(227a) Chemical Engineering and the Computational Grid

For the biological and chemical process systems communities,cyberinfrastructure must include a full spectrum of capability. Problem demands predict a rapid build up of need for increasingly sophisticated computational resources and the mathematical and computer science expertise to move to new plateaus of application code. There is high demand for computational reliability, excellent utility, and the ability to manage data now. Particularly strong is the need to manage dynamic data and information flow and the demand for tools to interpret data, mine data and extract knowledge. Another major need is the solution of very large-scale simulation and optimization models that are currently intractable. Visualization is seen as an increasingly important priority along with data interpretation. There is high demand for algorithm/software engineering and there is excitement about community-based models. Data security is not a value-add for the industry, but is recognized as a necessary priority.

As an example of cyber infrastructure development, this paper describes the build out of the University of California computational research enterprise to provide researchers the necessary spectrum of capability through grid access to university, inter-university and national resources. The ultimate vision of the University of California Grid Project is an computational research enterprise comprised of virtualized resources across geographical locations, where any given location could be optimized for a particular type or class of service and then be made available to a wide range of researchers through a unified, web-based Grid interface. Extremely high performance computational, visualization and data services can be made with greater capability and capacity for a given research group than any one group can provide through individual resources, resources can be better utilized and more researchers can have access to better resources. The research capacity and capability of the individual researcher and the research enterprise are raised up together without any one group or facility being over-burdened by an attempt to provide all services with their own resources.

Looking beyond UC, the UC Grid provides connectivity to TeraGrid resources at SDSC. Work is underway to provide Open Science Grid connectivity as well. The national cybert infrastructure has been funded largely by the National Science Foundation and the Department of Energy. The primary NSF initiative is under the heading of TeraGrid. Similar international efforts such as the United Kingdom's e-Science program follow similar approaches. In terms of current allocation percentages and overall current usage of the TeraGrid, chemistry, biology, chemical and biological processes and related fields have a rather light representation. Computational chemistry and bioinformatics are the two related subfields that are benefiting the most from the use of this capability.

The ultimate design goal is an overlay to computing, storage, modeling and simulation resources across the UC system with full capability integration with other Grids world-wide and into the future, UC Grid technology will enable grid-to-grid connectivity, which will connect the UC grid to other universities, research labs, and supercomputing centers around the country.