[tt] NS: Watchful eye keeps quantum computing on the boil
Premise Checker
<checker at panix.com> on
Sat Nov 17 19:59:28 UTC 2007
Watchful eye keeps quantum computing on the boil
http://technology.newscientist.com/article.ns?id=mg19626295.000&print=true
11 November 2007
Zeeya Merali
THEY say a watched pot never boils. Thanks to a quirk of quantum
mechanics, a similar thing can be said of atoms, which refuse to
decay while you peek at them. A new plan to exploit this effect may
overcome one of the biggest obstacles to building a quantum
computer.
In theory, quantum computers should be far superior to their
conventional counterparts because they will rapidly manipulate huge
amounts of information stored in quantum bits, or qubits. To build
a working quantum computer, the atoms that make up these qubits
must be "entangled" - inextricably linked so that making a change
to one qubit instantaneously alters the quantum state of its
partner. Unfortunately, entanglement is an extremely delicate
property, so qubits quickly disentangle, or decay, when they
interact with their physical surroundings, spoiling attempts to
carry out complicated calculations.
Now Sabrina Maniscalco at the University of Turku in Finland and
her colleagues believe they have found a way to slow down the
quantum disentanglement long enough to perform calculations by
"watching" the qubits' surroundings. The strategy is based on a
weird quantum phenomenon called the quantum Zeno effect, which was
predicted in the 1970s and demonstrated in the 1980s.
Essentially, says Maniscalco, the effect allows you to slow an
unstable atom's decay by measuring various aspects, such as its
energy level. "It's counter-intuitive, but in these experiments you
seem to freeze time."
However, using the effect to protect entangled qubits is far from
straightforward because, according to quantum mechanics, making
observations of quantum-scale particles alters their properties.
That means you risk breaking the very entanglement that you are
trying to shield, she says.
So rather than meddling with the qubits directly, the researchers
realised that they could delay disentanglement simply by observing
the region around them. They analysed how this could be done in the
case of two entangled qubits trapped in an optical cavity (see
Diagram). If these qubits disentangle, they emit a photon.
Curiously, the researchers found that setting up a detector to
watch for this photon at regular intervals would, in theory, delay
the emission of the photon - and by extension disentanglement - for
a few microseconds. Such indirect observation would temporarily
block the environment's ability to disrupt the qubits.
To induce the delay, the builder of a quantum computer would need
to perform a high-speed series of individual photon measurements,
says Maniscalco. "Freezing time" completely would be impossible,
because you would need to perform measurements with an infinitely
small delay between each one, which cannot be done. However,
measurements roughly every 100 nanoseconds - the best today's
technology could achieve - would be enough to win the extra few
microseconds needed for computation, she says.
Barry Garraway, a quantum physicist at the University of Sussex,
UK, is impressed, but notes that today's optical cavities usually
contain defects which may leak energy and disrupt the Zeno effect.
"The team will require the very best cavities, stretching our
current ability to make them to the limit," he says.
The team have submitted their work to Physical Review Letters
(www.arxiv.org/abs/0710.3914).
Related Articles
Quantum untanglement: is spookiness under threat?
http://technology.newscientist.com/article/mg19626281.400
2 November 2007
Quantum threat to our secret data,
http://technology.newscientist.com/article/mg19526216.700
13 September 2007
Trick of light advances quantum computing
http://technology.newscientist.com/article/mg19526145.200
26 July 2007
Quantum states undergo natural selection
http://technology.newscientist.com/article/mg19426104.700
25 July 2007
E-mail me if you have problems getting the referenced articles.
More information about the tt
mailing list