8. Basic Research and Human Resources

Basic Research

HPCC basic research has two goals -- to develop new methods for addressing fundamental technological limitations encountered as the field advances, and to develop the foundations for next and future generations of HPCC technologies.

The underlying basic research that develops the foundations for future generations of HPCC technologies can be characterized in seven broad areas: components, communications, computing systems, software and tools, intelligent systems, information management, and applications. Examples of accomplishments resulting from NSF basic research illustrate its impact. These include the following:

• The Ray Casting Engine project used VLSI (very large scale integration) design techniques and MOSIS (metal oxide semiconductor implementation service), an ARPA- and NSF-funded service for academic research and education, to design and build a new, application-specific, massively parallel computer to compute solid geometry quickly for mechanical design. This project resulted in new approaches to solid modeling, metrology, and the treatment of sculptured objects and led to examination of completeness and utility of ray representations.

• The Supercomputer ToolKit project used inexpensive, fast commodity hardware modules with static reconfigurable interconnect and dedicated software on each module to build a cost- effective application-specific computing machine that achieved supercomputer performance in simulating a model of the complete solar system. This dedicated system was used to carry out a 100 million year integration and proved planetary orbits to be chaotic. It was also used in engineering design to carry out real-time computations with sensors and actuators to design dynamically stable bridges.

• A project on the Correctness of Hardware and Software Systems used theoretical models to study complex systems with a large number of states, to study how to scale techniques for state-space blowup, and to verify that models of such systems satisfy specifications. The resulting techniques were used to identify coherency and potential deadlock errors found in the IEEE Futurebus+ protocol for application systems developed under the Navy's "Next-Generation Computer Resource" program.

• A Reduced Complexity Robotics project used modular hardware, simple sensors and actuators, efficient software design, rigorous scientific models and principles, theories on completeness, correctness, and complexity of plans for flexible parts handling and assembly systems that can be simulated, built, and programmed in a matter of weeks versus months and years. Such systems are directly and seamlessly controllable to produce any quantity of a selected assembly within a prescribed family of products, programmed directly from piece-part CAD files and manufacturing process sheets.

• A Multi-Media, Knowledge-Based, Evolutionary Biomedical Database System studies the retrieval of images by content rather than image ID; the integration of images, text, and domain knowledge; approximate or fuzzy query answering; and application and demonstration of these techniques on biomedical images and data, in particular hand growth and brain tumor studies. The work has led to the development of new spatio-temporal models for image representation, and the prototype system has potential for use in radiology and other areas using image databases such as manufacturing and digital libraries.

• A Robust Spoken Language Systems project that ARPA also supports focuses on the development of next-generation systems more capable of handling continuous speech, speakers' variabilities, disturbances from environmental noise, and larger vocabularies. The resulting system architecture has been licensed to U.S. West for enhancing telephone voice transactions. The U.S. 1995 Census will field test the system for interactive telephone interviews.

• The Mentat project is developing an object-oriented programming system that transparently distributes computations across geographically distributed, heterogeneous, computing resources. Applications in biochemistry for comparing sequence libraries and in electrical engineering for testing combinatorial circuits are executing in an environment of Sun, IBM, and SGI systems. Mentat software has been distributed to more than 150 sites.

• A Computational Molecular Biology project is studying the problem of how to locate genes and important functional signals in "raw sequences." Efficient combinatorial matching algorithms and algorithms for sequence reconstruction along with an innovative special biological chip as a DNA sample holder have been developed. This will allow the reconstruction of unknown DNA fragments from a set of common DNA sequences. The resulting binary sequencing by hybridization and customized DNA chips are expected to impact medical genetic diagnosis, DNA sequencing, DNA fingerprints for legal systems, and recombinant DNA technology.

Other applications that benefit directly from high performance computing and communications research include physics (for example, astrophysics, high energy physics, low temperature physics, and material sciences), chemistry (from basic chemical properties to chemical processing), biology (including genome understanding, molecular dynamics, neural and cellular models, and pharmacology), and a full spectrum of engineering disciplines (such as fluid dynamics, petroleum engineering, and structural dynamics).

NSF supports long-term investigator-initiated basic research in science and engineering disciplines represented by its Directorates for Biological Sciences, Engineering, Geosciences, Mathematical and Physical Sciences, and Social, Behavioral, and Economic Sciences. This research is generally focused on improving the ability of researchers to use high performance computing and communications to advance fundamental understanding in those disciplines. It frequently takes the form of advanced software development that ultimately has impact outside of the particular discipline. These Directorates also participate in Grand Challenge and National Challenge activities. Some specific activities are as follows:

• Through group awards and individual investigator awards, graduate and postdoctoral students are supported in order to increase the pool of talent trained in scientific computing and information processing. In FY 1995 NSF received 317 preproposals in response to a solicitation for multidisciplinary group-oriented research in Grand Challenges, National Challenges, enabling technologies, and challenges in computer science, mathematical sciences, and problem-solving environments. Between 15 and 20 multi-year awards are anticipated in FY 1995.

