Math whiz gets to show up his elders

The San Diego Union - Tribune. San Diego, California. Aug 21, 2001; Jeff Ristine, Staff Writer

Abstract:
"What [Kevin Chan] did should be regarded as surprising," said Robert Leary, a senior staff scientist at the center and Chan's mentor. Resources to do such work tend to be limited, Leary said, and while others previously "found an answer for 78 (atoms), Kevin found a better answer."

These nanoscaled atomic clusters are too small to be directly observed. But using supercomputers, Chan eventually simulated a double icosahedron -- two 20-sided clusters of atoms stuck to one another at the side like soap bubbles containing a combined total of 78 neutral atoms.

Kevin Chan found an atomic needle in a mathematical haystack.

During an internship at the San Diego Supercomputer Center, the Del Mar resident and Harvard University sophomore amazed the pros this summer with his work involving clusters of atoms and the energies they contain.

After more than 20,000 hours of computer processing time, Chan, 18, discovered a hypothetical arrangement for 78 atoms that seems to represent an ideal in nature: an energy minimum.

To a specialized group of scientists and mathematicians, it was a big deal. Chan found a new energy minimum for one particular atomic cluster and thus added one small bit of information to the field.

"What Kevin did should be regarded as surprising," said Robert Leary, a senior staff scientist at the center and Chan's mentor. Resources to do such work tend to be limited, Leary said, and while others previously "found an answer for 78 (atoms), Kevin found a better answer."

Kevin Chan, with a model of an icosahedron, actually two, one inside the other. He solved a computational problem involving atoms forming the structure. (Photo: Jim Baird / Union-Tribune)

Chan used a mathematical strategy developed by Leary to make his discovery just a few weeks into his internship.

"Dr. Leary had mentioned that it was possible we would find some new structures, but this was kind of unexplored territory," said Chan, a graduate of The Bishop's School in La Jolla.

Esoteric as it may appear, however, the chore holds at least passing relevance -- get this -- to mad cow disease.

What?!

Mad cow disease, a human equivalent called variant Creutzfeldt- Jakob disease and other deadly neurodegenerative conditions all involve a misfolded, infectious form of protein called the prion.

Good prions and bad ones are "folded" differently, and current thinking holds the difference has something to do with the geometric configuration of their atoms. There are good shapes and bad ones; infectious prions are associated with the bad ones, and they have different energies.

In doing mathematical work on atomic clusters, scientists hope to gain insights on how proteins get to their lowest energy state.

Which brings us to Chan's work, called global optimization.

Chan, who is majoring in math, was participating in Research Experiences for Undergraduates, a National Science Foundation program to give students a chance to see what research is like and work on important projects in their field.

The San Diego Supercomputer Center, a national laboratory for computational science and engineering at UCSD, welcomed him for 10 weeks.

Chan said he hadn't worked on atomic clusters before, but knew something about global optimization and enjoyed studying symmetry.

Kevin Chan was "surprised and excited" when he realized he had made a research breakthrough. "When you're an undergraduate, you're just kind of learning what's going on and how to conduct research," he said. (Photo: Jim Baird / Union-Tribune)

Leary introduced him to a method he had developed called "basin- hopping," which methodically jumps from one geometric configuration to another in search of lower and lower energies, or "valleys" as Leary calls them. The very bottom is the lowest energy configuration preferred by nature for that cluster.

"There are a lot of valleys," said Leary. "If you wanted to put a number (on it), it would be a 1 with 30 zeros after it."

Leary likens the forces between atoms to the tension of a spring, which can be pulled or stretched. These forces hold the cluster together in a particular geometry that represents the lowest energy configuration.

These nanoscaled atomic clusters are too small to be directly observed. But using supercomputers, Chan eventually simulated a double icosahedron -- two 20-sided clusters of atoms stuck to one another at the side like soap bubbles containing a combined total of 78 neutral atoms.

Chan and Leary threw as much computational power at the cluster simulations as they could muster, and even used the center's IBM Blue Horizon -- one of the most powerful computers in the academic world - - for part of their work.

"At one point we were using 256 processors at a time," Leary said.

Previous efforts by more experienced scientists to accomplish the same goal had arrived at a much different shape. Chan's is lower in energy and therefore appears to come closer to representing the true, physical form of one particular class of 78-atom clusters in nature.

Among other reasons, scientists want to know the shapes of these clusters because their properties tie into bigger concerns in nature.

"We believe that the algorithms that we're using on the clusters can be used on proteins," said Leary.

Chan said the new structure "popped up unexpectedly" and he was "surprised and excited" when he realized it was something that hadn't been found before.

"When you're an undergraduate, you're just kind of learning what's going on and how to conduct research," he said. "Sometimes you don't get to go into the real interesting stuff that makes a real contribution to science. I was happy to be able to do that."

[Illustration]
2 PICS; Caption:

1. Kevin Chan, with a model of an icosahedron, actually two, one inside the other. He solved a computational problem involving atoms forming the structure.
   
   
2. Kevin Chan was "surprised and excited" when he realized he had made a research breakthrough. "When you're an undergraduate, you're just kind of learning what's going on and how to conduct research," he said.

(B-5); Credit: 1,2. Jim Baird / Union-Tribune