Randall Bramley - US grants

The goal of this research is to develop parallel system solvers for symmetric indefinite and non-symmetric systems such as those occurring in computational fluid dynamics, develop parallel techniques for solving time-dependent problems, and construct a rapid prototyping environment for developing and experimenting with algorithms in sparse matrix computations on shared and distributed memory multiprocessors. In particular, the are examined: (1) parallel solution schemes for solving Stokes-like linear arising in CFD and other augmented Lagrangian problems such interior point methods of linear programming, (2) incomplete Newton methods with Krylov or row projection solvers for the inner iteration, (3) solution of time-dependent problems using high-order methods in time, (4) defining and developing sparse matrix primitives analogous to the highly effective BLAS-3's for dense computations, and (5) implementing sparse factorization and solution schemes. The goal is portable efficiency and scalability, by parallelism on multiple levels. This in turn allows the work be used across a wide spectrum of existing and future architectures, instead of simply a particular machine.

This award is for a postdoctoral associate in Experimental Science. The associate, Xiaoge Wang, will be working on tools for parallelizing CFD codes.

9502292 Bramley The work proposed here involves research and teaching in scientific computation. The research primarily uses partial orthogonalization methods for solving applications that include incompressible fluid flows. First, a new class of incomplete orthogonalization preconditioners that preliminary results have shown to be more effective than standard preconditioners will be further developed and analyzed. Secondly, methods for solving constrained systems using orthogonal projectors will be further developed, and extended to nonsymmetric (full Navier-Stokes equations) problems, and other problems in fields such as electromagnetics. Both as a tool and a separate research interest a unified package of methods for computing orthogonal projections will be developed and distributed.

EIA-0116050
Michael A. McRobbie
Indiana University - Bloomington

MRI: Creation of the AVIDD Data Facility: a Distributed Facility for Managing, Analyzing, and Visualizing Instrument-Driven Data

This is a proposal for equipment acquisition under the Major Research Instrumentation (MRI) program to support research and student training across a broad range of instrument-driven data-intensive science areas. The proposed distributed facility for managing, analyzing, and visualizing instrument-driven data would address the data life cycle consisting of data capture and remote data reduction; high speed data transfer; real time data analysis and processing; data storage; data retreival; data analysis and postprocessing; data visualization; and the use of remote data stores. Among the research projects enhanced and enabled by the proposed facility are both computer science and applications area projects, for example work on end-to-end real time data management for remote control and use of beam-line systems by X-ray crystallographers

In the late 1980's the field of computational science and engineering emerged as a new discipline, one with a research core that generalized from its many applications and led to the creation os several academic programs at leading unversities. Computational science has since proven itself with many successes, but it has also evolved both in its tools, methodologies, and research challenges and goals. This workshop seeks to re-examine the area, and develop a new consensus on the directions it should be taking. A major goal of the workshop will be a document summarizing the accomplishments of the past decade, and prognosticating the future for scientific computing research.

0202048
Wise, David S.
Indiana University - Bloomington

RI: A Research Infrastructure for Collaborative, High-Performance Grid Applications

This project, developing an experimental infrastructure for distributed high performance computing, supports ten research projects extending the location-transparency that the Grid provides for computation resources to the full spectrum of activities which end-users require. Services being explored include software development, parallel code middleware, distributed software components for scientific computing, security for parallel remote method invocation,
managing large-scale data streams, and collaboration methodologies. The research builds on and extends the institutions collaborations with several national Grid research teams. In contrast to existing national and university infrastructure available through production machines, this research requires an environment tolerant of experimental network protocols, temporary middleware, and other system-level changes. The infrastructure will contribute to the following research projects:
a. Opie: basic work on parallel matrix algorithms that achieve high efficiency across many architectural platforms
b. LAM: middleware MPI implementations supporting hierarchical and fault-tolerant parallel computing
c. dQUOB: application of SQL queries to live data streams
d. RMI Security: basic research into security mechanisms for remote method invocation, allowing security to be traded off with efficiency
e. HPJ: High Performance Java creating a language platform for portable high performance coding
f. Grid Broker: reliable, robust publish/subscribe service for introducing fault tolerance into the distributed Grid environment
g. Community Grids Collaboratory: advanced collaboration capabilities with applications to both distance education and distributed communities
h. Xports: design of methodologies for remote instrument access and data management of the resulting extremely large data sets
i. Software Components: distributed software component model designed for applications that use parallel computing "nodes" in wide-area Grid environments
j. Science Portals: set of tools that allow programmers to build Grid distributed applications accessed and controlled from desktop environments and web browsers
Major improvements to infrastructure supporting all these projects include a 16-node cycle server and a large-scale file server as well as network upgrades to and within the building.

This proposal researches a Common Instrument Middleware Architecture (CIMA)
to improve accessibility of instruments and to facilitate their integration into the Grid. The proposed middleware will be based on current Grid implementation standards and accessible through platform independent standards such as the Open Grid Services Architecture (OGSA) and the Common Component Architecture (CCA). Emphasis will be placed on supporting a variety of instrument and controller types including creating a small implementation that can be used with tiny wireless controllers such as the Berkeley Mote sensor package as well as embedded PC-104- and VME-based controller systems. The proposed CIMA implementation will be evaluated in three settings representing a spectrum of shared instrument applications: X-ray crystallography at a synchrotron source, real-time acquisition of network performance data with embedded monitors, and small sensor network nodes using Berkeley Mote wireless sensors. The end product will be a consistent and reusable framework for including shared instrument resources in geographically distributed Grids.

