Next: Why and wherefore an Up: Goals, Objectives and Areas Previous: What's our plan?
Why is it a good idea?
The benefits can be best seen through two case histories. One case is
internal to Northeastern University; the other external.
Remarkably, both case
histories came about during the writing of the preproposal and proposal
for the Institute. In both cases, the right connections with other
Northeastern faculty came about only because the faculty were already
in discussions in the context of the Institute proposal.
Thus, even before the Institute is funded, some of the benefits of the Institute are bearing fruit--simply through better communication in the context of the proposal writing process. Of course, the proposal writing process is not a stable long-term activity to encourage this interdisciplinary collaboration. Resources (graduate and undergraduate students with appropriate training, seed funds to build prototypes for future grant proposals, etc.) are needed to institutionalize this unconscious potential.
- Case History 1: External opportunity - GAMMA software.
Professor David Budil, Chemistry, was contacted by Dr. Louis-Claude
Brunel, senior scientist at the National High Magnetic Field
Laboratory (NHMFL) at
Florida State University. The software GAMMA (General Approach to
Magnetic Resonance Mathematical Analysis) is developed there. The
imminent departures of a key developer and a graduate student has
caused the NHMFL to seek outside assistance, since the GAMMA
software is critical. Prof. Budil, a NHMFL collaborator with
expertise in spin and molecular dynamics, seemed a natural fit.
Though he himself cannot provide the software expertise, he can
refer the problem to the Institute which has the resources and
expertise and will be able to provide the support, perhaps on a
consultant basis.
In addition to its need for new theoretical capabilities to address emerging problems in structural biology, GAMMA faces some serious issues of software integration that must be addressed in order to expand both its capabilities and its user base. Such questions would ordinarily be outside the scope of Prof. Budil or any other single investigator's expertise. However, this issue of retrofitting new features is exactly within the scope of the Institute. Only through an existing organization, which is prepared with the proper resources, can we take advantage of opportunities to design software with worldwide impact and visibility. Our further pursuit of the GAMMA opportunity is described in Section
![[*]](icons/crossref.png)
.
- Case History 2: Internal opportunity - Brain Scans.
Prof. Miller has been working on brain scan simulations in software as
part of his work in 2=0the Center for Subsurface Sensing and Imaging Systems (CenSSIS6)CenSSIS.
His simulation was proceeding slowly. This task is highly computationally
intensive and must be done hundreds if not thousands of times as part
of a larger imaging algorithm. Luckily, many of these simulations are
independent of one another and thus can be distributed in parallel over
a collection of individual processors. Prof. Miller needed his software
to be parallelized for faster performance and also needed non-trivial
human resources.
The difficulty here is that distributed or parallel implementation of algorithms is a very specialized computer science task. While it is absolutely necessary for CenSSiS to attack real world problems, distributed computing is not a central focus of this Center's research and thus not something on which a 3=0the Center for Subsurface Sensing and Imaging Systems (CenSSIS7)CenSSIS funded research assistant could really work.
As part of writing the pre-proposal for the Institute, Prof. Miller became aware (a) that Prof. Cooperman was working on extensive parallelization of physics-based simulations and (b) that the 32-node Beowulf cluster, funded through a previous joint grant proposal by the Department of Electrical and Computer Engineering and the College of Computer Science was available. The two professors worked together to solve the problem. They hired and were able to train a bright undergraduate to use Cooperman's TOP-C parallel package and sophisticated software tools to do the job in a few months. The parallelization was successful, and the highly-qualified student is now prepared to go on to use the hands-on experience gained to rapidly extend this software.
Next: Why and wherefore an Up: Goals, Objectives and Areas Previous: What's our plan? David H. Lorenz 2002-03-25