Glimpsing glueballs in collider debris.

A calculation that took 2 years on a powerful special-purpose computer has provided evidence that a hypothesized subnuclear particle called a glueball actually exists.

The result suggests that glueballs may be observed in particle accelerators when electrons or protons and their antimatter counterparts collide at high energies. Until now, glueballs had gone unrecognized because theorists had been unable to provide sufficient information on distinctive characteristics that would distinguish glueballs from other particles.

Physicists James Sexton, Alessandro Vaccarino, and Donald Weingarten of the IBM Thomas J. Watson Research Center in Yorktown Heights, N. Y., describe their computation as "the largest single numerical calculation in the history of computing." They report their findings in the Dec. 18, 1995 Physical Review Letters.

The team based its calculation on a simplified version of the theory of quantum chromodynamics (QCD). This theory describes the force that binds different quarks and antiquarks together to create protons, neutrons, and other subatomic particles.

Just as an electrically charged particle generates an electric field, a quark gives rise to a so-called chromoelectric field. This force field can also be described in terms of the actions of particles called gluons, which shuttle between quarks, seemingly gluing them together.

Quantum chromodynamics theory predicts that under certain circumstances, gluons themselves can stick together briefly to form composite particles called glueballs. However, the great difficulty of solving the relevant equations had prevented theorists from determining the masses and lifetimes of these hypothetical particles.

To help guide the search for glueballs, Weingarten and his coworkers turned to a simplification of quantum chromodynamics. In this formulation, quarks and antiquarks sit at points in a finite, four-dimensional lattice, and gluons correspond to the links between these points.

By solving the equations for a large number of quark and gluon arrangements, researchers can deduce characteristics of quark-containing particles. Increasing the number of points and expanding the region covered by the lattice, while decreasing the distance between the points, brings this approximation closer to the continuous space and time of the full theory. But this improvement occurs at the cost of greatly increased computation time.

To speed up the calculations, Weingarten and his coworkers used an experimental computer designed and built especially for this task. Called the GF11, it has 566 processors, each a powerful computer in its own right.

In 1993, the IBM team succeeded in computing from theory the masses of eight quark-containing subatomic particles (SN: 5/22/93, p. 325). Soon after, they calculated that the lightest glueball would have a mass (expressed in energy units) of about 1,707 megaelectronvolts (MeV).

To determine whether such a glueball would stick together long enough to be observed in a particle accelerator, the researchers calculated the glueball's rate of decay into different combinations of other particles.

The calculation demonstrated that a glueball has a sufficiently long lifetime for the particle to be detectable. Indeed, it's possible that physicists have already sighted a glueball in accelerator experiments. The best candidate is a particle labeled fJ (1710), which appears as the product of a quark-antiquark annihilation and has a mass of 1,710 MeV.