by Paul Preuss | In 1929, dryly remarking upon the utility of the new quantum theory, the eminent physicist Paul Dirac said, "The underlying physical laws are completely known, and the difficulty is only that the exact application of these laws leads to equations much too complicated to be soluble." Forty years later, in the classical age of the mainframe computer, matters had improved—if only a little. As a student, Bill McCurdy, now Berkeley Lab's Associate Director for Computing Sciences, wrote code for research in physical chemistry. "We did our programming with punch cards, and you'd hand these to the machine operator, and later you would come back and find your printout from the line printer on a shelf—if it had run."

After yet another thirty years, scientific computation has been so radically transformed that many call it a third mode of the scientific method, taking its place with experiment and theory. Supercomputers reveal what cannot otherwise be guessed from theoretical equations, what cannot otherwise be found in experimental data; they let us see processes we could never have observed and materials that don't even exist—yet.

"Is the difference just that computers are bigger and faster? Or has there been real change?" asks McCurdy. "True, there's now more computing power in some automobiles than in an entire Apollo spacecraft, but what's happened is not just more speed. In 1968 we might have dreamed of extending statistical mechanics or understanding phase transitions, but we did not imagine graphical user interfaces and instantly changing thousands of lines of code with a few keystrokes. We could not have conceived of watching a running simulation of turbulent flow in an engine, for example, or of stopping it, changing it—maybe moving a piston wall—and starting over with just a few mouse clicks."

One result, says McCurdy, "is that the barrier between computer and researcher is disappearing. Because of that, computer simulation is becoming the domain of the experimentalist. The traditional division between experiment and theory will erode."

At the same time, supercomputers have opened unexplored paths of discovery by allowing scientists to simulate larger and more complex systems than ever before.

"We can now contemplate modeling and understanding the climate of a planet, or the intricate couplings of the millions of chemical and physical machines that make up a living cell," McCurdy says. "We are beginning the mature period of computation," an age in which emergent, unexpected, and astonishingly potent interactions of art and science invite comparison—as the pages that follow hint—to nothing less than the flowering of the Renaissance.


Next: Seeing It Whole


— A Tool for Discovery —
A Tool for Discovery | Seeing It Whole | Climate Modeling | Cosmic Tangle | A Curvier Form of Carbon



Research Review Fall '98 Index | Berkeley Lab