and British theorists. In the latter view, galaxies, which have about 100 billion times the mass of the Sun, formed first. "Things just fall together and merge, and then they merge with bigger things," Melott says. Galaxies gather together to form clusters, which, in turn, clump into superclusters. Conversely, in the pancake model the first structures are much larger, formed from blobs with the mass of hundreds of thousands of galaxies. Because the blobs would not have been perfectly even spheres, Zel'dovich envisioned that the mass would contract more quickly in one direction, forming a pancake-shaped object. Zel'dovich and his research group--including students Sergei Shandarin, now a professor at the University of Kansas, and Klypin--developed a mathematical approximation that rather accurately describes this pancaking process. But Zel'dovich's group had no means to visualize what came out of the equations. "What they had imagined was a disk here, a disk there, everywhere a disk, disk," Melott says. When the Russians were able to model their ideas on computers, they found something different. "That demonstrates the power of a computer model to surprise you," Melott says. "You do get thin surfaces, but they are in an interconnected network"--filaments and sheets. By the early 1980s, the pancake theory had fallen out of favor. Then in 1985, astronomers measuring the distance to thousands of galaxies found the same filament and sheet structures. "It was a big shock for some people," Melott says, "because it looked like nothing they had expected. Instead, it looked like predictions from the pancake theory." Meanwhile, during a 1983 visit to Moscow, Melott, Shandarin, and collaborators found pancake-like features in the first computer simulation of the cold dark matter model. Pancakes began popping up in other people's computer simulations as well! They proved to be a generic feature throughout a modern family of models, including the cold plus hot model favored by Primack. EXCHANGING IDEAS What Melott has done is adapt Zel'dovich's method for hierarchical clustering models. This modern updating of Zel'dovich's approximation theory best describes the evolution of superclusters--sheets and filaments included--in all kinds of models. Primack and colleagues have used Melott's approximation to test out variations of his cold plus hot dark matter theory. "You can take at least a first look at a lot of models and see if they're okay," Primack says. Klypin was a postdoctoral research associate at the University of Kansas, alongside Mellott and Shandarin, when he did his pioneering work on the cold plus hot model. Melott says that the approximation should also allow researchers to look at much larger models, which are currently limited by the amount of mem-ory and time available on a computer. Melott is organizing a one-day session at this June's conference of the American Astronomical Society that will look at the dynamics of large-scale structures. s Kenneth Chang is a doctoral candidate in physics from UIUC and is currently enrolled in the Science Writing Program at the University of California, Santa Cruz. GRAND CHALLENGE COSMOLOGY CONSORTIUM (GC3) One goal of the Grand Challenge Cosmology Consortium (GC3) is developing parallel versions of the most frequently used algorithms for cosmological simulations developing new, more powerful algorithms that expand the range of physical length and mass scales that can be simulated in a single computation. Many of these codes will be made publicly available in 1995. The consortium uses NCSA's CM-5 for simulating the formation of galaxy clusters and compares them with observations of clusters in the universe [see access, Spring 1994, pages 14-16, 28]. In January 1995, members of the GC3 convened at NCSA under the sponsorship of NSF. Those attending were Principal Investigator Jeremiah Ostriker, (Princeton University), Ralph Roskies (Pittsburgh Super-computing Center), Dennis Gannon (Indiana University), Edmund Bertschinger (MIT), and Lars Hernquist (University of California, Santa Cruz). NSF visitors were Morris Aizenman, executive officer of Astronomical Sciences Division; Loretta Inglehart, centers program director, Mathematics and Physical Sciences Directorate; and Richard Kaplan, centers program director for the Division of Advanced Scientific Computing. Mike Norman (NCSA/UIUC) hosted the site visit. GC3 is funded through a five-year HPCC Program Grand Challenge grant of $3.2 million. GC 3: http://zeus.ncsa.uiuc.edu:8080/GC3_Home_Page.html Primack's research group: http://physics.ucsc.edu/groups/cosmology.html Distribution of galaxies in two cosmological simulations showing (left) Cold plus Hot (neutrino) Dark Matter, (right) Cold Dark Matter. Blue represents 1011; red, 1012.5 solar masses. Filaments are sharper in the CHDM simulation; clusters are bigger in the CDM. Computations on NCSA's CONVEX C3880; visualizations on Almaden Research Center's IBM Power Visualization System. (Courtesy of Joel Primack and Anatoly Klypin) Joel Primack (Photo by Don Harris, Courtesy University of California, Santa Cruz) Adrian Melott (Photo by R. Steve Dick, Courtesy University of Kansas, Lawrence) Melott and collaborators have shown that there is a continuous transition between the Zel'dovich pancake and hierarchical clustering pictures in the universe. The top row represents the pancake model; the bottom left, hierarchical clustering. Clustering in the universe has proceeded in the way represented in the bottom right and middle images. (Courtesy Adrian Melott) MAPPING THE BODY'S ELECTRICAL FIELDS BY HOLLY KORAB Hundreds of times each second, the brain sends electrical impulses racing through the body's web of nerve cells to the motor neurons, where they initiate the chemical reactions that cause muscles to contract. About a century ago, scientists recognized that these excitation currents produce an electrical field which can be detected as small voltages in the skin or scalp. By measuring changes in the patterns of the body's electrical activity, researchers could detect some forms of heart disease and neurological disorders. Electrocardiograms of the heart (ECGs) or electroencephalograms of the brain (EEGs) measure these voltages; however, they provide physicians with only a snapshot of heart or brain activity. These glimpses help doctors spot disorders but are not always sufficient for diagnosing them. For the latter, doctors turn to other techniques; in rare cases, to surgery. Such is the case with some abnormal heart rhythms (arrhythmias). For patients who do not respond to drug treatment or pacemakers, cardiac surgeons often must open the chest and examine the heart with a roving, hand-held electrode or a flexible grid of surface electrodes to determine just where to operate. For temporal lobe epilepsy, neurosurgeons determine whether a patient who is not responding to medication has an operable form of the disorder by opening the cranium and attaching electrodes directly to the brain to identify whether the disordered electrical activity is highly localized (thus operable) or diffused over the entire brain. REVOLUTIONIZING DIAGNOSTIC PROCEDURES A team of computer scientists, physiologists, and physicians at the University of Utah are developing a diagnostic tool so that doctors may not have to undertake risky preoperative procedures. Using NCSA's POWER CHALLENGE, Christopher Johnson, associate chairman of computer science; Robert MacLeod, assistant professor of internal medicine; Peter Heilbrun, chairman of the Department of Neurological Surgery; and John Schmidt, research associate of computer science, are simulating the electrical fields emanating from the heart and brain to replace the snapshots from ECGs and EEGs with full-scale, 3D models. Johnson is team leader for Utah's Scientific Computing and Imaging Group (SCI). "Our goal is to develop these techniques to the point where we can provide information from painless measurements made from the body's surface," says MacLeod. "Now, this information is only available through highly invasive diagnostic procedures." The process employed by MacLeod and Johnson works like this. To "see" the electrical activity on the surface of the heart, technicians attach electrodes to a patient's skin at optimally selected sites across the chest and limbs. As in ECGs, these electrodes detect the changes in voltage that accompany each heart beat. Replacing the 3 to 10 electrodes in a conventional ECG with 16 to 192 electrodes, a process known as "body surface potential mapping," or BSPM, produces a more detailed, contoured map of electrical activity on the entire surface of the thorax for every instance in time. Although these body surface maps reveal more about the underlying electrical activity of the heart than does the standard ECG, the signals are somewhat smoothed and lack definition. To estimate the electrical activity right at the heart, the electrical readings are then incorporated into a large-scale finite element model of the human thorax constructed from MRI scans. Because the shape and constitution of the thorax affect how heart activity is transmitted to the skin, the researchers include all tissues and organs, including fat, in their model. Simulations of brain activity are similar. By placing an array of scalp electrodes, ranging from the standard 32 to up to 128, the researchers record the electrical activity from the brain (the EEG) on the surface of the head. They use these EEG recordings along with a geometrical