February 27, 2014

by Rob Mitchum, Computation Institute

After a successful first run last fall, the Computation Institute hosted a second round of lightning talks -- short talks about CI research and opportunities for students at UChicago and Argonne National Laboratory. This round of speakers focused on applications of computer and programming skills to areas ranging from social good to genomics to nanotechnology at CI centers and projects including Swift, Globus, the Data Science for Social Good Fellowship, and more. 

Ian Foster, Computation Institute

As the director of the CI, Foster delivered a broad overview of how the institute started and the research and resources it currently offers. Foster also briefly summarizes some of the research projects not represented in this lightning round, such as metagenomics research at Argonne, large-scale text analysis at the Knowledge Lab and ARTFL, computational economics, and the Hack Arts Lab.

"All of these applications have a common theme of very large-scale data analysis, often a lot of sophisticated computer modeling, and then the need to bring people together from different disciplines," Foster said. "That's a common theme of the work that we do."

Rayid Ghani, Data Science for Social Good

Last year, the Data Science for Social Good Summer Fellowship brought 40 students from around the country to Chicago for a summer of working on data projects with non-profits and government agencies. Ghani talked about the inspiration for the fellowship, the preparations for its second summer, and how his team is continuing to develop tools generated by the fellows for helping match high school students to colleges and optimizing lead inspections in Chicago.

"If you can work with computers and data and stats, you have these superpowers that very few people have," Ghani said. "These are superpowers because you can do magical things, like predict the future, you can influence people to change their minds and behavior, you can make organizations rational."

Mike Wilde, Swift

The world's most powerful computers today use parallel computing, a strategy for faster computation that allows hundreds of thousands of computing cores to work simultaneously on a single problem, turbo-charging simulation and data analysis. Swift is a parallel scripting language created by Wilde and colleagues to help scientists easily access this powerful approach. Wilde talks about their work building automated scientific workflows for areas such as drug discovery and climate science, and how Swift connects researchers with computing resources such as the Open Science Grid, Blue Waters, and the Argonne Leadership Computing Facility.

"Our goal is to help scientists use large-scale parallel computing more effectively," Wilde said. "We've created a language to make it easier for scientists to do what they always get stuck with on a day-to-day basis...the challenges of expressing how to run thousands or millions of copies of a scientific program and feed those results into the next chain of their workflow is really daunting."

Hakizumwami Birali Runesha, UChicago Research Computing Center

As the computational needs of academic researchers grow more complex, demand increases for high-performance computing and data visualization. In 2012, the University of Chicago created the Research Computing Center to provide these resources (including some tools developed by the CI) for faculty and students from all corners of campus. Hakizumwami Birali Runesha, director of the RCC, talked about the services and student job opportunities at the center, and mentioned a few of his favorite projects that have used the RCC, including a fossil study that transferred and analyzed thousands of CT scan images from a museum in Italy.

"The research really spans everything from physical sciences, biological sciences, and then humanities, Booth school, law school, and even the art school," Runesha said. 

Maria Chan, Argonne Center for Nanoscale Materials

Engineering is now precise enough to manipulate small amounts of atoms to create new types of materials that are stronger, more flexible, or more energy efficient. Argonne's Center for Nanoscale Materials is one of five national centers exploring the frontiers of this young science, and CI fellow Maria Chan uses computational modeling to interpret experimental results and develop the theory behind the center's work. 

"If you think of atoms as people, then atomic physics is like psychology or psychiatry and material science is like sociology," Chan said. "That makes nanoscale science is more like organizational management...how different individuals interact with each other and in small groups." 

Ravi Madduri, Globus

Increasingly, every scientist is forced to become a computer scientist in addition to their chosen discipline. Globus develops "services for science," helping tap the potential of cloud computing and science-as-a-service to help scientists with time-consuming computational tasks, such as data management, analysis of large datasets, and collaboration over long distances.Ravi Madduri is building cloud-based analysis platforms for working with large datasets in genomics, climate science, and other fields with computational challenges -- though as Madduri observed, "Any data that you cannot handle is big data for you."

"At CI you get to change the world, you can work on projects that have great impact," Madduri said.

Read the original versions of this article at http://ci.uchicago.edu/blog/lightning-talks-return-part-1 and https://www.ci.uchicago.edu/blog/lightning-talks-return-part-2