Testing theory by computing quark behavior.

It took nearly a year to do the calculations, but when the computer finally disgorged the numbers, physicists had for the first time extracted from theory predictions of the ratios of the masses of eight subatomic particles. These computed, theoretically derived ratios differ from experimentally observed values by less than 6 percent.

The results provide the strongest quantitative evidence yet that quantum chromodynamics (QCD) is correct, says Donald Weingarten of the IBM Thomas J. Watson Research Center in Yorktown Heights, N.Y. Quantum chromodynamics equations describe the characteristics and behavior of quarks and the peculiar force that binds different quarks and antiquarks together to create protons, neutrons, and other subatomic particles known as hadrons.

Weingarten and his co-workers report their findings in the May 10 PHYSICAL REVIEW LETTERS.

The IBM group did their calculation using an experimental supercomputer known as the GF11, designed and built at IBM specifically for this task (SN: 8/10/85, p.88). The computer, which fills a large room, has 566 processors--each a powerful computer in its own right -- that operate together in various combinations. "It's a big, complicated machine," Weingarten says. "It took a while to get it debugged."

Even with a specially designed super-computer on hand, the researchers had to adopt certain approximations to simplify their QCD calculation so that it could be completed within a reasonable time. For example, like most other groups studying QCD, they used a so-called lattice formulation of the theory, in which each point, or node, within the lattice corresponds to a particular set of quark and antiquark positions and a given geometry for the force field acting on the particles.

Researchers can then calculate the probability that a certain quark-antiquark combination will shift from its initial state to a new state - that is, from one node to another in the lattice. These transition probabilities provide the raw numbers from which theorists can deduce such particle characteristics as the mass ratios of hadrons.

Because the calculations must be done repeatedly for the many nodes in a typical lattice, this tedious but necessary procedure consumes a vast amount of computer time. As the lattice size increases and the spacing between the nodes decreases to bring this approximation of QCD closer to the full theory, the required number of calculations escalates tremendously.

The IBM effort represents the first full calculation of hadron masses from QCD, Weingarten says. Previous calculations by a number of other groups either were incomplete or were used as tests of the computational methods themselves.

"I feel that we've reached a milestone," Weingarten says. "On the other hand, the fun is just beginning. Now that we really know how to calculate real numbers, there's a whole bunch of interesting things ... we'd like to go out and learn [about QCD]."