• Through its institutional infrastructure and research instrumentation activities, in FY 1995 NSF provided resources for conducting research in scalable I/O, network applications in education and manufacturing, gigabit-speed networking research, virtual manufacturing, multimedia systems, basic experimental research, and for the NIE program described immediately below. In FY 1996 NSF will evaluate existing programs, develop infrastructure support for basic research needed for educational use of the NII, continue infrastructure support for virtual manufacturing and other NII activities, and increase NIE participation.

• The Computer and Information Science and Engineering (CISE) and the Education and Human Resources (EHR) Directorates at NSF have established the Networking Infrastructure for Education (NIE) program. NIE's goals are to build synergy among technology and education researchers, developers, and implementers; encourage innovation and experimentation by educational groups; and integrate networking technology with education reform. Co-funded with ARPA, NSF awarded 19 planning grants and policy studies, 11 multi- year projects, and 10 one-year supplements to existing awards. In FY 1995 between 100 and 125 proposals are expected. One focus of the new awards will be on bringing networking information to previously underrepresented communities such as tribal colleges.

• NSF funds basic research in gigabit speed networking including work at the gigabit testbeds described in Section II.1; gigabit switching systems; protocols and software structures for network management; resource discovery among collaborative information spaces in large decentralized environments; network information theory; multi-sender and multi-receiver network security; modulation, detection, and coding of reliable information storage and retrieval; all-optical networks; and optical technologies for computing and communications. In FY 1996 NSF plans to fund research aimed at more effective use of large information resources, solicit and make awards for wireless access testbeds, continue collaboration in a 10 Gb/s testbed with ARPA, and examine the convergence of computing, entertainment, and telecommunications from the standpoint of Internet experience.

NASA funds basic research in computer architectures, fundamental algorithms, computational complexity, networked and distributed computation, numerical analysis, and application-specific algorithms. NASA supports the Illinois Computer Laboratory for Aerospace Systems and Software (ICLASS) at the University of Illinois and supports research centers at its Ames, Langley, and Goddard facilities.

DOE supports basic research in applied mathematics, computer science, and computational science toward solving large scientific problems. Work is conducted at ten DOE laboratories and more than 30 universities; more than 60 undergraduate students and more than 40 graduate students are supported. FY 1995 accomplishments include:

• Software based on adaptive mesh refinement techniques for resolving three-dimensional fluid flow phenomena in complex geometries

• Security procedures for voice and data transmission based on sending the data encapsulated in a chaotic signal and decoding the signal using techniques for controlling chaos

• Software for resolving large-scale stochastic programming problems such as those found in large commercial electric power distribution systems, networks, and airline crew equipment scheduling operations

• Development and implementation of sparse matrix techniques for finite element software that performs crash simulation and analysis

FY 1996 DOE plans include:

• Research in inverse scattering, nondestructive evaluation, and discrete mathematics applied to parallel programming and genome sequencing and mapping

• Multidisciplinary, multi-sector research in DOE's Grand Challenge applications

• Initiating a graduate student program based at industrial and laboratory sites

NLM grants provide training in HPCC technologies for biomedical professionals and provide training for computing professionals in medical and health applications.

Education and Training

During the three years that the HPCC Program has formally existed, the community of expert users has grown from a small enclave of researchers in computing and communications technologies to large groups of researchers and students in all fields. The community of able users spans all sectors of the economy and all sectors of society throughout the U.S. and around the world. This growth in knowledge and ability was initially the result of hands-on and formal training at HPCC-funded facilities (described in Section II.5) and in industry; today this type of training and education is being conducted on line; at universities, colleges, elementary and secondary schools; at other places of learning such as museums and libraries; and in industrial training programs throughout the country.

In FY 1994 NSF established a Native American Telecommunications Technical Assistance think tank, joined with ED and the National Telecommunications and Information Administration (NTIA, part of the Department of Commerce) in sponsoring a workshop on the role of state networks in developing statewide networking infrastructure, and developed museum-based networking support activities in urban centers.

On-going NSF activities include the following:

• Joint NSF/ARPA funding of more than 4,000 student projects in VLSI fabrication by MOSIS

• The Superquest program that provides high school students and teachers with research experiences at its Supercomputer centers, undergraduate course and curriculum development, and research experiences for undergraduate students

• Pilot Educational Networks project to develop, implement, test, and evaluate applications of computing and communications to education. The NIE (described earlier) addresses issues of large scale networking for education. Issues include scalability, cost/benefits, policy, effective applications, and educational impact for user-driven models of computer networking in education.