model of a patient's head to localize abnormal electrical activity. Before electrical imaging procedures start showing up in clinics, however, the SCI team has more work to do. The geometric modeling and the process of "seeing" back (an inverse problem) involve tremendous amounts of computations that now require supercomputers or distributed high-end workstations. To hone their electrical imaging technique as a diagnostic tool, Johnson must develop more efficient mathematical and computational techniques. Early results have the team excited about the possibilities, assures Johnson, "but they are not there yet. "We have developed software for this project using C, C++, and Fortran. All software has been written by the SCI Research Group over a period of several years," says Johnson. "As is the case with most computa-tionally intensive users, we have used a variety of computers, including IBMs, SGIs, Crays, and Suns. We compute on just about anything we can get access to. In the recent past we were using clusters of IBM RS/6000s for the numerically intensive work, such as mesh generation and finite element analysis, and SGI machines for geometric modeling and visualization. "Now that SGI has produced such a powerful set of computing and graphics hardware, we can do both computing and visualization on a single platform. This has allowed for a straightforward transition between various high-end SGI workstations and NCSA's POWER CHALLENGE." COMMERCIAL APPLICATIONS Recently industrial biomedical researchers have shown an interest in techniques being developed by Johnson and his colleagues. This past year the SCI group began simulating the various configurations of electrodes and stimulation impulses for an experimental internal cardiac defibrillator designed by engineers at Pacesetter. The internal defibrillators are implanted in the chest with electrodes placed near a patient's heart, where they automatically detect and regulate an arrhythmic heart beat. As the heart begins to beat irregularly, the defibrillator applies a small jolt of electricity that returns the heart to its normal beat. A patient described the jolt as feeling like someone had given him a hard tap on the back, but when he turned around no one was there. "He knew he had just had a potentially life-threatening arrhythmia, but he was still walking around," says Johnson. Another use for this technology is in conjunction with cardiac pacers, says Johnson. Someday implanted electrodes may replace standard pacemakers to correct electrical abnormalities in the heart. DEVELOPING AN INTERACTIVE ENVIRONMENT These applications are only the start, insists Johnson. "We are now bringing together geometrical modeling, numerical analysis, large-scale computing, and scientific visualization to develop a 'computational steering package' that will allow users to design bioelectric devices and measure their effectiveness in an interactive graphical environment." Using interactive graphical input devices, engineers will be able to design biomedical devices, place them directly into the computer model, and automatically change parameters and boundary conditions as well as the level of mesh discretization needed for an accurate finite element solution. Johnson says that Utah's computational steering package could be adapted for use in other areas of research as well. WORKING IN NCSA'S ENVIRONMENT "The POWER CHALLENGE coupled with the mass storage system facilities at NCSA has been an invaluable resource for our large- scale simulations," says Johnson. "Otherwise it would have been extremely difficult, if not impossible, for us to conduct our research in such a timely fashion. The large memory, disk space, and multiprocessing capabilities of the POWER CHALLENGE has given us the opportunity to explore complex defibrillation scenarios that would have been impossible with standard workstation facilities. The payoff from these simulations is that engineers are now designing the next generation of implantable devices based on the results from these complex simulations. NCSA's POWER CHALLENGE provides us with a tool that allows us to explore a wide variety of designs to hone in on the most promising ones." NOTE: Work on the interactive steering system was recently presented by Johnson during a plenary talk at Supercomputing '94. s Holly Korab is a science writer in the Publications Group. Christopher Johnson and the SCI: http://www.cs.utah.edu/~sci Christopher Johnson (Photo by Terry Newfarmer, University of Utah; Courtesy of University Communications, University of Utah) Computer models from SCI, clockwise from upper left: Cutaway models of thorax in defibrillation studies (left) and heart (right) showing electrodes in green cavity and pink vessel. Color- coded lines indicate current density near the heart in which a tetrhedralized slice (blue) is superimposed around the heart (red). Finite element model of a head profile showing current densities on the surface of the scalp, skull, and brain. (Courtesy Christopher Johnson, SCI, University of Utah) NETWORK DEVELOPMENT--NCSA'S LINK TO THE FUTURE BY THOMAS KRAWCZYK Access to computers, workstations, and peripherals would be slowed greatly if it were not for cutting-edge local-area networks that tie them all together. While NCSA is pioneering the use of high- performance computing in the research environment, it is pushing the envelope of networking as well. "We're working with a wide range of projects that involve anything from local-area to wide-area networking and using several different technologies," says Randy Butler, technical program manager of the Networking Development Group. Along with team members John Quinn, Vijay Rangarajan, Von Welch, Paul Zawada, Patrick Dorn, and Jon Dugan, Butler's goal is to keep NCSA poised at the forefront of networking. ATM REVOLUTIONIZES CONNECTIVITY One such project involves paving the way for the new wave of local-area networking called Asynchronous Transfer Mode (ATM). ATM is a new global standard that many think will revolutionize connectivity. "It's the first big international standard that's been accepted by both the data and telecommunications vendors," explains Butler. "The protocol was designed to function at various rates, from 1.5 Mbps [megabits per second] all the way up to 9.6 Gbps [gigabits per second] and beyond without modifications. That's a real change from a lot of the other technologies in the past. It should make this migration [described below] a lot easier." In the past, any time physical media needed upgrading to a higher transfer rate, the protocols that ran on it also had to change. With ATM, changing from the current T3 connection (45 Mbps) to an OC3 line (155 Mbps) linking the NSF supercomput-ing centers will involve only a change of interface cards in the routers. None of the protocols, applications on the hosts, or applications on the routers have to change. This capability can be carried up to OC192 (9.6 Gbps) without having to modify the ATM protocol in any way. "The architectures and protocols for local-area networks have always been different from wide-area networks," says Butler. "Now it can be more seamless than it has been in the past, because ATM scales between the different physical media." Another advantage of ATM is that voice and video can run on the same physical media that carries data. "Right now, your data line has to be a separate physical medium than your voice line for your phone," says Butler. "ATM has the opportunity to bring those two together. You can take a single medium and use virtual circuits to segment it into separate lines for voice and data." Another ATM project involves NCSA's collaboration with UIUC's Computing and Communications Service Office, Department of Computer Science, and NCSA industrial partner United Technologies. "We're taking local-area ATM equipment and interconnecting it to examine performance and functionality," states Butler. "United Technologies has given us an ATM switch to use in testing multimedia capabilities between workstations." WIDE-AREA NETWORKING PROJECTS In wide-area networking, NCSA currently is involved in a project using ATM on the BLANCA gigabit testbed. The work is in collaboration with the University of Wisconsin-Madison, AT&T, Bell Laboratories, Sandia National Laboratory, UIUC's Department of Computer Science, and others. Developed from AT&T's Experimental University Network (XUNET), BLANCA spans the country at 45 Mbps, with the trunk from NCSA to Wisconsin's Madison campus reaching 622 Mbps. "We've just started a wide-area network project called the Very High Bandwidth Network Service, or vBNS," says Butler. "vBNS is an ATM-based network in collaboration with MCI and the NSF supercomputer centers. It will soon be online at 155 Mbps, with 622 Mbps planned for later. Right now we are ready to connect all of our FDDI [Fiber Distributed Data Interface] and HIPPI [High Performance Parallel Interface] networks. Eventually we will all have ATM local area networks, and vBNS will connect those." ADVANCING DISTRIBUTED COMPUTING AT NCSA NCSA's role in the advancement of HIPPI technology shows that there is more to distributed computing than developing fast wide- area networks that span the country. "HIPPI is a gigabit network technology we've been installing and integrating here for a couple of years," says Butler. "We have 14 nodes on our HIPPI network, and it is now a production-class network--far more reliable and available than a research network." NCSA's new SGI POWER