Schrödinger's cat, the enduring icon of quantum mechanics, has been defied. By making constant but weak measurements of a quantum system, physicists have managed to probe a delicate quantum state without destroying it – the equivalent of taking a peek at Schrodinger's metaphorical cat without killing it. The result should make it easier to handle systems such as quantum computers that exploit the exotic properties of the quantum world.
Quantum objects have the bizarre but useful property of being able to exist in multiple states at once, a phenomenon called superposition. Physicist Erwin Schrödinger illustrated the strange implications of superposition by imagining a cat in a box whose fate depends on a radioactive atom. Because the atom's decay is governed by quantum mechanics – and so only takes a definite value when it is measured – the cat is, somehow, both dead and alive until the box is opened.
Superposition could, in theory, let quantum computers run calculations in parallel by holding information in quantum bits. Unlike ordinary bits, these qubits don't take a value of 1 or 0, but instead exist as a mixture of the two, only settling on a definite value of 1 or 0 when measured.
But this ability to destroy superpositions simply by peeking at them makes systems that depend on this property fragile. That has been a stumbling block for would-be quantum computer scientists, who need quantum states to keep it together long enough to do calculations.
Researchers had suggested it should be possible, in principle, to make measurements that are "gentle" enough not to destroy the superposition. The idea was to measure something less direct than whether the bit is a 1 or a 0 – the equivalent of looking at Schrödinger's cat through blurry glasses. This wouldn't allow you to gain a "strong" piece of information – whether the cat was alive or dead – but you might be able to detect other properties.
Now, R. Vijay of the University of California, Berkeley, and colleagues have managed to create a working equivalent of those blurry glasses. "We only partially open the box," says Vijay.
The team started with a tiny superconducting circuit commonly used as a qubit in quantum computers, and put it in a superposition by cycling its state between 0 and 1 so that it repeatedly hit all the possible mixtures of states.
Next, the team measured the frequency of this oscillation. This is inherently a weaker measurement than determining whether the bit took on the value of 1 or 0 at any point, so the thought was that it might be possible to do this without forcing the qubit to choose between a 1 or a 0. However, it also introduced a complication.
Even though the measurement was gentle enough not to destroy the quantum superposition, the measurement did randomly change the oscillation rate. This couldn't be predicted, but the team was able to make the measurement very quickly, allowing the researchers to inject an equal but opposite change into the system that returned the qubit's frequency to the value it would have had if it had not been measured at all.
This feedback is similar to what happens in a pacemaker: if the system drifts too far from the desired state, whether that's a steady heartbeat or a superposition of ones and zeros, you can nudge it back towards where it should be.
Vijay's team was not the first to come up with this idea of using feedback to probe a quantum system, but the limiting factor in the past had been that measurements weak enough to preserve the system gave signals too small to detect and correct, while bigger measurements introduced noise into the system that was too big to control.
Vijay and colleagues used a new kind of amplifier that let them turn up the signal without contaminating it. They found that their qubit stayed in its oscillating state for the entire run of the experiment. That was only about a hundredth of a second – but, crucially, it meant that the qubit had survived the measuring process.
"This demonstration shows we are almost there, in terms of being able to implement quantum error controls," Vijay says. Such controls could be used to prolong the superpositions of qubits in quantum computing, he says, by automatically nudging qubits that were about to collapse.
The result is not perfect, points out Howard Wiseman of Griffith University in Brisbane, Australia, in an article accompanying the team's paper. "But compared with the no-feedback result of complete unpredictability within several microseconds, the observed stabilization of the qubit's cycling is a big step forward in the feedback control of an individual qubit."