• Pilots and large-scale models for educational applications of digital libraries

• The Maryland "Virtual High School for Computational Science" on the Internet, where students throughout Maryland learn science and mathematics by participating in collaborative application projects using computational science, guided by experienced teachers and other mentors. Three groups participate:

  -- Montgomery Blair High School, which has developed a collaborative teaching approach based upon computational science applications

  -- Maryland high school teachers and students, including those at isolated rural, inner-city, and suburban schools

  -- Government agencies, business groups, and network groups concerned about the schools' abilities to respond to the promise of advanced technologies

The NSF postdoctoral training program in computational science and engineering and experimental computer science continues. To date some 80 people received two years of support to pursue computational science across a spectrum of disciplines and in experimental computer science.

NASA has a broad-based program designed to produce rapid improvement in both K-12 and lifelong learning in science, mathematics, and engineering. The two components of this program are (1) open competitive solicitations called "cooperative agreement notices," and (2) an educational outreach program at seven NASA field centers nationwide.

In FY 1995 NASA solicited for projects to use the Internet in lifelong learning through the cooperative agreement notice "Education, Training, and Lifelong Learning in Aeronautics." Details can be found at:

ftp://quest.arc.nasa.gov/pub/can95/

Outreach program accomplishments at the field centers include:

• Ames Research center (Moffett Field, CA) -- Produced several popular videos on approaches to and benefits from using the Internet in the classroom

  http://quest.arc.nasa.gov/

  -- Established a popular educational resource Web page supporting teachers nationwide

  http://quest.arc.nasa.gov/trc/toc.html

• Dryden Flight Research center (Edwards, CA) -- Investigating how to use the Internet to assist challenged and emotionally disturbed students in the classroom

• Goddard Space Flight center (Greenbelt, MD) -- Established an Internet-based program for Goddard scientists to work with teachers in all 24 Maryland school districts to improve the teaching of Earth and environmental science

  http://viva.gsfc.nasa.gov/

• Jet Propulsion Laboratory (Pasadena, CA) -- Produced multimedia software to guide school children through a space exploration mission from defining objectives through designing and selecting instruments and platforms to analyzing scientific data

  http://www.jpl.nasa.gov/educ/education.html

• Johnson Space center (Houston, TX) -- Developed a knowledge robot (Knobot (TM)) to discover and provide materials on the Internet tailored to the needs of K-12 teachers

  http://www.jsc.nasa.gov/stb/ILIAD/Mosaic/iliad.html

• Langley Research center (Hampton, VA) -- Deployed Internet networks to public schools economically by using the local carrier telephone infrastructure

  http://k12mac.larc.nasa.gov/Networks/Models.html

• Lewis Research center (Cleveland, OH) -- Demonstrated prototype high bandwidth, low-cost Internet access to schools using wireless radio frequency technology

  http://www.lerc.nasa.gov:80/Other_Groups/K-12/K-12_homepage.html

Gonzaga High School students using NASA's Earth System Science Community Curriculum Testbed. These students are using the Internet to access and review student-published "chapters" submitted by students from many schools, including the Model Secondary School for the Deaf at Gallaudet University.

On-going DOE activities include Adventures in Supercomputing, which trains in-service high school teachers and middle school teachers at select sites, and the National High School Honors Program, which provides summer supercomputing enrichment for gifted high school students.

In FY 1996 DOE plans to complete its electronic computational science textbook. In FY 1995 and FY 1996 DOE plans to develop a similar science textbook. The agency also plans to conduct a networking technology assessment project to identify educational technology tools and curricula for a variety of schools and to produce a catalog of effective technologies for use by teachers, administrators, and state and local education officials.

DOE supports more than 50 students in its Computational Science Graduate Fellowship program.

NLM funds both planning and implementation grants to academic medical centers with the goal of developing, testing, and implementing generalizable systems linking administrative, clinical, educational, and research databases so that they appear as one to the user. Four grants continue; approximately seven planning awards and two implementation awards will be made in FY 1995.

NCRR funds undergraduate and graduate students in the biomedical sciences and biomedical scientists in the use of HPCC technologies. NCRR trains K-12 teachers to use interactive learning tools in classroom activities and in scientific-computer literacy.

NSA is developing a skill-based assessment and training model in order to maintain and enhance employee abilities in today's constantly changing workplace.

EPA is developing a "virtual support center" through which advanced environmental modeling tools and assistance are provided to a target user community. That community consists of Federal government personnel, environmental decision makers from state governments, and industrial personnel representing the Consortium for Advanced Modeling of Regional Air Quality. The user community works with developers and trainers to ensure that system capabilities and interfaces meet their needs. The center will be Internet-accessible to environmental decision makers across the country. EPA also funds its EarthVision computational science educational program for high school students, graduate student support, training fellowships, conferences, and workshops. In FY 1996 EPA will develop environmental simulation software to expose high school students to computational environmental analysis.

The ten ED regional educational laboratories have created a Technology Task Force to develop Internet-based information systems. Connectivity of all ten labs will be completed in FY 1995 along with information servers at nine of the labs.

ED's Teacher Networking Project funds networking and applications for teacher professional development, with priority given for teachers serving poor families.