By Jacob Berkowitz
PARTICLE PHYSICIST DR. ROBERT (BOB) MAWHINNEY paints a picture of the subatomic world that makes you want to hit the replay button. It’s dizzying. You lean closer, squint, as he describes his specialty, quantum chromodynamics (QCD), the study of the interactions among particles that have color charge — quarks and gluons.
OK, quarks aren’t that foreign. The six flavors of quarks are the bare-bones constituents of all matter. We are mostly quarks. Combine two up-quarks and a down-quark and you have a proton. As for color charge (hence the “chromo” part of QCD) it’s the quark-gluon equivalent of the electric charge in electromagnetism. Just as electrically charged particles interact by exchanging photons, in strong nuclear interactions color-charged particles interact by exchanging gluons.
When it gets messy, even for QCD devotees, is when you try to calculate the interactions between quarks and gluons. Sure, at high energies this world of strong quantum fluctuations is quite linear. It’s at low energies, at the level of the proton, that things get messy. A quark turns into a quark plus a gluon. Then it interacts with another quark via a second gluon, while the first gluon turns into a quark/antiquark pair, then back to a gluon, and is reabsorbed by the first quark.
“You just have this enormously complicated, seething nonlinear sea of virtual particles. It’s a perfectly posed computational problem,” says Dr. Mawhinney, a professor of theoretical physics at Columbia University. “We have great faith in the underlying QCD equations because they’re built on the principles of relativity and quantum mechanics, and at high energies, where we can calculate analytically, the results agree with experiment. But there are many phenomena predicted by the equations that we couldn’t calculate with pencil and paper. So, it comes down to a question of computational strength to be able to calculate the physical consequences.”
To do this, Mawhinney is part of a team from Columbia, the RIKEN Brookhaven Research Center (RBRC) at Brookhaven National Laboratory (BNL) and IBM that is designing and building the latest in a series of massively parallel computers to numerically probe the world of QCD. The 10-Tflops peak performance supercomputer, dubbed “QCD On a Chip” (QCDOC) is set to be booted-up at BNL in early 2004. QCDOC won’t just push the bounds of quark-gluon physics mdash; including understanding of the universe’s early evolution mdash; it’s also a world-leading demonstration of the power and promise of topical computing.
Custom Made, Please
“There are only a few machines in the world that are built specifically for a topic,” says Ed McFadden, who manages Brookhaven’s Scientific Computing Facility, where QCDOC will be housed. But the 40-year BNL veteran (whose first computational color problem had nothing to do with quarks, but rather with solving the four-color map problem) notes that these purebred machines are attracting more interest than ever, including the interest of DOE’s SciDAC (Scientific Discovery through Advanced Computing) program, which is partially funding QCDOC’s design.