LIVERMORE — Edmond Chow first came to Lawrence Livermore National Laboratory in 1986, as a "Superkid" — one of dozens of promising high school students invited to spend summers at the lab to work with supercomputers.

"I got my picture taken with the Cray 2, which at that time was the hottest computer around," Chow said. "It certainly broadened my scope, because I found out about all sorts of software and applications the average high school student does not have access to."

After earning his doctorate in computer science from the University of Minnesota, he was hired to work at the lab in 1998.

As the leader of the Complex Networks project at the lab's Center for Applied Scientific Computing, Chow's duties include helping the Department of Homeland Security sift through vast databases for leads in the war on terrorism.

Chow, 35, is as at ease with words as he is with numbers. In his spare time, the computational scientist writes and teaches poetry.

But given the abstract — and often secret — nature of his work at the lab, it's not always easy for him to describe to nonscientists exactly what he does for a living.

Often, he said, "They ask, and then their eyes kind of glaze over," when he tries to explain.

To a nonmathematician, even the problem that Chow tackled as a high school student growing up in Toronto sounds daunting.

Using an IBM personal computer, Chow calculated pi — the ratio of the circumference of a circle to its diameter — to a high degree of accuracy.


Advertisement

It was a challenging task for computers of the time, requiring many repetitive calculations. In order to solve the problem, Chow had to write software analyzing minute errors that might crop up. If allowed to "propagate," or repeat, the errors would compound themselves and throw the calculations off.

The problems Chow works on today are much more complex, and the computers he can call on to solve them far more powerful. But the key remains understanding how to break a problem down in order to harness the computing power at his disposal.

"As the size of the problem increases, the amount of work you have to do increases even faster," Chow said. "It requires (disproportionately) more (computing) time and memory."

During a recent visit, Chow explained a conceptual diagram of a problem he was working on for the Department of Homeland Security.

A grid sketched on the whiteboard in blue ink represented the problem. Red squares on top of the grid represented four layers of Blue Gene L, the world's fastest supercomputer. Each point on the grid represented a computer processor.

When fully built, Blue Gene L will employ 128,000 microprocessors, all running at the same time. Programming a computer that complex requires planning.

"Maybe three-quarters of the work happens before any code is written," Chow said of the process of mapping out a complex problem.

Chow's 1997 doctoral thesis involved showing how computers could be used to simulate complex real-life events that required them to solve millions of equations at once. Today, simulations are considered large if they involve hundreds of millions of simultaneous equations, Chow said.

Supercomputers can be programmed to model events such as the Earth's weather or a nuclear explosion. They also are being used to glean information from massive databases far too voluminous for humans to sift through.

"A lot of the effort at many places is how to get computers to read information, and then present the most important information in a summary a human can understand," Chow said.

In the past year, Chow's team has performed a "very large calculation" that involved reading 1 billion pieces of data, he said.

The job, performed for clients that included the Department of Homeland Security, was first proposed two years ago.

"We didn't know if we could do it," Chow said. "The computation involves searching a very complicated database with varied types of information. To do it required some fundamentally different approaches people had not thought about in the past, because they were thinking of simpler organizations of the data."

Chow's work on simulations earned him a 2002 Presidential Early Career Award for Scientists and Engineers, which recognizes the accomplishments of promising young researchers. Although awards are nice, Chow said he and his colleagues are motivated by the satisfaction they receive from their work.

"The reason a lot of us do this is for those rare times when we have a really good idea, and we get to bring it to fruition, deploy it, see it work," Chow said. "Those are rare times, but to me, that is where the reward is."

Born in Taiwan in 1969, Chow emigrated to Canada two years later. He and his wife, Ruth, settled in Dublin four years ago after living in Livermore and Pleasanton.

His job at the lab is a never ending one, and can become intertwined with everyday life.

"I don't want to characterize people as workaholics, but it is very easy to blur what is work, and what is your own free time," Chow said of the work ethic at the lab. "I've gotten a lot of good ideas in the shower."