PORTLAND, Ore.—Quantum computers are being pursued by all the major research labs worldwide, including even new-comers like Google. However, IBM claims its 30 years of experience in quantum computing research shows Google is all wrong in its "linear" design, because IBM's "square tiled" design can solve both of the most important problems in quantum computing plus can scale to any size needed in the future.
IBM's search for the right architecture for quantum computers began in 1981 when it attended Nobel Prize winner Richard Feynman's first workshop on the "physics of information" where he laid out the concept of a quantum computer.
Over the subsequent 34 years IBM has developed Feynman's theory, invented its own architecture and conducted experiments that today have led to the devices that can reliably scale to any number of qubits—extending Moore's law indefinitely since it would on take only 50 qubits to outperform the fastest supercomputer on the Top500.org supercomputer list.
"We are reporting on a four-qubit system today, and already experimenting with an eight-qubit system," Jerry Chow, manager of experimental quantum computing group at IBM Research Center (Yorktown Heights, N.Y.) told EE Times. "But unlike other designs, the series of problems we have just solved can scale to a superconducting quantum computer with any number of qubits."
The current prototype (see photos) is supercooled to 15 milliKelvin, using a commercially available super-refrigerator, and has solved the two most important problems in quantum computing today, according to IBM, that is—simultaneously correcting for bit-flip and phase-flip errors as well as being completely scalable to any size.
IBM's four superconducting quantum bits use a square lattice to detect both types of quantum errors--bit-errors and phase-errors--for the first time.
(Credit: IBM Research)
"On the march toward quantum computers error correction is the most important problem, because qubits are not as robust as normal computer bits," Chow told us. "Qubits are very fragile and can be spoiled by all sorts of noise—in the environment and in the rest of the system."
There are two main types of errors that must be corrected in quantum computers—bit-errors (flipping a 1 to a 0 or a 0 to a 1) and phase-flip errors (which cause signals to subtract from each other instead of add). Unfortunately, it is very difficult to solve both these errors at once. To do so, IBM had to go to a square architecture—instead of a linear-array architecture like Google's, which according to IBM can only solve one of the problems or the other but not both. As a result, IBM's square array of four qubits is quadruple redundant, but it makes an error-free scalable quantum computer possible.
IBM's square layout is the best configuration to add more qubits and scale to larger systems.
(Credit: IBM Research)
"You need nearly perfect qubits to do meaningful calculations with a quantum computer, so you have to dedicate some qubits to error correction," Chow told us. "In fact, we are working on an eight-qubit architecture now to correct other possibilities of error in a single qubit and believe that it may take 13-to-17 qubits or maybe even more to make a single qubit perfectly reliable."
The reason is that qubits do not just hold a single 0 or 1 like regular bits, but hold a superposition of values, "they are part 0 and part 1, which is why it is so hard to perfectly preserve then during quantum calculations," Chow told us.
In the current four-qubit architecture, two qubits hold the value to be preserved, while the other two qubits tell you whether there have been any bit-flip or phase-flip errors, respectively.
IBM's Jerry Chow inspecting a qubit in the quantum computing lab at IBM's T.J. Watson Research Center.
(Credit: Jon Simon/Feature Photo Service for IBM)
Next IBM is shooting for the Holy Grail of quantum computing—the perfect incorruptible qubit—that could form the basis of a real quantum computer scalable to any size. To achieve that goal, IBM plans to extend the current architecture by adding more qubits—in an 8-bit surface-code lattice for now—that will offer even better protection and correction of quantum errors on the road to the perfection they need to start marketing real quantum computers.
"It may take 8, 13, 17 or even 49 qubits in a scalable lattice to realize the Holy Grail of perfection, and the architecture may have to change to rectangular or hexagonal or some other symmetrical configuration, but we now believe it is definitely an achievable goal."
IBM's quantum computing architecture uses scalable building blocks, detects and corrects quantum errors, and will be able solve problems conventional computers cannot.
Once a quantum computer is perfected it will not only be able to crack any encryption code today and make new uncrackable codes, but will also allow researchers to understand physical processes that are impossible to simulate of traditional computers For instance all the molecular interactions that make common processes work—allowing designers to create materials impossible to imagine today, derive insights from unstructured Big Data in an instant, and truly understand useful chemical reactions that today were only found by trial-and-error.
"Quantum computers will spawn a new era of innovation across all industries," Chow concluded.
IBM's work was partially funded the U.S. IARPA (Intelligence Advanced Research Projects Activity) multi-qubit-coherent-operations program.
— R. Colin Johnson, Advanced Technology Editor, EE Times
Join over 2,000 technical professionals and embedded systems hardware, software, and firmware developers at ESC Boston May 6-7, 2015 and learn about the latest techniques and tips for reducing time, cost, and complexity in the embedded development process.
Passes for the ESC Boston 2015 Technical Conference are available at the conference’s official site with discounted advance pricing until May 1, 2015. The Embedded Systems Conference and EE Times are
owned by UBM Canon.