CHALLENGE Array is also interconnected by the HIPPI network. The first five CONVEX Exemplar HYPERnodes are connected to the FDDI network. In the third quarter of 1995, they are expected to support HIPPI, when all 8 HYPERnodes are online. The CM-5 is already connected to the HIPPI network. The NCSA HIPPI network also connects to the BLANCA network and is connected through BLANCA at 622 Mbps to the HIPPI network at Madison. NETWORKING CONFERENCES The Networking Development Group was able to demonstrate its expertise in HIPPI development and management at Supercomputing '93 [see access, Spring 1994, page 36]. "Our team was responsible for the HIPPI network used at that conference," says Butler. "It represented the largest HIPPI network at the time--with 16 HIPPI switches and over 50 installed systems. We designed and installed the network, worked with all the exhibitors and vendors, supported it, and disassembled it afterwards." Again the Networking Development Group lent its expertise at Supercomputing '94, through HIPPI and ATM network management support and by serving on networking panels. The team also contributed to the network design and implementation for the Second International World Wide Web Conference '94 in October at Chicago. PUSHING THE EDGES If they are to keep pace with the growth in HPCC, Butler and his group have their work cut out for them. "Supercomputers have pushed the development of high-speed networking because researchers are able to create so much data. They also want to take advantage of each machine's unique capabilities through a distributed environment. Typically you don't have the resources you need at your own site, but if you had supercom-puters at both ends of the network, you could link into machines specialized for different types of work. It's a combination of large datasets and the specialization of machines that calls for better and better networks." s Thomas Krawczyk is a freelance science writer. NCSA's Network Development Group: (left to right) Randy Butler, technical program manager; Paul Zawada; Von Welch; Jon Dugan; John Quinn; Vijay Rangarajan; and Patrick Dorn (Photo by Thompson-McClellan) NEW TECHNOLOGY WORLD WIDE WEB SERVERS AT NCSA BY ROBERT MCGRATH, NANCY YEAGER, AND ADAM CAIN In early 1993, NCSA released NCSA Mosaic, considered by many to be the first useful Internet browser.[1] It has proved unbelievably popular--today hundreds of thousands use it to access information over the Internet. At roughly the same time, NCSA started a World Wide Web server. As Mosaic grew in popularity, the demands on NCSA's Web service exploded. Meeting this demand has been a serious challenge for two NCSA Computing and Communications (C&C) teams--Information Systems and Networks. The NCSA Information Systems Group pioneered the technology of scalable information servers, an economically and culturally crucial area. NCSA Mosaic opened the door and helped millions of people look for and get information over the Internet. The center also pioneered the scal-able server (information provider) side of this technology. The Information Servers Team is leading the way in investigating and developing new information server technology. With the explosive growth in requests for information from our server, NCSA has been forced to invent new technologies to keep up with demand. NCSA WEB SERVICE The NCSA server offers information on the center's HPCC systems, groups, publications, software, and, of course, the NCSA Mosaic Home Page. Members of the Web user community named NCSA's server as the Best Overall Site in the BoWeb '94 (Best of the Web) competition [see access, Summer 1994, page 22]. The number of requests for information from the center's Web server has increased continuously. As of this writing, the NCSA Web service receives more than 500,000 requests each weekday. During peak hours, the service gets 50,000 requests per hour, which is roughly 1,000 per minute or over 30 requests per second-- for hours on end! [2, 3] This phenomenal load has present-ed a significant technical challenge. No ordinary computer is designed to respond to 30 network requests per second for several hours at a time, and not many computers can do this at all. As the load has grown, individual computers used as servers have struggled and, inevitably, failed under this unprecedented stress. This experience gave the two C&C teams the opportunity to study how servers work under extreme pressure--whether we wanted it or not. Any single computer has an upper limit of Web traffic it can handle. Our load already exceeds the limit for most computers. We quickly realized that we needed to use more than one computer to meet the demand. The NCSA solution can be applied to other situations. Before we could use multiple servers, several key problems had to be overcome: o Addressing information--the "www.ncsa.uiuc.edu" address had to map to multiple computers o Balancing the load--requests had to be apportioned among multiple servers o Distributing information--each server needed all the information available from the service We have solved these problems by adding servers to keep up with demand. Starting with one server in February 1994, we had four servers in May, six in September, and eight in December. NCSA will continue to add servers throughout the coming months and, possibly, years. THE NCSA WEB SERVICE ARCHITECTURE Addressing information Web servers find documents through addresses called Uniform Resource Locators (URL) [4] that have the form: http://www.ncsa.uiuc.edu/filename.html. Part of the URL is a server's Internet Protocol (IP) address (in this case, www.ncsa.uiuc.edu). In order to connect to www.ncsa.uiuc.edu, a computer uses network software to translate the name of the computer into the numeric address of that computer on the network. The standard behavior is to translate one name into the address of one actual computer. How, then, could the single name www.ncsa.uiuc.edu be spread out over more than one computer? At NCSA this translation is done by a widely used version of the Domain Name Service (DNS). We needed to have the name www.ncsa.uiuc.edu refer to more than one computer, which was not possible with standard DNS software. So a new feature was added to the NCSA network name service: round-robin DNS. Whenever a computer asks the NCSA network name service "What is the IP address for www.ncsa.uiuc.edu?" it receives one of several addresses, in a simple rotation, rather than only one. [5] Load balancing Round-robin DNS effectively makes the name www.ncsa.uiuc.edu a logical address that resolves to several physical addresses. It also provides load-balancing among the several servers because the address of each server is given out just as often as every other server. It turns out that this is a rather imperfect load balancing mechanism because other systems use name-to-IP translations for a while rather than asking over and over for the same translation. Even with its shortcomings, round-robin DNS has spread the load fairly evenly among NCSA's multiple servers.[8] In short, it has worked--so far. Distributing information The NCSA Web service provides information from a tree of files containing HTML documents, images, audio files, and movies. In order to use multiple servers, each server must be able--one way or another--to retrieve all the documents. Otherwise NCSA would have several Web services, rather than a single one. Compounding the problem is the large and changing set of information being served: tens of thousand of files amounting to many megabytes of storage with new files added daily. One approach would be to replicate all the information to be served on each server. Clearly it would be very difficult and expensive to maintain identical copies of thousands of files on four, eight, or 32 servers. A second approach would be to divide the information among the servers. Some documents (notably the What's New pages) are more popular than others. These high-demand items present a serious challenge because the server (or servers) that offers the most popular information will often be overloaded, while the other servers will not be able to help. This rather defeats the purpose of adding extra servers. The solution to this problem is to cache a copy of the most used information on each of the servers. We solved the problem of distributing information by putting the document tree in the center's distributed file system, the Andrew File System (AFS) from Transarc Corporation. [6, 7] AFS provides a single consistent view of the file system to each Web server, so that each server sees exactly the same documents. This effectively replicates the whole document tree and because AFS caches files on the local disk, the most popular documents are generally available on local disk on each server. NCSA's multiple Web servers do not have any knowledge of each other. This independence allows NCSA to use a variety of hardware as servers. Almost any system that can be an AFS client can be made into an NCSA Web server in under an hour. The round-robin approach could be made to work with any efficient distributed file system. Much research remains to learn how well distributed file systems in general (and AFS in particular) work with large files, such as images or movies, and how well they will work in the long run. It may not be possible to precisely replicate our solution everywhere, but the basic idea should be widely applicable. References and URLs 1. NCSA Mosaic Home Page. URL: http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/ NCSAMosaicHome.html 2. Robert McGrath, "What We Do and Don't Know About the Load on the NCSA WWW Server." NCSA Colloquium (September 28, 1994). URL: http://www.ncsa.uiuc.edu/InformationServers/Colloquia/ 28.Sep.94/Begin.html 3. Mosaic Stats. URL: http://www.ncsa.uiuc.edu/SDG/Presentations/Stats/WebServer.html 4. Tim Berners-Lee, "Uniform Resource Locators (URL): A Syntax for the Expression of Access Information of Objects on the Network." URI Working Group (21 March 1994). URL: http://info.cern.ch/hypertext/WWW/Addressing/URL/url-spec.txt 5. Paul Albitz and Cricket Liu, DNS and BIND in a Nutshell. O'Reilly & Associates, 1992. 