Bringing High-Powered Computing to Other Fields

Thursday, November 8, at 12 noon, U.S. Eastern time

Sirin Tekinay leads the Cyber-Enabled Discovery and Innovation program and is chairwoman of its implementation team at the National Science Foundation. She joined the Division of Computing and Communications Foundations at the agency as its program director for theoretical foundations in communications research in 2005. She is also an associate professor of electrical and computer engineering at the New Jersey Institute of Technology, where she has been a faculty member since 1997. She holds seven patents.

A transcript of the chat follows.

Dan Carnevale (Moderator):
    Welcome to today's Brown Bag chat. I'd like to thank Sirin Tekinay for taking the time to answer our questions. Let's get started.

Question from Dan Carnevale:
    I'd like to start off with more information about Cyber-enabled Discovery and Innovation. Is this a grant program just for big universities with big supercomputers? How are smaller research programs contributing?

Sirin Tekinay:
    Thank you! This was a point I wanted to clarify before saying anything else: CDI is NOT based on high-performance computing. It is based on "computational thinking," i.e., computational concepts, models, methods, algorithms and tools. It promotes "mental" component of computation as much as, if not more than, the "metal." The basic philosophy of the program is to broaden participation to traditionally underrepresented groups, and unconventional partnerships. In fact, the larger scale, research center type activities will be supported starting 2009; in 2008, we are welcoming smaller research programs.

Question from Charles Greenberg, Yale University:
    Would you recommend that grant proposals include the academic library and IT structures in either the development or local evaluation stage?

Sirin Tekinay:
    First of all, let me say these are excellent topics that can be subject matter in CDI proposals. Please note, we have been avoided prescribing topics or application areas in the program solicitation. Rather, we have formulated general themes including computational tools, virtual organizations, building and assessment of socio-technical structures, and demonstration of advancement in research and education of science and engineering. The aim is to transform research and education such that transformative outcomes can be obtained. We call this the "transformative squared" effect.

Question from Jeffrey Brainard, Chronicle of Higher Education:
    How will this project build on previous federal programs on computing, such as the Networking and Information Technology Research and Development Program?

Sirin Tekinay:
    CDI will leverage past investments in such endeavors as Cyberinfrastructure, and learning programs. NSF's part in NITRD that you mentioned, has been a factor in formulating CDI. We would like to see CDI become more than an NSF program, rather, a vision shared by the federal agency family and industry, to transform all frontiers of science and engineering. Therefore, we are encouraging partnerships beyond federally funded academic research: industrial collaborations, international partnerships...

Question from Dan Carnevale:
    So how does this advanced computational thinking work, and what are the results of such an approach?

Sirin Tekinay:
    Computational thinking is the approach scientists and engineers (therefore, future generations of scientists and engineers) have to take to make sense out of data, derive knowledge, get insights into complexity, build new socio-technical systems... essentially move science forward.

Question from Dan Carnevale:
    What are some examples of the areas and disciplines that can benefit from this type of computing that we wouldn't necessarily expect?

Sirin Tekinay:
    Excellent question! All areas of science and engineering: from life sciences to hard engineering problems, from math to social and behavioral, including economic and political sciences... For example, before computational means could help us, biologists were unable to decode the human genome. Geoscientists could not produce real time maps showing global scale phenomena with data from satellites in real time. Of course, the most obvious example is the Internet allowing us to have this forum right now! We would like to see it "jump" to its next generation, given what it was engineered for originally versus what we need it for now.

Question from Thomas Cundari, University of North Texas:
    Hello,

Is an emphasis to bring together teams from different disciplines (e.g., chemistry and engineering) to develop novel ways of solving multi-scale issues, and, in doing so, help better train the next generation of modelers to better succeed in the innovation process.

Sirin Tekinay:
    Absolutely! We are looking for those partnerships that can end up in "transdisciplinary" findings, drawing the best from a multitude of disciplines that were not interacting to date.

Question from Dan Carnevale:
    There seems to be a big trend toward virtual environments. How do they really enhance research and learning? Do the results really justify the computing power needed to create the virtual reality?

Sirin Tekinay:
    Assessment of virtual organizations (VOs)is a part of CDI. VOs do not necessarily mean virtual reality; they are socio-technical systems that enable remote access to resources, collaborations, education and training that would otherwise not be possible, or have logistical costs. Their feasibility is expectedly manifold: energy savings, for instance, in transportation costs, come to mind.

Question from Jeff Stanton, Syracuse University:
    One of the thematic areas in the RFP is "Understanding Complexity in Natural, Built, and Social Systems." Can I safely infer from the "Social" in this statement that proposals focusing on the transformative analysis and use of social science data could be considered viable?

Sirin Tekinay:
    Sure. Social, behavioral, and economic sciences are already interacting with other fields such as computer science, communication and information systems, biological modeling, etc. We fully expect innovative proposals in these areas and beyond.

Dan Carnevale (Moderator):
    We're about half way through the chat. Keep those questions coming.

Question from Yongwu Rong, George Washington University:
    The idea of CDI encourages people to collaborate each other. Is the track record of past collaboration important? Say, if I have a research team with good ideas along the line of CDI, but no history of collaboration, would that make it rather non competitive?

Sirin Tekinay:
    On the contrary. We are encouraging NEW partnerships. Certainly, a track record of recent success is a plus, but not a must. CDI is looking for discovery and innovation via new collaborations that traditionally were not in place.

