Green Silicon
Stefan Wild
(Page 2 of 3)
“It’s not about making the models less detailed,” Wild says. “The original modeler is thinking of replicating some physical phenomenon, not the larger goal of using the model to improve some system. If they were to make the model faster, they’d try and make it simpler. What we’re doing is looking for mathematical ways to keep the detail, but boost the speed.” For Wild, optimization is about bringing math to the people — to turn an unusable model into a dinner-table standard.
It’s why he sought out his doctoral advisor, Cornell engineering professor Christine Shoemaker. She’s a world leader in the application of sophisticated computations to solve environmental problems and, like Wild, takes a “math with a mission” approach.
It’s also why Wild’s DOE CSGF practicum stint at Argonne National Laboratory, outside of Chicago, was so valuable.
The Laboratory for Advanced Numerical Simulations in Argonne’s Mathematics and Computer Science Division is a world leader in optimization technologies. The lab’s Toolkit for Advanced Optimization (TAO) is a collection of high-quality, high-performance codes, primarily for distributed computing applications, used by hundreds of researchers within DOE, industry and academia in the United States and beyond.
Wild spent the summer of 2006 at Argonne, working with practicum advisor Jorge Moré to create a new code for the optimizer’s toolkit: an algorithm specifically designed for engineers with computationally expensive models.
“The optimization technique that Stefan is developing is different from most,” says Moré, an Argonne staff scientist who develops algorithms and software for large-scale optimization problems, such as modeling nuclear energy production.
“Most of the optimization techniques we’re currently using require additional information from the user,” Moré says. “The technique that Stefan’s developing doesn’t require additional inputs, and this makes it much simpler and more user-friendly.”
« Previous 1 | 2 | 3 | Print Next »
