Quantum computers are surprisingly random – but that’s a good thing

Quantum computers can produce randomness much more easily than previously thought, a surprising discovery that shows we still have much to learn about how the strange realm of quantum physics intersects with computation.

Randomness is a key component of many computational tasks – weather forecasting, for example, involves simulating atmospheric behaviour many times over, each time with a slightly different initial configuration chosen randomly. For quantum computers, arranging their quantum bits, or qubits, in random configurations to produce results is one way that researchers have attempted to demonstrate quantum advantage, where quantum computers can do tasks that are effectively impossible for classical machines.

Setting up these random configurations essentially means shuffling the qubits and the way they link together multiple times, similar to the way you would shuffle a deck of cards. But just as a larger desk of cards is more unwieldy to shuffle than a smaller one, this process was thought to take much longer as you added more qubits to your system. Because more shuffling increases the chances of ruining the qubits’ delicate quantum state, this meant that many useful applications that relied on randomness were thought to be limited to small quantum computers.

Now, Thomas Schuster at the California Institute for Technology and his colleagues have found that these random sequences can be produced with fewer shuffles than we thought, which opens up the possibility of using randomly arranged qubit sequences that would previously have been too complex to implement on larger quantum computers.

To show this, Schuster and his team imagined dividing a collection of qubits into smaller blocks, and then mathematically proved that these blocks could each produce a random sequence. Then, they proved that these smaller qubit blocks could be “glued” together, creating a well-shuffled version of the original set of qubits in a way that you wouldn’t necessarily expect.

“It’s just very surprising, because you can show that similar stuff does not hold for random number generators in classical systems,” says Schuster. For example, shuffling a deck of cards in blocks would be very noticeable, because cards in the top block would always stay near the top. This isn’t true in the quantum case, because the quantum shuffling creates a random superposition of all possible reshuffles.

“This is a much more complicated object than a classical shuffler. For example, the ordering of the top cards is no longer fixed, because we are a superposition of many possible re-orderings, so if I try the classical approach above and measure the location of the top cards after shuffling, I will just receive random outcomes each time, which contain no information about the shuffling whatsoever,” says Schuster. “It’s really a kind of new and intrinsically quantum phenomenon.”

“This kind of random quantum behaviour we all expected to be extremely hard to generate, and here the authors showed that you could do this essentially as efficiently as you can imagine,” says Pieter Claeys at the Max Planck Institute for the Physics of Complex Systems in Germany. “It was a very surprising finding.”

“Random quantum circuits have a plethora of uses as ingredients in quantum algorithms, and even for demonstrating so-called quantum supremacy,” says Ashley Montanaro at the University of Bristol, UK. “The authors already identify numerous applications in quantum information, and I expect that others will follow.” For example, it would make it easier to do the kind of quantum advantage experiments that researchers have previously done, though Montanaro cautions that this doesn’t in turn mean that reaping the practical benefits of such advantage is any closer.