THE GREAT Quantum NUMBER CRUNCHER - research into quantum computer networks - includes related articles on quantum physics - Abstract

IF SOMEONE SUCCEEDS in building a quantum computer -- and the odds of that look better every day -- the information age MAY NEVER BE THE SAME.

BEHOLD THE WORLD'S MOST FEEBLE computer network: sitting in a classroom building at Caltech, it connects a grand total of two processors, crosses all of a basement corridor, and transmits a whopping single bit of information. Or would if it were working, which it isn't. "We think it would be nice if it were up and running by the new millennium," says Caltech researcher Jeff Kimble.

Given the pathetic specs, it might seem just a little surprising to hear that Kimble's network is widely considered to be one of the most challenging projects in all of computer science. That's because the one bit of dam his network is designed to transmit won't be an ordinary "1" or "0" of the sort that everyday networks traffic in. It will be a mixture of the two--a so-called quantum bit, or "qubit."

Kimble is trying to build the world's first quantum computer network. In a sense, he's getting a little ahead of himself, since neither he nor anyone else has yet come close to building a practical quantum computer--a computer, that is, that makes calculations on data in the weird multiple-reality state that is the hallmark of quantum mechanics. Still, the benefits of this stunningly radical approach to processing promise to be so great that the young field of quantum computing has been steadily attracting researchers not only from computer science but also from physics, math, and chemistry. Just a few years ago most computer scientists doubted that a quantum computer could ever be built. Now the tide of opinion seems to be turning, and the past year or two have seen some important advances. Neil Gershenfeld, for example, a physicist at MIT, has actually built a simple quantum computer. It can't do much--what it does is pick out one name from a list of four but it does it faster than a conventional computer.

What's the big deal about quantum computing? Imagine you were in a large office building and you had to retrieve a briefcase left on a desk picked at random in one of hundreds of offices. In the same way that you'd have to walk through the building, opening doors one at a time to rind the briefcase, an ordinary computer has to make its way more or less serially through long strings of 1's and 0's until it arrives at an answer. Of course, you could speed up the briefcase hunt by organizing a team, coordinating a floor-by-floor search, and then getting them all back together again to compare results. Ordinary computers can do this sort of thing, too, by breaking up a task and running the components in parallel on several processors. That sort of extra coordinating and communicating, however, exacts a huge toll in overhead.

But what if instead of having to search by yourself or put together and manage a team, you could instantly create as many copies of yourself as there were rooms in the building, all the versions of yourself could simultaneously peek in all the offices, and then--best of all--every copy of yourself would disappear except for the one that found the briefcase?

That's an example of how a quantum computer could work. Quantum computers would exploit the fact that under certain conditions the denizens of the atomic-scale world can exist in multiple realities--atoms and subatomic particles can be simultaneously here and there, fast and slow, pointing up and down. How? Not even physicists agree on that one, but countless experiments over the past seven decades have verified the bizarre phenomenon. By thinking of each of these different atomic states as representing different numbers or other types of data, a group of atoms with all their various combinations of potential states could be used to explore simultaneously all possible answers to a problem. And with some clever jiggling, the combination representing the correct answer could be made to stand out.

CONVENTIONAL COMPUTER CHIPS are getting so jammed with ever tinier components that they may soon hit their physical limits in power and speed; some researchers are hoping that quantum computers might break through those barriers. But although a number of research teams are struggling mightily, even the most optimistic among them don't expect to do more than demonstrate some almost uselessly simple devices within the next three years or so.

Even then, the quantum future is not guaranteed. Any computer--quantum or otherwise--can't do much good unless it can be programmed to perform a practical task. And many researchers have been wondering whether quantum computers will be able to tackle real-world computing problems--or at least run them significantly faster than conventional computers can.

Most applications actually won't lend themselves to quantum computing. That's because the typical computer task, like calculating the orbit of a satellite or rotating a graphic image, requires computer logic that proceeds in serial fashion, each step depending on the results of the preceding one. Quantum computing can't speed up that sort of task. There isn't much advantage in having multiple selves, for example, if instead of looking for a briefcase in a single room, you had to assemble a wristwatch out of parts scattered throughout all the rooms. Whether one person was to do the job or a thousand copies of that person, someone would still have to walk into each room, grab a component of the watch, and then add each piece, one at a time, in the correct order, to the wristwatch-in-progress. The desired result--in this case, a completed wristwatch--requires that every searcher does part of the job; no one's contribution can be discarded.