KDI Home    Contact Us   

Jerry Marsden's ambitious KDI project to advance simulations of ocean and earth systems brought together mathematicians, computer scientists, and geophysical scientists from four California universities (Caltech, UC San Diego, UC Davis, and UC Santa Barbara). Marsden and his group overcame distance, bureaucracy, and everyone's busy schedules to achieve success, according to an NSF evaluation study of the KDI program.

To collaborate on projects like this one, scientists have to overcome obstacles of geographic distance, disciplinary training, institutional structures, and sometimes even personality. James Watson and Francis Crick, physicists-turned-biologists who discovered DNA structure in 1953, shared the same office in Cambridge and talked for hours and days on end. But their achievement depended on the independent work of Rosalind Franklin, a crystallographer, at King's College in London. Franklin was skeptical, and it was only through a friend's recommendation that Watson and Crick obtained the critical photographs needed to make their advance.

One of the first steps Marsden and his collaborators took when they were awarded a special KDI grant was to schedule annual workshops, bring postdoctoral researchers onto the project, and institute a program for postdoc and student exchanges. The researchers created Web sites to report on meetings and share papers. They created tutorial lectures, and continued communication and visits throughout the project. Postdocs helped supervise and monitor the work of students. In a workshop for KDI grantees, mathematician Steve Shkoller said these systematic steps to foster communication across the disciplines helped the mathematicians on this project identify the real needs of physical scientists, which then allowed them to develop mathematical tools.

Other successful projects identified by the evaluation also faced obstacles in their collaborations. Daniel Joseph's project aimed to create 3-D simulations fundamental to the chemical process for oil exploration and recovery used in industry. The project linked engineers and computer scientists from Minnesota, Texas, Pennsylvania, and Stanford. One strategy this collaboration adopted was to ask the postdocs to learn techniques from the other discipline. For example, computer science (CS) postdocs learned about computational fluid dynamics (CFD), and CFD postdocs learned about CS. Another strategy the project adopted was to hold seminars at one university every 6 months to evaluate progress. This multi-disciplinary research led to a new discipline—"direct numerical simulations for multiphase dynamics."

Even projects within the same university can experience obstacles to collaboration. Mark Embrecht's impressive project to explore the discovery of new pharmaceuticals through database mining involved researchers from the engineering, chemistry, and mathematics departments at Rensselaer Polytechnic Institute. Though weekly group meetings were convenient because all principal investigators were based on the same campus, two who attended the KDI workshop remarked on how long it took to get the collaboration going. They estimated that it took almost a year before they were really able to make progress. Because the grant was only for 3 years, as soon as results starting coming in (e.g., their team won two data mining competitions), they had to focus on securing additional funding.

Integrating knowledge from different disciplines and learning to work together in a harmonious way are only two of the many challenges that KDI researchers faced. A project at Arizona State University under the direction of Anshuman Razdan had the goals of developing a software library kernel, tools for data archiving, and an Internet-accessible interface to let people construct customized search engines. Because the project involved researchers from areas such as engineering, computer science, art, and anthropology, there were no clear norms within the university for advising graduate students across departments. The international journals or conferences available for multi-disciplinary research were few, and other members of the research community did not readily see the value of this research. Despite these institutional barriers to collaboration, the project successfully developed a software database that had immediate application to important problems in the biosciences, in biotechnology, and in anthropology.

Back to Top of Page