For all that computers can do, they’re useless when they’re not running. That’s not the case for quantum computers. Using a rudimentary quantum computer, a team of physicists consistently determined which of two operations the computer would have performed, even though it didn’t actually perform them.
The experiment, reported in the Aug. 21 Physical Review Letters, delivered an unprecedented 85 percent success rate for what’s called counterfactual computation: exploiting quantum mechanics to glean the information a computer would have provided had it run.
Future counterfactual schemes could enable the safe imaging of objects that are vulnerable to light, the researchers say. But some physicists aren’t sold on the result, mainly because it’s difficult to define the concept of a computer running or not running in the wishy-washy quantum world. At the very least, the study “contributes to the debate of how we understand quantum mechanics and how it compares with our normal ideas of logic,” says Matthew Rakher, a condensed matter experimentalist at HRL Laboratories in Malibu, Calif.
The realm of the tiny is both bizarre and exciting because of its uncertainty. A photon, for example, can seemingly take two paths simultaneously to reach a destination, a phenomenon known as superposition. In 1993, two theoretical physicists showed that if a photon-triggered bomb were placed in one of the paths, a photon passing through would occasionally reveal the presence of the bomb without setting it off. Five years later, theoretical physicist Richard Jozsa, now at the University of Cambridge in England, replaced the bomb in the thought experiment with a quantum computer. He showed that if the computer was turned on and programmed to perform a calculation, then an experimenter could learn the result without the computer doing anything. But the success rate was limited to 50 percent, no better than guessing.
Fei Kong, a quantum physicist at the University of Science and Technology of China in Hefei, and colleagues tried to beat that limit. The researchers built a simple quantum computer and programmed it to run either one of two operations. Each operation assigned a different value to the spin of a nucleus.
Rather than let the computer do its thing, the researchers set up a detour to probe which operation the computer would have performed. They put the nucleus in a superposition of multiple spins and then repeatedly manipulated the spin using radio pulses. A trick known as the quantum Zeno effect (SN: 11/20/10, p. 20) prevented the computer from running. After tweaking the nucleus 17 times, the researchers made a definitive measurement of the spin. If the spin had a value of 1, then they knew the computer would have run the first operation; if it was 2, then the computer would have performed the second operation. The scheme worked 85 percent of the time.
Although a team including Rakher experimentally demonstrated counterfactual computation in 2006 (SN: 2/25/06, p. 117), this setup far surpasses the theorized 50 percent limit, meaning it was better than simply guessing between the two operations in advance. “That’s where this magic turns out to be very exciting,” says study coauthor Liang Jiang, an applied physicist at Yale University. If single photons and nuclear spins can extract information unobtrusively, Jiang says, then perhaps scientists could employ just a handful of photons to deliver a complete image of a light-sensitive protein under the microscope.
Jozsa and Rakher are more cautious, noting that physicists are still deciphering the magic of counterfactual computation. “This paper adds to the discussion,” Rakher says. “Questions like, ‘Did the computer run or not run?’ don’t mean the same thing [in quantum mechanics] as they do in everyday life.”