A Canadian company with substantial venture capital backing claims to have built a "quantum computer" that will ultimately solve problems beyond the power of conventional systems - and will demonstrate it over a live link next week.
While most scientists believe a useful system is at least 20 years away, D-Wave (dwavesys.com), based in Burnaby, British Columbia, says it has a breakthrough in a field that already promises revolution. The company says it expects to sell quantum computers next year that can solve knotty problems from protein structure to financial optimisation.
The company has persuaded investors to put $20m (£10m) into its radical new take on quantum computing and promises to show a prototype next Tuesday in its labs.
A quantum computer can, in theory, be vastly more powerful than a conventional one because it uses quantum properties to carry out multiple calculations simultaneously. That's because quantum systems (such as various properties of electrons) simultaneously embody different "states". The best known simile is that of Schrödinger's cat, the thought experiment in which a cat in a box is both alive and dead - until someone opens the box.
In a quantum computer, every quantum bit (or "qubit") is simultaneously both 0 and 1. Put two qubits together, and you have a system whose values are simultaneously every value from 0 to 3. A system with only 300 qubits is in 1090 (one followed by 90 zeros) states simultaneously - more than the number of atoms in the known universe. If (and it's a big "if") you can frame your calculation in the correct way then rather than grinding through each individual step of the calculation (what is 2+2? Add 2 to 0, add 2 to 2, read the result) the quantum computer will move directly to the correct answer. What is 2+2? The quantum state: 4. (Wikipedia's full, and very mathematical, description is here).
But quantum computers literally stop working if you look at them. If any interference, even thermal noise, gets in from the outside world, quantum states "collapse". The cat is alive or dead, the bits are 0 or 1, not both, and the computer loses its magical multiplicity. So far, quantum computers have only been isolated long enough for a few thousand operations - too short to do anything really useful. Some scientists, such as Michael Dyakonov of the University of Montpellier in France, believe thermal noise makes quantum computing as impossible as perpetual motion.
But D-Wave's "Orion" is designed to collapse: it uses a so-called "adiabatic" process, in which the quantum states evolve towards the answer. Noise actually helps this, according to D-Wave's founder, the scientist Geordie Rose. His Orion system is a 16-qubit chip, built with the metal niobium using conventional lithography, and cooled to just above absolute zero.
"The quantum states are like the notes of a chord," says Seth Lloyd, professor of mechanical engineering at MIT, who helped develop adiabatic quantum computing (AQC) theory. "If you could hear quantum states, you would hear a complex chord, changing towards a single note, which is the answer."
The trade-off is that an AQC solves only one problem. It takes any set of inputs and settles into the one state that solves that problem for those inputs. Orion solves a theoretical magnetic field problem, called the two-dimensional Ising model, which would take exponential amounts of time on a normal computer. It can solve more useful problems, such as protein folding and financial optimisation, after a conventional computer translates them into the Ising model.
With 16 qubits, it won't do anything a conventional computer can't, but D-Wave hopes to add qubits quickly if the unproven technology works. "The jury is out," says Lloyd. "It's a long shot, but they've gone about it in the best possible way: they've said 'Let's build it and see'."
Others are less optimistic. "My gut instinct is that I doubt there is a major 'free lunch' here," says Professor Andrew Steane of Oxford University. "That means I doubt that this computing method is substantially easier to achieve [in the present of noise and imperfection] than any other."
No comments:
Post a Comment