Custom-assemble five fluorine atoms with a few other atoms, and the product is a molecule. It’s also a computer. The problem that it’s solved is a simple one, but the exercise provides experimental evidence that a quantum computer can handle certain mathematical problems more efficiently than can a conventional computer.

Though a practical quantum computer still may be decades away, “this result gives us a great deal of confidence in understanding how quantum computing can evolve into a future technology,” says Isaac L. Chuang of the IBM Almaden Research Center in San Jose, Calif.

He and his collaborators announced the feat last week at the Hot Chips 2000 conference at Stanford University.

A quantum computer exploits the quantum mechanical nature of tiny particles, such as electrons or atomic nuclei, to encode information as quantum bits, or qubits. Whereas an ordinary bit has at any time a value of either 0 or 1, a qubit can also take on both values at once. Because a quantum computer can act on these multiple states simultaneously, it’s potentially many times as powerful as a conventional computer (SN: 1/14/95, p. 30).

For their quantum computer, Chuang’s team designed a molecule in which the nuclei of five fluorine atoms interact with each other to produce a five-qubit system (SN: 1/18/97, p. 37). The researchers report that they used radio pulses to “program” the nuclei into specific quantum states. Nuclear magnetic resonance (NMR) instruments could then detect the results.

The investigators tackled a permutation problem. Roughly speaking, computing the order of a permutation is akin to finding the shortest path through a hidden maze of rooms, each with precisely one exit and one entrance, connected by one-way passages. The goal is to get back to the starting point most quickly.

Last year, Richard Cleve of the University of Calgary in Alberta showed theoretically how a quantum computer could solve a particular version of this problem in fewer steps than a conventional computer.

Chuang and his coworkers tackled a four-room version of the permutation problem. They used radio pulses to program the molecules, which were dissolved in a liquid, so that two qubits encoded the starting room, and three qubits represented, in effect, the eight possible systems of one-way passageways.

Then, the researchers obtained the molecules’ NMR spectrum, and its pattern of lines gave the correct answer in each test. In essence, the result emerged in one computational step, whereas a conventional computer would have required more steps.

“We anticipate more-complex algorithms, involving a few extra qubits, may be possible using our current approach,” comments team member Lieven M.K. Vandersypen of Stanford.

However, “such implementations do not provide sufficient information about critical issues for the future of quantum computation, such as how much control we actually can achieve over a quantum computer,” says Emanuel Knill of the Los Alamos (N.M.) National Laboratory.

Earlier this year, Knill and his coworkers put together a seven-qubit quantum computer made up of six hydrogen and four carbon atoms. Although the researchers didn’t perform a complete computation at that time, they demonstrated a reliable, efficient procedure for setting up qubits for a computation.