The Office of Advanced Scientific Computing Research (ASCR) supports research in computational technology and laboratory technology research, subprograms that underlie a variety of Department of Energy missions.

ASCR's primary mission, carried out by the Mathematical, Information, and Computational Sciences subprogram, is to discover, develop, and deploy the computational and networking tools that enable researchers in the scientific disciplines to analyze, model, simulate, and predict complex phenomena important to the Department of Energy. To accomplish this mission the program fosters and supports fundamental research in advanced scientific computing - applied mathematics, computer science, and networking - and operates supercomputer, networking, and related facilities. The applied mathematics research efforts provide the fundamental mathematical methods to model complex physical and biological systems. The computer science research efforts enable scientists to efficiently run these models on the highest performance computers available and to store, manage, analyze, and visualize the massive amounts of data that result. The networking research provides the techniques to link the data producers; e.g., supercomputers and large experimental facilities with scientists who need access to the data. The two topics that follow support this scientific computing mission.

The Laboratory Technology Research subprogram funds high-risk, multidisciplinary research partnerships between the DOE's Office of Science multi-program national laboratories and private industry. Projects supported explore applications of basic research advances in the investigation of problems, over a full range of scientific disciplines, whose solutions have promising commercial potential.

The Department of Energy (DOE) supports a wide range of research activities in mathematics, information, and computational sciences to support distributed high-end computing, remote instrumentation and data storage, and scientific collaboration in DOE. Many emerging energy research problems require coordinated access to distributed resources - people, data, computers, and facilities. These problems generate an enormous amount of data that must be remotely analyzed and visualized by scientists in geographically distributed scientific facilities. The distributed and dynamic character of this large-scale resource sharing calls for secure, high performance network infrastructures and services to support collaboration on a national and international scale. The current Internet, built with commodity components and optimized for low-speed, best-effort applications, lacks the capability and functionality to deliver the level of performance and security required by DOE scientific applications and collaborations. Therefore, research is needed to enhance the performance and capabilities of network services, transport protocols, network control and management, and other features that contribute to the overall security, survivability, and scalability of contemporary networks. Grant applications must clearly state how the proposed research will be beneficial to the long-term mission of DOE and at the same time make significant contributions to the general subject areas. Grant applications are sought only in the following subtopics:

a. Agile Optical Network Technologies - New technology is needed to harness and extend the unprecedented bandwidth capability offered by Dense Wave Division Multiplexing (DWDM) to the DOE high-performance computing and high-end scientific applications. Grant applications are sought to develop agile optical network technologies leading to the development and deployment of highly reconfigurable optical network components, optical routing and switching technologies, scalable link layer framing protocols, and optical interconnects for supercomputers and cluster computing. Areas of interest include: (1) scalable non-SONET (Synchronous Optical Networks) framing mechanism for Internet Protocol (IP) over DWDM at OC-192 (Optical Carrier level 192 ( concatenated at 2.5 Gbps) rates and above; (2) optical interconnects for the interconnection of supercomputers, cluster computing hardware, and storage devices; (3) Multi Protocol Label Switching (MPLS) over dynamically reconfigurable lambda-switched networks; and (4) performance characterization of legacy protocols (Point-to-Point Protocol (PPP), High Data Link Control (HDLC), and Ethernet frames) operating over terabits/sec optical links. Grant applications must address optical network issues of specific interest to the DOE computing environment.

b. Resource Discovery and Sharing - Grant applications are sought for research and software tool development in technologies that support coordinated resource sharing in areas such as resource discovery, resource access, authentication, authorization, etc. Tools should be developed using an integrated services approach and may be, for example, robust full-featured servers with one to many services that incorporate best practice concepts emerging from the grid community.

c. Network Security - Grant applications are sought to develop tools and techniques for: (1) secure and fair means for enabling application and user access and control of network resources; (2) smart network management (i.e., highly capable network management agents, tools, and stations) that adapt to a dynamic network infrastructure; and (3) innovative techniques for intrusion detection using Artificial Intelligence (neural networks, fuzzy logic, etc.) and advanced statistical techniques. The tools and techniques must be applicable throughout the high speed network. Grant applications must address appropriate public key infrastructure (PKI) research that supports these efforts and that is interoperable and consistent with industry-driven and government PKI.