6. M. Satayanarayanan, "Scalable, Secure, and Highly Available Distributed File Access." IEEE Computer 23, no. 5 (May 1990). 7. NCSA AFS Users Guide version 2.1 (September 1994). URL: http://www.ncsa.uiuc.edu/Pubs/ NCSASysDir.html 8. Eric Katz, Michelle Butler, and Robert McGrath, "A Scalable HTTP Server: The NCSA Prototype." First International Conference on the World Wide Web, 1994. URL: http:/ www.ncsa.uiuc.edu/InformationServers/Conferences/CERNwww94/ www94.ncsa.html s Robert McGrath, Nancy Yeager, and Adam Cain are members of the Computing and Communications Group. OTHER WEB ACTIVITIES AT NCSA NCSA is a hub for research and development in information server technology including scientific information and images, online libraries, industrial applications, and community service. The projects are different, but they all share the need to break new ground in information services. C&C teams are currently involved in the URN (Uniform Resource Name) project that is establishing guidelines for unique document tags and locator information. They are also investigating digital commerce security issues, such as user authentication, digital signatures, and encryption as well as document access control. (Both will be covered in future issues of access.) Below are URLs and article references for some recent C&C collaborative projects covered in this and previous issues of access. CCNet: see article on page 24. Astronomy Digital Image Library URL: http://imagelib.ncsa.uiuc.edu/imagelib.html The Daily Planet URL: http://www.atmos.uiuc.edu Digital Library Initiative research: see article on pages 6-7 SHTTP (secure servers) standard URL: http://www.commerce.net/information/standards/drafts/shttp.txt NCSA's scalable Web service includes a collection of independent servers, a document tree in a distributed file system (Andrew File System), and a round-robin Domain Name Service. (Data supplied by Michelle Butler, NCSA C & C; Illustration by Marshall Greenberg) ncsa mosaic update NCSA MOSAIC AWARDS NCSA Mosaic has been recognized throughout business, industry, and academia as an award-winning product, and the list keeps growing. Larry Jackson, NCSA Mosaic technical manager, said at one recent awards ceremony: "Outside groups don't perceive the boundaries between client, server, Web, protocols, user interface. They tend to lump them all together. Accordingly, I expect this is not specifically a 'software developer's' award, but an award to anyone connected with the creation of the total package the user encounters." Industry Week's (December 19, 1994) Technology of the Year Award--one of six awarded this year--was bestowed on NCSA Mosaic in a ceremony at the center on February 2, 1995. Charles Day, editor-in-chief, made the presentation. (See page 21.) Information Week (December 19, 1994) ranked NCSA Mosaic at the top of its list of the Ten Most Important Products of 1994. NCSA Mosaic placed above Microsoft Windows NT, Apple Power Macintosh, Lotus Network Notes, and Fore Systems' ForeRunner. The December 12, 1994, issue of Fortune magazine lists NCSA Mosaic as one of the publication's Products of the Year: "This software is transforming the Internet into a workable web connecting data users and sources, instead of an intimidating domain of nerds." NCSA Mosaic shares the honor for 1994 with the Mighty Morphin Power Rangers, Olds Aurora, and Wonderbra. A panel of Apple Internet users from the Advanced Technology Group of Apple Computer Inc. in Cupertino, CA, awarded NCSA a Cool Tools Award for NCSA Mosaic for the Mac. The Berkeley Macintosh User's Group (BMUG)--the largest of the Mac Users Groups, with about 12,000 members--awarded NCSA and UIUC one of their Fall 1994 Choice Product awards for NCSA Mosaic at the 1994 Macworld convention in Boston. The December 1994 issue of Unix Review magazine, distributed in advance at COMDEX, gave NCSA Mosaic the Networking Award in their 1994 Outstanding Product Awards for a UNIX application. PC magazine (a Ziff-Davis publication) chose NCSA Mosaic for the 1994 PC Magazine Technical Excellence award in the Communications Software category. The award was presented by Rick Ayre, executive editor, and Nick Stam, technical director for hardware at PC Magazine Labs, during COMDEX. NCSA Mosaic won a Most Valuable Product Award from PC Computing magazine (a Ziff-Davis publication) in November 1994. The award--for the Most Valuable Product in the Communications/Network group, Online Services category--was given in Las Vegas at the annual COMDEX convention. John Dvorak and Dennis Hayes presented the 1994 Dvorak PC Telecommunications Excellence Award for Outstanding Multimedia Internet Utility to NCSA Mosaic in August 1994. The Dvorak PC Telecommunications Excellence Awards are presented by Dvorak and Hayes Microcomputer Products to "pioneers whose vision and commitment to telecommunications and online activities have helped increase the vibrant, expanding network of ideas and information that exists today." NCSA Mosaic was awarded the 1994 MacUser-ZiffNet/Mac Share- ware Award for Telecommunications during the Third Annual MacUser/Ziff-Davis Interactive Shareware Awards ceremony at the August 1994 Macworld Expo in Boston. InfoWorld magazine's editors named NCSA Mosaic the 1993 Product of the Year in Bob Metcalfe's Industry Achievement Award category [see access, Summer 1994, page 26]. "Our editors' Product of the Year Awards recognize superior achievement in 11 categories of personal computer hardware, software, and networking products. Coming as it does from men and women with a thorough, objective understanding of personal computer and networking technologies, your award represents a unique mark of excellence for your very distinguished product," states Editor-in-Chief Stewart Alsop. NCSA Mosaic wins InfoWorld's 1993 Product of the YearAward, 1994 MacUser-ZiffNet/Shareware Award, and 1994 Dvorak PC Telecommunications Excellence Award for Outstanding Multimedia Internet Utility (left to right). (Photos by Thompson-McClellan) 1994 Most Valuable Product Award from PC Computing and 1994 PC Magazine Technical Excellence Award. "You're helping to make our lives better. . . You're helping us get around on that [information] highway," said Charles Day, editor-in-chief of Industry Week, as he presented the publication's Technology of the Year Award (1994) to NCSA Mosaic. Accepting were (left to right) Terry McLaren, Larry Jackson, Briand Sanderson, [Day], Tom Redman, and Dave Thompson of NCSA's Software Development Group. (Photo by Tony Baylis, NCSA Media Technology Resources) NCSA Mosaic Trophy Case: http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/Awards/ MosaicAwards.html ITFC The Internet Technologies Federal Consortium (ITFC) brings together a group of federal agencies with an interest in the Internet, information tools, and NCSA. The goals of the ITFC are to assist federal agencies in reinventing their internal operations and information dissemination, provide a forum for technology transfer, help provide resources for continued NCSA Mosaic development, and give members the chance for direct input to NCSA developers. Participating agencies are: Central Intelligence Agency (CIA), Defense Technology Information Center (DTIC), General Services Administration (GSA), National Institute of Standards and Technology (NIST), National Library of Medicine (NLM), National Oceanographic and Atmospheric Administration (NOAA), Nuclear Regulatory Commission (NRC), National Security Administration (NSA), and National Science Foundation (NSF). The primary focus for NCSA is to provide the participating agencies with the support and technical assistance they need to solve problems of information dissemination and communication to enhance technology sharing. The work includes providing a WWW server with information specific to the consortium members; training; collaborating on small useful software applications using the CCI (common client interface); technology transfer; and watching the net for new, cool, and useful developments the agencies might apply to their WWW pages. education CRAZY ABOUT YOU, KID: JOHN ZIEBARTH MAKES PLANS FOR NCSA'S EDUCATION AND OUTREACH PROGRAM BY PAULETTE SANCKEN John Ziebarth has been aroundNCSA for a few months. In fact, he recently moved to his permanent office just three doors west of mine. We had met upon his summer arrival at NCSA, but prior to this interview contact was limited. Ziebarth's reputation preceded him. I knew he was lauded as an education and computer guru from the University of Alabama in Huntsville, and his arrival was eagerly anticipated by the NCSA education staff. As I prepared for our interview, I anticipated hearing Ziebarth's savvy strategies and precise plans concerning his vision for the Education and Outreach Program. I wasn't disappointed. COMPUTATION HERE TO STAY Ziebarth makes