Quantum computer keeps it simple

August 13/20, 2003
By Eric Smalley, Technology Research News

Page One

Skulls gain virtual faces

Viewer explodes virtual buildings

Tool blazes virtual trails

Quantum computer keeps it simple

News briefs:
Video keys off human heat
Interference boosts biochip
Device simulates food
Motion sensor nears quantum limit
Molecule makes ring rotor
Carbon wires expand nano toolkit








Quantum computers promise to be fantastically fast at solving certain problems like cracking codes and searching large databases, which provides plenty of incentive for overcoming the tremendous obstacles involved in building them.

The basic component of quantum computers, the qubit, is made from an atom or subatomic particle, and quantum computers require that qubits exchange information, which means the interactions between these absurdly tiny objects must be precisely controlled.

Researchers from the University of Oxford and University College London in England have proposed a type of quantum computer that could greatly simplify the way qubits interact.

The scheme allows qubits to be constantly connected to each other instead of repeatedly connected and disconnected, and it allows a computer's qubits to be controlled all at once, said Simon Benjamin, a senior research fellow at the University of Oxford in England. Global control is a fairly unconventional idea that "allows you to send control signals to all the elements of the device at once instead of having to separately wire up each element," he said.

The scheme can be implemented with different types of qubits. A common type uses the spin of an electron. Electrons can be oriented in one of two directions, spin up and spin down. These are analogous to the poles of a kitchen magnet and can represent the 1s and 0s of computer information.

Key to the potential power of quantum computers is a weird trait of quantum particles like electrons. When an electron is isolated from its environment, it enters into superposition, meaning it is in some mix of both spin up and spin down.

Linking two qubits that are in superposition makes it possible for a quantum computer to examine all of the possible solutions to a problem at once. But controlling how two qubits interact is extremely challenging, said Benjamin. Qubits "must be made to talk to each other, and when the operation is over they must be made to stop talking," he said.

In traditional quantum computing schemes that use electron spins, pairs of qubits have a metal electrode between them. When the electrode is negatively charged, it repels the negatively charged electrons that make up the qubits, keeping them separated. But giving the electrode a positive charge draws the electrons toward each other, allowing them to interact by exchanging energy. Allowing the qubits to interact for half the time it takes to completely swap energy is the basis of two-qubit logic gates.

The energy of the two qubits has to be resonant or errors can arise, but off-resonant energy can also be harnessed, said Benjamin. Particles resonate at specific energies in the same way that larger objects vibrate more readily at certain frequencies. Different energies can be more or less resonant with each other much like certain musical notes sounding better together than others. "Something that we were used to thinking of as a source of error could in fact be a means of controlling the computer," he said.

The researchers' proposal replaces the electrode with a third electron. These three electrons are constantly interacting, but they don't always exchange energy. When the middle electron is off resonant, the qubits are blocked from exchanging energy. This way, the interaction "is always on, but we can effectively negate it by ensuring that the energies of neighboring spins are completely incompatible," said Benjamin.

Avoiding electrodes is useful for several reasons. Fabricating qubits with electrodes between them "will require a fantastic degree of control," said Benjamin. "If a particular pair of electrons are too close, then the interaction will be jammed on, and if they are too far away then the interaction will be jammed off," he said.

Electrodes can also knock qubits out of superposition. "Each electrode can act as an [antenna], channeling electromagnetic noise from the room-temperature world right down to the qubits," said Benjamin.

The researchers took their proposal a step further by removing the need to control electrons individually. Every change to the energy of the electrons is applied to the whole device. The researchers divide a string of qubits into two groups, odd and even, with every other qubit in one group. A set of six specific changes to the energies of the electrons covers all of the logic gates required for quantum computing, according to the researchers. Quantum programs would consist of timed sequences of the changes.

The main disadvantage of the researchers' proposal is that it could require as many as two spins per qubit rather than the usual single spin, which would make for a larger device, said Benjamin. "Right now experimentalists are struggling to make even two qubits in solid-state systems," he said.

The researchers' work is valuable because it extends the range of candidates for quantum computing, said Barry Sanders, a professor of quantum information science at the University of Calgary in Canada. The work is "stoking the fires of creativity so that we physicists can dream up other quantum computing realizations that lead to easier control and less experimental complexity," he said.

There is a growing realization that there are many ways to perform qubit operations, said Robert Joynt, a physics professor at the University of Wisconsin at Madison. The Oxford and University College London work is significant for people trying to make a real machine, because it means that the constraints on the hardware are a lot looser than people thought at first, he said. This research "is particularly nice since it gets rid of the usual need to precisely tune two-qubit operations."

The researchers are currently exploring how the method would work in a two- or three-dimensional array of qubits, said Benjamin. "We'd also like to build up a more detailed description of how to implement our scheme with specific technologies like... electron spin," he said.

Researchers generally agree that practical quantum computers are two decades away. It is possible that quantum computers capable of computations that are impossible on conventional computers could be built within ten years, said Benjamin.

Such systems "will be mainly of interest to the scientific community because they will involve using quantum computers to simulate other quantum systems, such as fundamental biological processes," said Benjamin. "These first quantum computers may require an entire lab built around them, and may be treated as a national or international resource for research -- a bit like today's supercomputers or... particle accelerators."

However, it is also possible that quantum computing research could stall if there's not enough experimental progress in the next few years, said Benjamin. "It's possible that quantum computing is an idea born before it's time. Our technology may simply be to crude to achieve it," he said.

Benjamin's research colleague was Sougato Bose. The work appeared in the June 20, 2003 issue of Physical Review Letters. The research was funded by the Royal Society, the Oxford-Cambridge-Hitachi Nanoelectronics at the Quantum Edge project in England, and the National Science Foundation (NSF).

Timeline:   10-20 years
Funding:   Corporate, Government, University
TRN Categories:  Quantum Computing and Communications
Story Type:   News
Related Elements:  Technical paper, "Quantum Computing with an Always-On Heisenberg Interaction," Physical Review Letters, June 20, 2003.


Home     Archive     Glossary     Resources     Research Directory    Letters

Offline Editions and Reports     TRN Store     Feedback      Letters     Comments     About TRN

Find out about TRN Services for Web sites and print publications.

For permission to reprint or republish this article, please email trn@trnmag.com.
© Copyright Technology Research News, LLC 2000-2003. All rights reserved.