d. Network Engineering - Grant applications are sought to develop advanced tools, technology, and services to: (1) monitor, analyze, and manage multiple layers of heterogeneous networks; (2) perform effective and scalable routing/switching including best effort and priority traffic, reliable multicast, real time, and variable or flat accounting/costing mechanisms and protocols for differentiated services; (3) manage lead user infrastructure, which entails the dynamic and secure concurrent support of production and network research traffic (i.e., multiple policy traffic that may be in conflict with one another) on the same infrastructure, as well as the de-aggregation of tributaries; and (4) develop remote (wireless), reliable and secure network connectivity at higher speeds than currently commercially available for real time operation (to include two way videoconferencing, visualization, and collaboration) and capable of supporting a variety of protocols.

7. Next Generation Internet (NGI)
http://ngi.gov

8. QBONE (Cooperative Advanced Quality of Service Testbed) http://qbone.internet2.edu/

9. Rajagopalan, B., et al., IP Over Optical Networks: A Framework, (Internet Engineering Task Force Internet Draft). (URL: http://www.ietf.org/internet-drafts/draft-ietf-ipo-framework-00.txt)

The Department of Energy (DOE) sponsors a wide range of research activities in mathematics, information, and computational sciences to support high-end computing that is required to solve leading edge scientific problems critical to DOE's missions. The solution of these problems requires hundreds of gigaflops to multiple teraflops of computing power. Currently available computer architectures are based on clusters of mid range symmetric multiprocessor nodes, built with commodity components and optimized for mid-range commercial or scientific applications. In order to achieve acceptable performance on large clusters where 1000s of processors must effectively work on a single application, significant specialized software is needed. Therefore, to enhance the performance and capabilities of these clusters, improvements are needed in operating systems, system management software, performance monitoring and improvement tools, data management software and visualization software.

Grant applications must clearly state how the proposed research will benefit DOE's long-term mission and at the same time make significant contributions to the general subject areas. In addition, grant applications must demonstrate that the proposed research is scalable to at least 1000s of processors. Priority will be given to technical approaches that build upon open source software and which support interoperability across all operating systems important to the high-end community, including both proprietary and open source systems. Grant applications are sought only in the following subtopics:

a. System Management Software - Cluster management encompasses a wide range of tasks such as configuring new nodes, monitoring the state of the system and rebooting when necessary, and upgrading and installing new soft-ware. System vendors have supplied management tools for their products, but clusters built by end users are typically administered with ad hoc tools that are particular for a given site. Research in large clusters has generally not focused on systems management approaches or techniques. Grant applications are sought for innovative ways to manage large clusters of commercial computers. Areas of interest include power and fault management, software configuration control across nodes, and hardware and software status monitoring.

In addition, current operating systems are not optimized for clusters with 100s to 1,000s of processors and may incorporate design elements that raise significant barriers to achieving high end user application performance. Therefore, grant applications also are sought to develop improved operating systems that are based on portable methodology and that enhance the delivery of high performance to end user scientific applications in a large cluster environment.

b. Performance Monitoring and Management Tools - An important barrier to the efficient use of large cluster systems is the inability, in present systems, to sustain more than a small fraction of peak performance on priority applications. In order to provide a higher percentage of peak performance to end users, research and development is needed to develop software and programming techniques for the more effective utilization of large cluster systems. Grant applications are sought to develop tools that will increase the ability of scientific users, writing software in Fortran, C, or C++, to monitor and improve the performance of their software. Examples of important issues include memory utilization, cache performance, identification of communications bottlenecks, and optimization of data layout.

c. Data Management Tools - Existing data management tools characteristically focus on the needs of business systems and are not fully responsive to the data management needs of the scientific community. Grant applications are sought to develop database management systems approaches for high throughput parallel I/O and complex queries of large scientific databases; for agent methodology for feature extraction and complex query operations; and for tools for user-driven and automatic clustering, reclustering, or replication of objects to maximize retrieval efficiency.

d. Visualization Tools - A new generation of scientific visualization tools are needed to effectively display voluminous and complex scientific datasets. Grant applications are sought for visualization tools that support vector/tensor field visualization in 3-D; for remote and collaborative visualization methods; for the characterization of simulation, experimental, and visualization errors/uncertainties; and for adaptive, multiresolution, parallel and scalable visualization algorithms.

4. U.S. Department of Energy, Office of Science
http://www.science.doe.gov

Return to Solicitation Table of Contents