Question from Bob Harbort, Southern Polytechnic State U:
    We are a masters level teaching university. Do you have a suggestion how we might participate?

Sirin Tekinay:
    Institutional diversity is one of our priorities. In CDI, our usual "broadening participation" criterion is further broadened. We look forward to undergraduate, and graduate level research and education efforts pushing the envelope.

Question from Dali Wang, SURA:
    Would you mind talking about the review procedure, any different from previous program, such as ITR, etc.

Sirin Tekinay:
    We are ready for the challenges of a large volume, multi-disciplinary review process. We have a big team of NSF officers in place, in addition to a multi-tiered submission process spanning eight months: Letters of Intent will help us set up the review panels; preliminary proposals will then be panel-reviewed; finally, full proposals will be invited to be reviewed in panels. We have learned from previous programs of comparable size and scope, and we are drawing from the best practices of the entire foundation, to make sure we are up to the "golden standard" NSF boasts among funding agencies.

Question from Jeffrey Brainard, Chronicle of Higher Education:
    The NSF has labored -- and struggled at times, it seems -- to develop a working definition of "innovation" as a criteria for grant awards. How will you define the term for the NSF's request for proposals for this grant competition? Can "innovation" be meaningfully defined in a few sentences?

Sirin Tekinay:
    We have had many hours of discussion on how we can define "transformative" research. The words "potentially transformative" are now officially in NSF's review criteria. Clearly, we cannot predict the future and judge if a proposal is going to produce transformative outcomes. However, we can rely on the researchers' creativity and expert reviewers' and NSF officers' judgements to evaluate proposed work in terms of having that potential.

Question from Tom Cundari, U, of North Texas:
    Would it be an advantage to integrate experimentalists into a CDI proposal to provide "real world" systems for analysis, and also raise the potential for putting the results of "new ways of computational thinking" into practice?

Sirin Tekinay:
    We expect experimental research will be supported by CDI; as part of all three themes of the program: for example, experiments will enable advancements in deriving knowledge from data, obtaining new insights into complexity, or building and assessing virtual organizations with remote experiments and data sharing.

Question from Jim Zheng, Medical University of South Carolina:
    If we apply "computional thinking" to a disease related area, will that be encouraged, since disease related research are typically funded by NIH?

Sirin Tekinay:
    CDI will likely overlap, especially in application areas, with other funding agencies. The defining difference will be innovations in, or innovative use of, computational means, in addition to demonstration of advancing a multiplicity of disciplines. For instance, disease outbreak models are also useful for understanding the proliferation of spam on the Internet.

Question from Jeff Stanton, Syracuse University:
    Thanks for your earlier answer. One hears "through channels" that NSF is (or maybe just panel reviewers are) encouraging PIs and their teams to "open source" any software that is developed from NSF funded work. For the CDI competition, how might a PI's chosen IP licensing or distribution model affect the viability or attractiveness of a proposal?

Sirin Tekinay:
    Sure, Jeff. First of all, as a federal agency, we would like to see the most public good out of our investments. CDI especially promotes computational means to be useful in areas that did not take advantage of such means before. CDI investment is likely to result in savings in the way research and education are conducted: efficiency in the use of computational concepts, models, methods, algorithms, and tools will avoid re-inventing the wheel with data collection, software writing, and tool building.

Question from Bob Harbort, Southern Polytechnic State U:
    Would proposals dealing with development of interdisciplinary educational programs involving high performance computing fit into this program?

Sirin Tekinay:
    Precisely!

Question from Dave Leisawitz, NASA Goddard Space Flight Center:
    Are scientists employed by Government research labs eligible to propose, perhaps in collaboration with our colleagues in academia?

Sirin Tekinay:
    There are exceptions, however, as a rule, NSF cannot directly fund other federal agencies. Scientists with such affiliations are certainly welcome to contribute to CDI proposals as intellectual leaders or consultants.

Question from Yongwu Rong, GWU:
    There will be Type I and Type II awards. Does a Type I proposal compete just with other Type I proposals, or does it compete will both types? And, if Type II proposal is competitive, but with excessive budget, would it be possible to trim it down to a Type I award?

Sirin Tekinay:
    The review processes for the two types of proposals are kept separate; e.g., as you said, Type I proposals will compete with other Type I proposals. The type of a proposal is self-identified by the proposal team, and it is not defined by the budget. It is defined by the total effort and scope of the proposed work. Therefore, the type of a proposal is not allowed to change through the review and award processes.

Question from Jeffrey Brainard, Chronicle of Higher Education:
    For some grant competitions, the NSF has limited the number of applications that a research institution or investigator may submit, a rule that has irked some universities. Does the NSF plan such a limit for these grants, and if so, what is the rationale?

Sirin Tekinay:
    As specified in the solicitation, there is no institutional limit for CDI submissions; however, a researcher can be named in at most two proposals per competition cycle.

Dan Carnevale (Moderator):
    That's all the time we have for today's chat. Thanks again to Sirin Tekinay for providing such great answers to all our questions.

Sirin Tekinay:
    Thank you all for participating! Please keep in touch. We look forward to your innovative ideas!

Dan Carnevale (Moderator):
    Be sure to join us next week when John Nicholas of Beloit College talks about disaster preparedness.