Broad impact and intellectual merit. The proposed work will have important ramifications in the development of many instrument-driven Grid computing projects, and by extension, to many science education programs. Further implications exist for the development of industrial standards for networking instruments and international e-Science collaborations, as well as developing and evaluating new ubiquitous computing methodologies. The co-PI's are active in the Global Grid Forum, the DoE CCA Forum, and other Grid and HPC standards efforts, and together with industrial and DoD projects these will be used to evolve and promulgate the results of the proposed work. The co- PI's also have a track record of mentoring students from underrepresented groups, and have education projects at all levels from K-12 to postdoctoral.

The CrystalGrid Framework (CGF) project will research the acquisition, transport, and curation of data over the entire data space of the field of X-ray crystallography, addressing methods for managing wide heterogeneity in data representations, formats, data containers, administrative domains and diverse instruments and equipment. Until recently, individual labs have simply imposed local homogeneity of format and procedure, and not stored lab-dependent metadata. This ad hoc system is limited, however, as crystallographers begin to cross between labs to accomplish their research objectives, and as increasing numbers and sizes of output data streams leave less time for each investigation. Local workflow must be made explicit, procedures must be formally described, and the history and assemblages of data expressed in an open, shareable way. Creation and management of complete, accessible records for each experiment is critical, as well as heterogeneity in data acquisition and management across the field.

To meet that need, this project will develop a framework of web service interfaces and data and metadata systems addressing the whole spectrum of crystallography. Project participants and collaborators will leverage existing projects, such as Reciprocal Net and Common Instrument Middleware Architecture, that address narrower issues in the problem domain. The CGF will also draw on collaborating projects with overlapping areas of interest, such as the UK-based Comb-e-Chem project. The resulting framework will be a useful environment for crystallographic investigations and an extensible platform on which new web-based applications can be built.

The CGF project involves the classic problem of dealing with heterogeneity in data, procedures, and instruments in the crystallography application space, and another classic problem in integrating the entire data collection, transport, and curation requirements of the domain into a seamless beginning to end system. The challenge is to create a virtualization system that manages heterogeneity in more than a single aspect and to provide vertical integration using only open, extensible, and interoperable standards and methodologies.

While the project constitutes research into pertinent computer science problems, the plan for performing the research is centered on producing a product (the CGF) that will immediately be useful in addressing emerging technical problems in the field of X-ray crystallography. Within crystallography, one of the specific goals is to make structural results accessible that might otherwise never be seen, and so the CGF will help increase the body of scientific knowledge and improve the return on federal investment in the large numbers of x-ray diffractometers and associated instruments nationwide. Although the project targets specifically a few hundreds of crystallography labs worldwide, the software and methods created in it are intended to be reusable for any science moving from individual lab practices to a shared, global collaboratory system. In sciences such as high-energy physics and astronomy, the scientists have long shared single, unique, large instruments and had to create shared data management and instrument metadata. CGF is likely to be useful in other scientific disciplines which still use widely-distributed lab-based instruments that now need to be linked in data grids.

This project, creating a Data Capacitor and a Metadata/Web Services server, addresses two clear and widespread challenges: the need
-To store and manipulate large amounts of data for short periods of time (hours to several days) and
-For Reliable and unambiguous publication, discovery, and utilization of data via the Web.

The Data Capacitor, a 250 Terabyte short term data store with very fast I/O and the Metadata/Web Services server, a robust server, enable the institution and collaborators to adopt and depend upon the Web services for exchange of research data. Research and development efforts at IU will create the tools required for the Data Capacitor to be used to its fullest. Progress and research possibilities in many disciplines have been fundamentally changed by the abundance of data now so rapidly produced by advanced digital instruments. Scientists face the present challenge of drawing out from these data the information and meaning contained within. IU has established a significant cyberinfrastructure composed of high performance computing systems, archival storage systems, and advanced visualization systems spanning two main campuses in Indianapolis and Bloomington, and connected to national and international networks. This institution enhances its infrastructure in ways that will result in qualitative changes in the research capabilities and discovery opportunities of a broad array of scientist that work with large data sets. The Data Capacitor is expected to become a development platform and testbed for new cyberinfrastructure, as well as a proof of concept for large capacity, short-term storage devices. On the other hand, the Metadata/Web Services server enables the institution to establish a leadership position in standards-based data dissemination in many fields.

Broader Impact: The Data Capacitor enhances current practice in relevant scientific communities, enables technology transfer and commercialization, develops a 21st century workforce, and ensures public understanding of the value of science. Deliberate use of objective metrics in all areas of broader impact ensures that new discoveries, technology development, educational activities, and public information efforts translate into benefit for the scientific community and society as a whole. Women and underrepresented groups will be drawn into computing-intensive sciences and applications of computing.