< The University of San Francisco: News

Computer Science Plots Earthquakes



Investigating earthquakes means studying past seismic events or, increasingly, analyzing elaborate computerized simulations – the latter of which Associate Professor of computer science Christopher Brooks will do as part of a shared grant worth $1.8 million from the National Science Foundation.

The multi-year shared grant allows for collaborative research from the University of San Francisco’s Brooks and scientists at the Information Sciences Institute at the University of Southern California, Lawrence Berkeley National Laboratory, and the University of Delaware through 2012.

Brooks intends to use $100,000 of the grant to make complex computerized modeling faster, easier, and less mysterious.

“The old-school way to do this is to look at past events – what happened last time there was a quake,” Brooks said. “Unfortunately, this doesn’t help with new types of quakes, and doesn’t account for any changes to the area since the last quake.”

From earthquakes and climate change to astronomical star formation, more and more of today’s scientists use large-scale simulations run on hundreds or thousands of clustered computers to study natural phenomena. But, with so many computers dependent on one another, crashes are frequent and simulations can take unexplained days or weeks to complete.

In one of the simpler cases Brooks’ team has solved, scientists noticed a prolonged delay in a simulation. Following weeks of painstaking analysis of computer logs, it was discovered that one of the hundreds of computers working on the simulation was set to the wrong time, delaying the others.

“On one hand, this is a trivial sort of error and not very interesting,” Brooks said. “On the other hand, it would’ve been very hard for the scientists to realize what the problem was without the analytical tools we’ve built.”

Using the $1.8 million grant, the team intends to engineer additional software tools to improve simulation project oversight, management, and debugging so that scientists can spend more time doing science and less time figuring out why a computer isn’t working properly, Brooks said.

By running such simulations, scientists hope to not only be able to predict earthquakes but, eventually, their severity. Information gleaned from simulations could lead to such practical applications as the improved construction and retrofitting for buildings, highways, and dams, Brooks said.

Included in the grant are funds for USF computer science students to become involved in developing tools to help scientists better monitor and debug their research – work that could open opportunities for USF students to work at Lawrence Berkeley National Laboratory, Brooks said.

Written by Edward Carpenter »usfnews@usfca.edu