it very clear where he stands on technology and education: it's not a fad. He explains that theory and experimentation have been the mainstays of science and education research, but computation is now on equal ground. Each educational level is addressed in Ziebarth's plans. Graduate-level education has been using high-performance computing and doing computational science as a matter of course for well over eight years. Undergraduates are novices in this means of research; computational science is just beginning to play a part in high school and even middle schools. Training the educator has become a priority of the NCSA Education and Outreach Group. Training teachers, keeping in contact with them, making sure they have access to the latest technology--as well as an expert to consult with--has proven beneficial and far-reaching. The next great "frontier" will be the training of teacher educators so that eventually every new teacher will be prepared to use technology in the classroom. "The constant changes in technology will cause changes in education," says Ziebarth. "It's NCSA's job to stay involved and in-step with these changes and to adapt them to the education arena." Currently there are five NSF-funded programs being administered at NCSA. Plans are to pursue funding for the continuation of current programs and the addition of new programs. oThe Education Affiliates Program (EA), funded by NSF's Education and Human Resources (EHR), extends NCSA's outreach to K-12 institutions to learn how an HPCC center can partner with other oganizations to support teachers and educational improvement. oThe Resource for Science Education (RSE) Program, also funded by EHR, brings visiting educators to NCSA to interact with staff and scientists and to utilize testbed environments that NCSA helps to establish. oSuperQuest is a national computational science competition for secondary students and teachers designed to change high-school science and mathematics curricula by providing the resources and training necessary to stimulate the computational exploration of science [see access, Fall 1995, page 17]. oThe Networking Infrastructure for Education (NIE) program, funded by EHR, provides a testbed for deploying a variety of high- speed network technologies in rural and suburban environments. NCSA will evaluate the impact of infrastructure needs in networking, teacher enhancement, curriculum change, and ongoing support for these activities. oThe ChemViz program develops curriculum materials for teachers to cover modern atomic theory and introduces students to atomic bonding and molecular orbital theory [see access, Spring 1994, page 22]. WELL CONNECTED Ziebarth believes that wholesale access to technology by schools, businesses, and communities is only a matter of time and money. The content and quality of what is accessed is his real concern. "This is an area where NCSA needs to focus. What is the benefit of having this connectivity? How does educational information get there? How is it evaluated? Does it get evaluated? How does this information get disseminated?" asks Ziebarth. Answers to these questions will help drive the new direction in educational outreach. THE ENGAGEMENT Getting communities and businesses involved in technology is the other aspect of the Education and Outreach Group. The stakes are high to be in the first phalanx of "those that are connected" to the Internet. Although many are rushing to get on the Information Superhighway, what's the big hurry? Businesses may liken getting connected to buying a new car model--maybe it's best to wait a few years and let them get the bugs worked out before buying. Yet waiting may lead to repercussions down the line. Businesses and communities, like educators, need good online content and a reason to be connected. "This connection needs to be affordable and justifiable," Ziebarth reminds us. What goes on in this area in the next several years will dictate whether being a stop on the superhighway is profitable or not. If it isn't profitable--for both the consumer and the business--it won't last. "We can help educate the community and industry to make sure what they contribute to the Internet is needed, wanted, and has longevity," Ziebarth explains. Recently developed NCSA outreach projects include: oIllinois Learning Mosaic (ILM)--an online resource for Illinois educators, parents, and students that will provide information about access to technology-driven programs and courses from K- Ph.D. Ultimately every school in Illinois will be on the network and in the ILM database. oCCNet--a collaboration between NCSA and the Champaign County Chamber of Commerce designed to steer the county onto the global electronic superhighway. Areas being addressed are agribusiness, community and government resources and libraries, education, health care, small business, and Geographic Information Networks [see access, Fall 1994, page 24]. Under Ziebarth's direction, NCSA will continue to be an information and training resource for current outreach participants, while adding new dimensions and target groups to the program. MULTIPLICATION IN ORDER Ziebarth is also looking to double the number of full-time staff in his area. Team leaders will head up three major spheres of interest: (1) educational research, (2) science and curriculum issues, and (3) networking and outreach. Newcomers to this group will find a leader who is dedicated to advancing computational technology from the classroom to the boardroom. Those that come into contact with him through outreach efforts will discover a well-informed, dedicated educator. s Paulette Sancken is a public information specialist in the Publications Group. JOHN ZIEBARTH PROFILE oGraduated from Rantoul Township High School, Rantoul, IL oReceived bachelor's degree in physics and math, and master's degree in math from Eastern Illinois University oCompleted doctorate in aerospace engineering with an emphasis in computational fluid dynamics at Mississippi State University oWorked for NASA, Rockwell International, Colorado State University, Argonne National Lab, and University of Alabama, Huntsville oWorked most recently at the University of Alabama in Huntsville and with the Alabama Supercomputing Center, where he developed the Alabama Precollege Supercomputing Program (APSP), an opportunity for high-school students to learn about computational science and supercomputing. The APSP was adopted by the Department of Energy for its Adventures in Supercomputing Program and by NASA as its Explorations in Supercomputing Program oActed as Educational Chair for Supercomputing '94 oElected Educational Chair for Supercomputing '95 oRecently married Beth Ann Bucher, a chemistry teacher at Centennial High School, Champaign, IL oEnjoys golf and snow skiing John Ziebarth (Photos by Thompson-McClellan) NSF GRANT NETWORKS ILLINOIS SCHOOLS The UIUC is distributing ten grants totalling approximately $132,000 to local school districts and educational organizations to facilitate the creation and use of telecommunications networks in grades K-12. The money will also be used for curriculum development and teacher training. The award is provided from a one-year, $0.5 million grant received from the National Science Foundation by a collaboration of university units: NCSA, the College of Education, and the Department of Computer Science. The funding comes from NSF's Networking Infrastructure for Education (NIE) program. NCSA'S ROLE "We have a special interest in science and mathematics education," says Scott Lathrop, NCSA project coordinator. "We see this program as a collaborative effort to bridge the gap between laboratory science and classroom education to enable students to learn and perform science with the same tools used by researchers." The faculty from the College of Education bring many years of experience in curriculum development and teacher training to benefit this project. The Department of Computer Science faculty and students are providing their experience in computer networking to develop networking recommendations for the schools involved in the project. "Schools participating in the project," says Lathrop, "will help achieve NIE's goals for creating a testbed networked educational community throughout 1995." Lathrop says that the NSF grant came about in part because NCSA has studied the role of high-performance computing and communications in K-12 education since 1990. A series of workshops "generated momentum for the introduction of high-performance networking, computational science, and visualization into local and regional schools." In addition to training teachers, NCSA has made its facilities available as a "living laboratory" for community teachers to instruct their students. CCNET PROVIDES FOUNDATION Champaign County Network (CCNet) was developed by the Champaign County Chamber of Commerce to catalyze the creation of a telecommunications infrastructure with high-speed networking connections throughout Champaign County [see access, Fall 1994, page 24]. CCNet has a number of task forces focused on the uses of the network in the community. The CCNet Education task force is led by the superintendents of the Champaign and Urbana districts. The local superintendents and their staff have been collaborating with several UIUC units and CCNet to prototype networking solutions. This experience and foundation will be used for many of the school connections in the NIE project. Schools involved Grant recipients, selected from written proposals, are