D.V. and G.V. Chudnovsky, who have calculated pi to eight billion digits, plotted the first million as a random walk in 3-D form, shown here.

Beginning 3.14159..., never cycling, and going on forever, pi's digits certainly look random. Indeed, everybody "knows" they are, but nobody has proved it or understands why.

David Bailey may be getting close. Bailey is chief technologist for the National Energy Research Scientific Computing Center (NERSC) and an innovator in the field of experimental mathematics -- a pursuit that has brought him and his colleagues tantalizingly close to solving one of the mysteries of pi.

Bailey was born in Provo, Utah, where his neighbors, mathematicians on the faculty of Brigham Young University, became his role models. "I remember learning in the fifth grade that there were formulas you could use to calculate surfaces and volumes" -- for example, that a sphere's surface area is equal to pi times its diameter squared -- "I thought this was wonderful!"

Armed with curiosity and books from the public library, Bailey taught himself enough math to win school prizes: one was a reference he still uses, CRC Standard Mathematical Tables. As an undergraduate at Brigham Young, his precocity landed him a job assisting his physics professor in programming computers to do science. "I cut my teeth on computers at BYU."

When he received his doctorate from Stanford in 1976, the market for mathematicians was so bad that Ph.D.s were driving cabs and delivering mail. Bailey's computing experience saved the day. He worked for institutions like TRW Inc. and SRI International until 1984, when he moved to NASA Ames Research Center.

There Bailey was in charge of "shaking down" NASA's first Cray-2 supercomputer; to test its ability to sustain a long calculation, he had two of its four processors compute the first 29 million digits of pi. "Much to the consternation of Seymour Cray, they came up with two different answers. It took nine months to get the bugs out."

In 1998 Bailey joined NERSC, charged with evaluating supercomputer performance and improving numerical algorithms for running scientific calculations. He is coauthor of an extensive library of high-precision software and heads the Performance Evaluation Research Center, a group funded by the Department of Energy's Office of Science to improve high-performance computing.

Bailey's collaborations with mathematicians like Jonathan and Peter Borwein, Richard Crandall, and Helaman Ferguson have yielded important advances. Among his prizes are the Sidney Fernbach Award for high-performance computing and the Mathematical Association of America's Chauvenet Prize and Merten Hasse Prize, awarded to gifted expositors.

Throughout his career, he has never lost his love for pure mathematics.

"To this day I live in two worlds, theoretical math and scientific computing," Bailey says. "I'm trying to marry these two by applying advanced computing to problems in pure mathematics. Experimental mathematics is the outcome."