Scientists engineered a high-tech computer that can run several
simple programs by harnessing the relationships between tiny particles,
or quantum mechanics, according to a new report Wednesday.
The
work represents a leap in the field of so-called quantum computers,
which store information differently than traditional machines, computing
experts said.
Classical computers use binary bits of information, whose values alternate between 0 and 1, to store data.
Quantum
computers use smaller units—“qubits”—which can simultaneously be 0
and 1. That duality underpins their rapidity and potential supremacy
over conventional devices like laptops and servers.
The new
device combines magnets, lasers and five individual ions, or charged
atoms, trapped in single file. Each atom represents singular qubits. The
team of researchers, at the University of Maryland in College Park,
used the mechanism to run three different basic algorithms, including a
primitive version of the algorithm that underlies encryption technology.
“It’s wonderful work,” said Krysta Svore, a research manager at Microsoft Research,
Microsoft Corp.
’s experimental arm, who wasn’t involved with the study.
Government organizations, academic labs and companies like
Alphabet Inc.
subsidiary Google, Microsoft and
International Business Machines Corp.
believe quantum machines will facilitate advances in materials
science, chemistry, space exploration and artificial intelligence. They
are investing millions on developing so-called universal quantum
computers that could execute any program without tweaking the hardware
on which that software runs.
The new prototype, described in the
journal Nature, is a step in that direction—“a very clear demonstration
of flexible programmability and universality on a single hardware
platform,” said Mark Saffman, a University of Wisconsin-Madison
physicist who wasn’t involved in the work.
The lead researcher, Chris Monroe, is using the technology to launch a startup dubbed ionQ Inc., according to the paper’s first author, graduate student Shantanu Debnath.
Previous attempts have made strides toward, but haven’t quite achieved, such versatility, said Daniel Lidar, the director of the University of Southern California Center for Quantum Information Science & Technology in Los Angeles.
To
run their algorithms, the researchers devised the equivalent of a
quantum-scale laser-light show. They split a single laser beam and
precisely varied the times and intensities at which each mini-ray struck
atomic qubits. Their ensuing vibrations can entangle any pair such that
they behave like a unit. Such particle pas-de-deux give quantum
machines their speed.
Depending on the operation, qubits light up
a certain way; a detector sees that fluorescence and uses it to read
the information the qubits hold.
Simple calculations had an
accuracy of 98%. For more complex ones, like the encryption algorithm,
accuracy plummeted to roughly 60%. In the quantum world, scientists aim
for precisions higher than 99%. At that level, they can start harnessing
quantum mechanics to correct errors that naturally build up in quantum
systems, an important hurdle to making them workable.
They haven’t yet shown they can do that, said Hartmut Neven, a director of engineering at Google specializing in quantum computing. Mr. Debnath says they are tackling the problem.
The
researchers must also prove they can build a reliable system bigger
than five qubits, said Dr. Saffman and others. Such scalability is among
the biggest quandaries in quantum computing. To solve commercially
interesting problems that can’t be cracked in a timely manner with
classical computers, experts estimate somewhere between hundreds to
millions of qubits will be necessary, depending on complexity. Most
experimental systems have handled only a handful.
D-Wave Systems
Inc. has a 1,152-qubit system, though scientists have called into
question the reliability and “quantum-ness.” D-Wave stands by the
quality of its product, which it says is used by companies like Lockheed
Martin and Google, according to CEO Vern Brownell.
There is
disagreement over what flavor of qubits will ultimately dominate:
individual trapped ions in a line, like the ones in the new study, or
superconducting qubits, which are essentially a collection of atoms
acting together as a quantum unit. Commercial companies like Google and
IBM prefer superconducting qubits because they are faster and can be
manufactured similarly to conventional chips. Trapped-ion qubits are
more reliable, but may be difficult to scale. (Microsoft is going after
an altogether different approach.)
Isaac Chuang, a
quantum-information scientist at the Massachusetts Institute of
Technology, thinks it is likely future quantum machines will fuse
several technologies. Such hybrids are already emerging. In June, Google
published a study, also in Nature, describing a prototype that reliably
merged two architectures. That machine was also regarded as a step
toward a multipurpose quantum computer.
Write to Daniela Hernandez at
daniela.hernandez@wsj.com
Source:http://www.wsj.com/articles/scientists-harness-quantum-physics-to-build-a-programmable-computer-1470243605