the Champaign School District, including Bottenfield Elementary, Centennial High School, Central High School, Columbia Elementary, South Side Elementary, and Westview Elementary; Urbana School District, including Wiley Elementary, Leal Elementary, and Urbana Jr. High School; Charleston School District, including Carl Sandburg Elementary and Charleston Jr. High School; Danville High School; Discovery Place, Champaign (science museum for children); Mahomet-Seymour Combined School District; Springfield District; Teutopolis High School; Unit Seven Schools in Tolono; and University Laboratory High School. s CCNET EXPANDS Three new Champaign County Network (CCNet) sites have been added to the system, Alaina Kanfer of NCSA's Educational Outreach Team recently announced. The Champaign-Urbana News Gazette's Urbana office and Champaign's Central and Centennial High Schools now receive data at high speed from the Internet via TV cable with a telephone return path. The Champaign-Urbana Convention and Visitors Bureau is slated to be connected into CCNet by April 1995. Local exhibits may be accessed from the WWW at either of the URLs given below. s Local exhibits: http://www.prairienet.org/ippages.html (Prairienet) http://www.prairienet.org/SiliconPrairie/info_hwy/w3servers.html (CCNet) industrial partners NCSA HOLDS PARTNER USER MEETING BY FRAN BOND The first Industrial Partner User Meeting was so well received that plans are already underway for a second meeting. Thirty participants from NCSA's Industrial Partners Program attended a day-long session at the Beckman Institute in November 1994 to prepare for hardware changes at the center [see access, Fall 1994, pages 4-7]. "The main focus of the meeting," said Joe Blackmon, program manager of NCSA's Industrial Partner Program, "was how NCSA can help industrial partner computational users transition from the CRAY Y-MP and the CONVEX C3880 systems to the new high- performance systems--the SGI POWER CHALLENGE and CONVEX Exemplar--and what applications will be available on these machines." NCSA Director Larry Smarr addressed the group on the direction of NCSA's scalable metacomputing environment. He wants the center's Industrial Partners to "get on the learning curve early" as NCSA transitions to a scalable microprocessor metacomputer. Moving into NCSA's new environment will be more satisfactory for users, Smarr pointed out as he stated that every job running on a CRAY Y-MP system would "fit into the memory of one processor on the SGI POWER CHALLENGE." Aspects of transitioning partner applications to the new systems, timelines for doing so, and system administration issues were introduced by Lex Lane, associate director of NCSA's User Services. John Towns, chair of NCSA's Computer Policy Committee, described the Exemplar and the CHALLENGE in terms of their software environments and vendor participation. Convex representative Scott Free and SGI representative Jeff McDonald talked about the benefits and features of their respective machines. The compatibility of popular third-party applications on the Exemplar and the CHALLENGE was stressed. After lunch, break-out sessions on a variety of topics were offered. Session topics and their leaders were as follows: computational fluid dynamics, Danesh Tafti; finite element analysis, Fouad Ahmad; chemistry, Balaji Veeraraghavan; molecular dynamics, Shankar Subramaniam; chemical engineering, Jay Alameda; business computing, Michael Welge; digital libraries, Bruce Schatz; and information systems, Larry Jackson. The meeting concluded with a discussion on NCSA's industrial user environment. The group recommended holding another meeting the next year. The user meeting was held in conjunction with the second Industrial Partner Advisory Council that convened on the following day. The major item passed from the user meeting to the advisory council was a discussion on how the partners and NCSA could work together to communicate to software vendors about applications and flexible licensing agreements. NCSA's Corporate Officer John Stevenson said, "I am especially pleased with the partner participation and the results of the meeting. It is important that we continue to receive this type of partner input, and hopefully we can have another meeting in 1995." s Fran Bond, editor of access, is a member of the Publications Group. Chemical Engineering Break-Out Session discussants were (left to right) Ken Bishop (University of Kansas); Willis Bell (Eli Lilly); Jay Alameda (leader of NCSA's Chemical Engineering Applications Group), session leader; Haruna Cofer (NCSA); and Linda White (Eli Lilly). (Photo by Tony Baylis, NCSA Media Technology Resources) The Tribune Co. becomes NCSA's 12th industrial partner . . . Read all about it in the next issue of access. center activities MILESTONES Honor NCSA research scientist Eric Jakobsson, biophysicist in the Applications Group, recently was elected a Fellow in the American Physical Society. "For the elucidation of ion transport through biological membranes by computer modeling of polypeptide, ion, and water motions" is the citation on his induction. Awards Ping Fu, NCSA research programmer in the Software Development Group (SDG), won the Best Paper Award at the 28th International Conference on System Sciences in the biotechnology systems track. Her collaborators are Herbert Edelsbrunner (UIUC computer science professor), Mike Facello (SDG), and Jie Liang (Applications Group). NCSA publications won three awards in the 22nd Annual Chicago Technical Publications, Art, and Online Communication Competition of the Society for Technical Communication (STC), Chicago division. access won second place in the art competition for magazine design (Fall 1993, Spring and Summer 1994). Collaborating recipients were Linda Jackson, designer; Fran Bond, editor; and Melissa Johnson, managing editor. For cover design, the Summer 1994 issue placed third. Winners were John Havlik, concept and illustration, and Bond, concept and research. "Focus on Fractals," designed by Jackson, took third prize in the poster category. Image creation was by Robert Panoff, NCSA senior research scientist, and Michael South, former NSF site intern at NCSA. Bond was invited to judge selected entries in the publications category. s Rich Kendall, Jim Long, and Scott Coyle (left to right), former Cray Research Inc. onsite representatives at NCSA to maintain the machines, were honorees at a farewell reception. Altogether, their time at NCSA totaled 17 years. Long remains at NCSA in the Computing and Communications Group. (Photo by Tony Baylis, NCSA Media Technology Resources) NATIONAL ACADEMY OF ENGINEERING INDUCTS SMARR NCSA Director Larry Smarr was inducted as a member of the prestigious National Academy of Engineering (NAE) for "leadership in high-performance computing and communications" in February. "I am thrilled with the election," Smarr said. "However, I think it is symbolically honoring all of those who have worked so hard to create the evolving information infrastructure in America and the world." Election to the National Academy of Engineering is among the highest professional distinctions in engineering, honoring those who have made important contributions to engineering theory and practice and demonstrated unusual accomplishment in pioneering new technologies. Smarr is one of 77 selected for membership this year. He was nominated by the NAE Section 5, Computer Science and Engineering. Smarr, a relativistic astrophysicist, is widely recognized as one of the principle catalysts in making supercomputing power available to academic researchers. In 1984 he helped convince the U.S. Congress to fund the program that created the National Science Foundation's supercomputing centers, one of which is NCSA. He was also influential in the development of a national network to connect the centers' processing power to remote university and industrial users. Before centers like NCSA were established, supercomputers were almost completely unavailable to the academic community. In the 1970s and early 1980s, Smarr traveled abroad or worked under clearance in national laboratories to run his research problems on supercomput-ers. Smarr, 46, joined the UIUC faculty in 1979 after conducting postdoctoral research at Princeton, Yale, and Cambridge Universities, and as a Junior Fellow at Harvard University. He holds a doctorate in physics from the University of Texas at Austin, master's degrees in physics from Stanford University and the University of Missouri, and a bachelor's in physics from Missouri. Smarr is a Fellow of the American Physical Society and the American Academy of Arts and Sciences. In 1990 he received the Franklin Institute's Delmer S. Fahrney Medal for Leadership in Science and Technology. With William Kaufmann III, he co-authored Supercomputing and the Transformation of Science, published in 1993 by Scientific American Library. s DISNEY HONORS LOCAL TEACHER "Walt Disney Salutes American Teacher Awards" honored Marcy Vancil, Urbana kindergarten and first grade teacher at Flossie Wiley School, in Washington, DC. Recognized as a teacher who represents the best in her profession, Vancil was one of three finalists in the Early Childhood Category. She has been involved in a number of workshops offered through NCSA's Education and Outreach Program and has participated in the networking of Urbana public schools. Out of thousands of applicants, 60 American teachers from the U.S. and abroad were chosen to represent 12 categories. Of that 60, 36 