Boffins from America's standards-setting body NIST, the University of Maryland (UMD), and the California Institute of Technology believe there's a way to make quantum computers correct many of their own errors – which would help overcome one of the major design challenges for such devices.
Physicists Simon Lieu, Ron Belyansky, Jeremy Young, Rex Lundgren, Victor Albert, and Alexey Gorshkov described their theory in "Symmetry Breaking and Error Correction in Open Quantum Systems," a paper published Tuesday in the journal Physical Review Letters.
Quantum computers calculate using quantum bits, or qubits, which represent the state of a quantum system. The results of the computations can be derived by measuring characteristics of subatomic particles within that system, such as the polarization of photons or the spin of electrons.
However, qubits are finicky. Managing them is a bit like herding cats – one of which played a starring role in the famous quantum superposition thought experiment devised by physicist Erwin Schrödinger and is memorialized, if it's really dead, by the term "cat state," which describes a superposition of multiple photon states.
"Error correction is especially important in quantum computers, because efficient quantum algorithms make use of large scale quantum interference, which is fragile, i.e. sensitive to imprecision in the computer and to unwanted coupling between the computer and the rest of the world," explained Andrew Martin Steane, professor of physics at the University of Oxford, in a 2006 paper on the subject [PDF].
"This makes large scale quantum computation so difficult as to be practically impossible unless error correction methods are used."
Boffins from China push quantum computing envelope for 'supremacy' in emerging photon fieldREAD MORE
Researchers working on quantum computers expect they will need between 5 (Laflamme) and 9 (Shor) extra qubits, depending upon the type of noise being mitigated, to correct an arbitrary 1-qubit error. That's why scaling current 50-100 qubit quantum computers up past a million qubits, in order to run practical, general-use quantum systems, is going to be non-trivial.
Microsoft's approach to the problem involves topological qubits, which in theory will need less error correction than vanilla qubits because they're less fragile. But topological qubits require materials that haven't yet been developed, in addition to other breakthroughs.
Simon Lieu, a postdoctoral researcher at the Joint Quantum Institute (JQI) and the Joint Center for Quantum Information and Computer Science (QuICS), both of which are research partnerships between UMD and NIST, told The Register in a phone interview that he and his colleagues have proposed using photonic cat qubits in a way that can self-correct certain kinds of errors, though not all of them.
A photonic cat qubit, he explained, is one where you have an even state with, for example, 16, 18, or 20 photons, and another that has an odd state with, for instance, 17, 19, or 21 photons, creating superpositions of photon states. Thus it's like the cat in Schrödinger's experiment – alive and dead, or in this case even and odd – until it's observed.
The miracle here is that by using these photonic qubits, the error correction we're describing does not require additional qubits
Lieu and his colleagues have devised a way to herd these quantum cats that compensates for potential information loss. It involves creating a cavity with two mirrors and bouncing photons back and forth so the light creates an interference pattern that represents quantum information. The pattern should be robust enough to allow incoming photons to replace ones leaking out of the system and sustain the represented system's information.
"The miracle here is that by using these photonic qubits, the error correction we're describing does not require additional qubits," explained Lieu.
Lieu likened the scenario to the movement of a free-standing punching bag that, once hit, wobbles and eventually returns to its equilibrium position. "Our work classifies different noise channels, types of punches, that can correct," he said.
The technique should work for certain types of errors, like bit flips, but not for others. Phase flips, for example, he said, would still require additional redundant qubits to make corrections. But in this yet-to-be realized cat wrangling, less error correction is necessary.
To correct arbitrary 1-qubit errors, the technique will reduce required overhead – the extra qubits needed for fixing things – by a factor of three, he said. Lieu said the technology to demonstrate this theory is within reach and noted that there are several groups studying photonic cat qubits.
"All the ingredients are there, we just need to put them together," he said. ®