finalists--three in each category--were flown to Washington with their spouses. During National Education Week, they were treated like royalty as primary guests of Disney. "The most spectacular part of this honor was meeting the 35 teachers," says Vancil. "I have 35 new brothers and sisters in the profession." She says the group quickly bonded and "appreciated each other's uniqueness." While in the nation's capitol, the group chose one person to represent all teachers for the "Walt Disney Salutes the American Teacher" promotional campaign, which is sponsored by the Disney TV Channel and Campbell Soup. As part of the process, videos of 60 teachers selected in the first round were produced by Disney. Teachers were interviewed and filmed in their local schools along with students, parents, and school administrators. These brief videos (six and one-half minutes) are airing throughout the school year on the Disney Channel. Vancil's video will be shown in the fall. Showings of the awards ceremony continue through this year. An Educational Advisory Committee, selected from 20 professional organizations in education, chose the winners. s At the beginning of the year, NCSA's last vector processing machines were removed from service. The CRAY-2 (top) and the CRAY Y-MP systems (center) were taken apart and removed from the Machine Room in the Advanced Computing Building (bottom). (Photos by Tony Baylis and Lynn Gephart, NCSA Media Technology Resources) THE SECOND INTERNATIONAL WWW CONFERENCE '94 MOSAIC AND THE WEB "[THE WEB] IS MORE THAN JUST A TECHNOLOGY AND INTELLECTUAL NETWORK; IT IS A WAY TO BRING PEOPLE TOGETHER FROM AROUND THE WORLD," SAID TONY RUTKOWSKI OF THE INTERNET SOCIETY AS MORE THAN 1,300 WORLDWIDE WEB/NCSA MOSAIC USERS CONVENED IN CHICAGO LAST OCTOBER TO PROVE HIS POINT. ALMOST AS MANY WANNA-BE ATTENDEES MADE UP THE CONFERENCE WAITING LIST AS THOSE WHO GOT IN. ENTHUSIASM AND OPTIMISM FILLED THE AIR AT THE SECOND INTERNATIONAL WORLD WIDE WEB CONFERENCE WHERE DISCUSSANTS LINGERED ON AFTER PRESENTATIONS AND WHERE BIRDS OF A FEATHER SESSIONS LASTED WELL INTO THE EVENINGS. BY CONFERENCE END, PLANS WERE UNDERWAY FOR MEETINGS INTO 1996. ATTENDEES PROJECTED AN ATTEMPTED REGISTRATION OF 3,000 TO 4,000 PARTICIPANTS FOR THE THIRD MEETING SCHEDULED FOR APRIL 10-14, 1995 IN DARMSTADT, GERMANY. (ACTUAL REGISTRATION IS LIMITED.) FUTURE LOCALES ARE SET FOR BOSTON, MA (FALL 1995) AND PARIS, FRANCE (SPRING 1996). REGIONAL CONFERENCES--IN AUSTRALIA, FOR EXAMPLE--ARE ALSO PLANNED. EXTRAPOLATING FROM THE CURRENT RATE OF GROWTH ON THE NET, RUTKOWSKI PREDICTED THERE WOULD BE 100 MILLION GLOBAL USERS BY THE YEAR 2000. IF SO, MANY MORE WEB CONFERENCES AND OTHER SPIN-OFFS ARE GUARANTEED! FOR INFORMATION ABOUT THE THIRD WWW CONFERENCE, SEE THE URL BELOW. S by Fran Bond, editor of access Third International WWW Conference information: http://www.igd.fhg.de/www95.html This page, top to bottom: Poster session (Photo by Fran Bond); Grand reception (Photo by Mitch Kutzko); Conference session (Kutzko); NCSA Director Larry Smarr, keynote speaker (Photo by Tony Baylis); Mary Laplante, executive director of SGML Open, chairs Product Announcements Session (Bond) This page, top to bottom: Vendor Exhibit (Photo by Linda Jackson); Conference T-shirt (Jackson); Internet Access Room (Jackson); Registration (Kutzko); Ira Goldstein, vice president of the Open Software Foundation Research Institute and conference co-chair (Kutzko) "The ability to tame the oceans of information and make them navigable to the layman is the real benefit of Mosaic. . . . Few tools today do it better." --Unix Review U.S. Vice President Al Gore, Jr. demonstrating NCSA Mosaic to Jacques Santer, president of the European Commission, at the G7 "The Information Society" exhibition. Standing left to right are White House aide David Lytel, NCSA Associate Director Charlie Catlett, Vice President Gore, President Santer, and NIIT Executive Director Troy Eid. (Photo courtesy of G7 Conference) Still from "A Short Cut Home" created by Tony Rubey, former NCSA visiting artist. Rubey's video was featured in the "Globonet" episode of Future Quest aired on PBS. It also appears in "NCSA Mosaic/Glimpse of the Future," an NCSA-produced video. (Courtesy of NCSA Media Technology Resources) NCSA AT G7 CONFERENCE IN BELGIUM "This was NCSA's debut in Europe," says Charlie Catlett, NCSA associate director of Computing and Communications, NCSA's exhibit lead at the G7 Exhibition Event in Brussels, Belgium, February 25-26. The Event, designed to show the benefits and capabilities of the Information Revolution in today's global society, was held concurrent with the G7 Ministerial Conference on the Information Society. The European Commission (EC), chaired by President Jacques Santer, was host. The conference was attended by ministerial representatives from the G7 and European Union countries who are responsibile for telecommunications and information technology. Among the topics for discussion was the development of the Global Information Infrastructure. The Event complemented the conference's content by showcasing technology's advantages to society. NCSA was one of 15 U.S. organizations invited to exhibit information technology. U.S. Secretary of Commerce Ronald Brown led the U.S. delegation, and Vice President Al Gore gave the keynote address. NCSA's role at The Event demonstrated NCSA Mosaic as a window into the Global Information Infrastructure. Several NCSA videos were shown as well. Representing NCSA along with Catlett were Lex Lane, NCSA associate director for User Services; Roy Campbell, NCSA research scientist and professor in UIUC Department of Computer Science; Tom Fischer, NCSA computing utilities analyst in Computing and Communications; and John Ziebarth, NCSA associate director for Education and Outreach. NCSA's exhibit was supported by Silicon Graphics, Inc., an NCSA strategic technology partner, and AT&T, an NCSA industrial partner. s VROOM ONLINE VROOM, a virtual reality room using CAVE (Cave Automatic Virtual Environment) facilities, is now on the World Wide Web available for viewing in NCSA Mosaic. All the VROOM exhibits shown at SIGGRAPH 94 [see access, Summer 1994, pp. 3-14, 26] are now online at the URL below. VROOM: http://www.ncsa.uiuc.edu/EVL/docs/VROOM/HTML/OTHER/ HomePage.html NCSA ANIMATIONS ON PBS BY LYNN GEPHART, USER SERVICES Several NCSA animations were featured in episodes of Future Quest, a Public Broadcasting System (PBS) science educational series that premiered in the fall of 1994 and continued through March 1995. Future Quest topics included time travel, robotics, virtual reality, genetic engineering, and cybernetics. Each episode integrated clips from classic science fiction films and television shows to compare fictional visions of the future with scientific realities. Shows featuring NCSA footage were "Globonet" and "Doomsday." s BIOLOGICAL IMAGING WORKSHOP The Second Workshop on Advanced Computing for Biological Imaging will be held April 28-29, 1995, at the Beckman Institute. The purpose of the workshop is to provide a forum for discussion of integrating imaging modalities and applications used in biological research with advanced computing systems. The goal of the workshop is to bring together researchers and developers for an exchange of applications, techniques, and technologies. Hosts are the Beckman Institute and NCSA. Bridget Carragher (Beckman Institute) and Clinton Potter (NCSA) are organizing the workshop. Workshop attendance is limited to 50 participants to foster a proper environment for discussion. Dead-line for registration was March 1, 1995. For further information, contact Judy Jones, ACBI coordinator, by phone (217) 244-5582 or by electronic mail acbi@ncsa.uiuc.edu (Internet). http://kepler.ncsa.uiuc.edu:6666 SGI TRAINING DATES Monthly training dates have been announced for NCSA's new SGI POWER CHALLENGE system. Session dates, scheduled now through June, are available at the URL below. For further information, academic users contact Deanna Spivey at (217) 244-1996 or ncsa-train@ncsa.uiuc.edu (Internet). Industrial partner representatives, contact your representative. NOTE: All dates are subject to change. Registration is required at least 10 days before a workshop. A cost-recovery fee may apply to some sessions. s http://www.ncsa.uiuc.edu/General/Training/cal_complete.html GUIDES TO NET NAVIGATION BY GINNY HUDAK-DAVID Go into just about any bookstore these days and you are sure to find a section of books dedicated to the Internet, the World Wide Web (WWW), online subscription services, and NCSA Mosaic. The choices can be overwhelming. To help you make your selection, two new Macintosh-based books are reviewed here. A QUICK TOUR Ventana Press' Mosaic Quick Tour for Mac (subtitled Accessing and Navigating the Internet's World Wide Web) is a quick-to-read book with a breezy and fun approach. Here is author Gareth Branwyn's transition from the nitty-gritty chapters to the rest of the book: "Now that we have the basic technical aspects out of the way and we're ready to roll, let's go for a stroll through the Web to see what kind of trouble we can get ourselves into. Cyberspace beckons." The stroll starts off with an explanation of the Web, hypermedia, and Mosaic's menus--in short, all the stuff you need to know to understand the value of a Web browser and how to use NCSA Mosaic. While this introduction to the Web takes 70 pages, it is not overwhelming because of Branwyn's style. Another chapter runs through a sample Web session (dubbed "web walking"), along the way explaining the various Mosaic icons and file types you encounter. This jam-packed chapter also introduces the HyperText Markup Language (HTML) and takes you through setting up a home page. The fifth and final chapter ("site seeing") shows you some Web displays that the author likes. Visually this book has an open look, making it easy to read. Scattered throughout are hot tip boxes, screen displays, and icons that hold your attention from page to page. A HANDBOOK PLUS SOFTWARE Among the latest offerings from pioneering O'Reilly & Associates Inc. is The Mosaic Handbook for the Macintosh by Dale Dougherty and Richard Koman. Exhibited at the Second WWW Conference in October 1994, this volume comes bundled with a disk containing Enhanced NCSA Mosaic software. (O'Reilly is a licensee of Spyglass Inc.) In the handbook, the authors explain the WWW and introduce Mosaic, explore the Web, explain HTML, walk through creating a home page, and (in an appendix) list the Mosaic menus and menu options. The foreword says the book is "more than a description of the Mosaic interface; it's a guide to navigating the Internet." The O'Reilly volume spends considerable time on searching the Web, using other Internet services, and customizing your copy of Mosaic. The Mosaic Handbook is a lot like the other offerings from O'Reilly--dense pages but with a nice layout that helps you quickly find what you are looking for. Liberal use of illustrations and screen displays help make the book less intimidating than it could have been. SIMILARITIES, DIFFERENCES Both books cover similar territory, although they are organized differently. Dougherty and Koman, on average, provide more details on each topic than does Branwyn, and they also give a nice history of the development of Mosaic. Their style is straightforward, and their instructions are clear. Branwyn's volume is lighter in tone, which some might find more inviting. Both books reference their publisher's WWW services--the Ventana Visitors Center from Ventana Press and the Global Network Navigator (GNN) from O'Reilly. These references are not intrusive or limiting because both books include countless URLs for other Web sites. So which, if any, book should you buy? It depends. If you want a quick tutorial on the WWW and NCSA Mosaic, then spend $12 on the Mosaic Quick Tour. If you want more detailed information as well as software, then hand over $24.99 and get the O'Reilly book. If you are on a limited budget, take an online look at NCSA Mosaic for Macintosh User's Guide and the NCSA What's New page. They aren't as thorough as the books reviewed here, but you can't beat the price! s Ginny David is a member of the Publications Group and prepares WWW files. Ventana Visitors Center: http://www.vmedia.com/vvc Global Network Navigator: http://gnn.com/GNNhome.html NCSA Mosaic for Macintosh User Guide: http://www.ncsa.uiuc.edu/SDG/Software/MacMosaic/Docs/ MacMosa.0.html NCSA What's New: http://www.ncsa.uiuc.edu/SDG/Software/Mosaic/Docs/whats- new.html abbreviations ARPA Advanced Research Projects Agency CTC Cornell Theory Center HPCC High Performance Computing and Communications NASA National Aeronautics and Space Administration NCAR National Center for Atmospheric Research NCSA National Center for Supercomputing Applications NII National Information Infrastructure NSF National Science Foundation PSC Pittsburgh Supercomputing Center SDG Software Development Group SDSC San Diego Supercomputer Center SGI Silicon Graphics Inc. TMC Thinking Machines Corp. UIC University of Illinois at Chicago UIUC University of Illinois at Urbana-Champaign URL Uniform Resource Locator VEG Virtual Environments Group WWW World Wide Web documentation orders Articles in this access may refer to items that are available through the NCSA Technical Resources Catalog. To receive a copy of the catalog, send your request to Orders for Publications, NCSA Software, and Multimedia [see ncsa contacts, page 2]. accessing NCSA's servers Many NCSA publications (e.g., calendar of events, user guides, access, technical reports) as well as software are available via the Internet on one of three NCSA servers: anonymous FTP, Gopher, or the World Wide Web. If you are connected to the Internet, we encourage you to take advantage of the easy-to-use servers to copy or view files. Anonymous FTP IP address: ftp.ncsa.uiuc.edu (141.142.20.50) Gopher server IP address: gopher.ncsa.uiuc.edu World Wide Web IP address: www.ncsa.uiuc.edu NCSA WWW Home Page: http://www.ncsa.uiuc.edu/General/NCSAHome.html NOTE: References in access to a World Wide Web URL (Uniform Resource Locator) refer to the server address and file location information used by NCSA Mosaic software to retrieve documents. If you have any questions about accessing the servers, contact your local system administrator or network expert. Instructions for accessing the anonymous FTP server follow. downloading from anonymous FTP server A number of NCSA publications are installed on the NCSA anonymous FTP server. If you are connected to Internet, you can download NCSA publications by following the procedures below. If you have any questions regarding the connection or procedure, consult your local system administrator or network expert. 1. Log on to a host at your site that is connected to Internet and running software supporting the FTP command. 2. Invoke FTP by entering the Internet address of the server: ftp ftp.ncsa.uiuc.edu or ftp 141.142.20.50 3. Log on using anonymous for the name. 4. Enter your local login name and address (e.g., smith@ncsa.uiuc.edu) for the password. 5. Enter get README.FIRST to transfer the instructions file (ASCII) to your local host. 6. Enter quit to exit FTP and return to your local host. 7. The NCSA publications are located in the /ncsapubs directory. All brand and product names are trademarks or registered trademarks of their respective holders. access National Center for Supercomputing Applications 152 Computing Applications Building 605 East Springfield Avenue Champaign, IL 61820-5518 u Spring 1995 vol. 9 no. 1""--------"---"""'''''''''''''''''''''"q"t" @ (,p--;qIo(c)ITXd )-?d}- 6\`"f/u 9 I M w # > B l p 6 O f j n )FJ! --`Jp " 2 D \ ` w (c) "&CUz&8,>Wy}o'AEb>Yk37HUg}! -co"Sfjnr J"Z,OS+/t9im3quCp/r'M''"((\(`(()/)t))*6*}**o*+/+s++,,H,,----Y--(c)--.2.k../3/z/00,000o00121r112:2|233F3g3k33454Z4^4y455F55566Y667*7=7A777788_88889=! ! -a9=999999:):M:Q:h:::::--:;;E;I;M;|;;;;;;;;< < +>/>s>>?:?>?\??@(@k@@A=AQAUAAB#B`BdB (TM)BCCC`CCD(DFDJDDEEcEEEEF7FzFFFG+GoGGGH4HyHII-I"IbI! ! -aIIJ&JfJ (R)JK,KZK^KKL-LpLLM-M1MmMMN6NaNeNNOOGOOPP!P%PkP (R)PPPQEQQ-RR/R3RwRSSSSaSoST0TtTxTTU8U~UUUV*VpVV'VVW=WWWXEXXY Y^D^H^^^^^_"_&_*_._I_[____`&`W`[``a%abafaab2bubbccHccd dOdd"eeCeGeeffVffg$gjg(c)gghhFhh i i5i9i{ijjJjj-kkZkklllLllm mgmommm"nnLnn-oo[oop"p&pippq3qrq! ! -aqqr:rrrrsDsst tTttu ucuouuuv;vvww&w*wfww-w'xxZxxyydyyyz0zzzzzz{%{i{{{|4|v|||} }I}}~~~S~`~d~~!%e.JN"?C] V(h! -a--'[%`qu#`$(f-s=5v>CVA`d (TM)*9= $ha (R)+Y]S!-1r:VZy! -! aGqu@Dos-X!d-"-f----.--V--Z----"$"j! ! -a"j"""""4"r""'=''' 'I'''V&+/mq:>HCGKO(l$b'm="h8lp@ V'm,s6:|0y! -c-c)nLY!%PI-Y-T"V-`o-"c$iuy"ALHL~D[_cgy}--'aI Lz! ! -az~ GXim HK04t8u:> IY]Znr-a (R)$>~.r@z> I c (TM) ? S f j S! ! -aS # h SK"\!X\*q3v(c)8}F]ax#d (TM)$(k (TM)!%eR=|,0u@dh (TM)6y-4-v! ! -a-v--- (R)--';? = !!!!!0!4!J!N!!"""X""#+#p####$C$$$%%S%%%&&N&&''''W'' '((Q((( ))N))))**U***--+ +O++++,;,s,,-3-t---.+.i.../-/[//! ! -a/0/0s001>1B1122G2K22223?3y344>4444455555"5&5e55--5'66K66"777U7788[888"99V99::_:o:;;;-;";/;3;L;;;;;<:<}<< (R)<=-=q==>:>K>O>k>>?.?d?h?o?@&@n@@@A"AbA! ! -aAAB*BqB(c)BBC0CrC}CCDDP04! ! -a4z> &*m9NRRd0^bfj59 Ly} ML\"f.^b'h(c).<@Ds$W!S8PT! ! -a'a,c!%X|J L')no+p159~Hosw buyBo)omoo*.kJ (TM) (R)- (R)q (R) (R)(c)(c)(c)Z(c)(c)! ! -a(c)(c)(c)(c)(c) (TM) (TM)P (TM)p (TM)t (TM) (TM) (TM) (TM)6GKO-bf-r "3x|E R_coH(DH)bf NRVr"Z-'*p@"! ! -a_Y#ko-\`|F-_:F--D"_ So(c)3xC04mq"9EIMQbfJ-A^}! ! -a15MQ ; [ _ { -h#6:lp!\+ew{ (R) X-/-d-p- (TM)--- --A--k---- "" "D"R"V"""""""*"I"e"i"m"q! ! -K" "p@"q #/8?ISw]gq| @q""} $-'1:2CM2U]Wgvnx (TM)(Ep ")"pP.)- D = : EG>6E=GC?IC,E4-C@ !D"D#$?%A&'(F)>*+,C-./E0G192C3)4156"p! "J$9=I]q (TM)`"jzS-v/ATevS4(c)"q !" (R) T " P"""""""3 (TM)3333338888889:+:0:~::HHLXG(HHLd'h=/--RBH-:LaserWriter 8 YChicagoNew YorkGenevaMonacoVeniceLondonAthens San Francisco Cairo Los Angeles Zapf DingbatsBookmanN Helvetica NarrowPalatino Zapf ChanceryTimes HelveticaCourierSymbolMobileOptima! Avant Garde"New Century SchlbkI Avant Garde BookObliqueB Avant Garde DemiBI Avant Garde DemiObliqueI Bookman LightItalicB Bookman DemiBI Bookman DemiItalicNI Helvetica Narrow ObliqueNB Helvetica Narrow BoldNBI Helvetica Narrow BoldOblI New Century Schlbk ItalicB New Century Schlbk BoldBI New Century Schlbk BoldItGaramondI Garamond LightItalicB Garamond BoldBI Garamond BoldItalicI Optima Oblique B Optima Bold (R)BI Optima BoldOblique StencilCutAstroAndesite TrueType Eurostile SapirSans WindsorDemiGaramond NarrowT SapirSansItalNCSASpecChar "MT Extra P ArchitectMyriad Roman Myriad BoldGillSans BoldItalic GillSans BoldGillSans ItalicGillSansGillSans LightItalicGillSans Light GaramItcTGaramItcTCon"Industria Solid"MathematicalPi 6"MathematicalPi 5"MathematicalPi 4"MathematicalPi 3"MathematicalPi 2"MathematicalPi 13 (TM)Copperplate33bc3Copperplate32bc3Copperplate31bc3Copperplate29bc3Copperplate32ab3Copperplate31ab3Copperplate29ab8AGaramond Titling8 AGaramond8AGaramondExp8AGaramondExp Bold8AGaramond Bold8AGaramondAlt9Helvetica Compressed:+Carta:0Brush Script:~LB Helvetica Black:L Helvetica Light:GillSans ExtraBold&&&E[f4 &"o"p"t'"p .