Ordinary Computers Could Beat Google’s Quantum Computer After All

Superfast algorithm place crimp in 2019 claim that Google’s machine had achieved “quantum supremacy.”

If the quantum computing era dawned three years earlier, its rising sun might have ducked behind a cloud. In 2019, Google scientists claimed they had passed a milestone called as quantum supremacy when their quantum computer Sycamore performed in 200 secs an abstruse calculation they said would tie up a supercomputer for 10,000 years. Currently, researchers in China have done the computation in a few hrs with ordinary processors. A supercomputer, they state, could beat Sycamore outright.

“I believe they are right that if they had accessibility to a huge enough supercomputer, they should have simulated the … task in a matter of secs,” states Scott Aaronson, a computer scientist at the College of Texas Austin. The development takes a bit of the shine off Google’s claim, states Greg Kuperberg, a mathematician at the University of Golden State, Davis. “Getting three-hundred feet from the summit is less exciting than getting there.”

Nevertheless, the promise of quantum computing continues to be undimmed, Kuperberg and others say. Furthermore, Sergio Boixo, principal scientist for Google Quantum AI, said in an email that the Google group knew its edge could not hold for very long. “In our 2019 paper, we stated that classical formulas would certainly boost,” he said. However, “we do not think this classical approach could keep up with quantum circuits in 2022 and beyond.”

The problem

The “issue” Sycamore solved was designed to be challenging for a conventional computer but as simple as possible for a quantum computer that manipulates qubits that could be set to 0, 1, or– thanks to quantum mechanics– any combination of 0 and 1 at the same time. h Sycamore’s 53 qubits, tiny resonating electrical circuits made from superconducting metal, can encode any number from 0 to 253 ( about 9 quadrillions)– or even all of them at once.

With all the qubits set to zero, Google scientists applied to single qubits and pairs a random but fixed set of logical operations, or gates, over twenty cycles, then read out the qubits. Crudely speaking, quantum waves representing all possible outputs sloshed among the qubits, and the gates produced interference that reinforced some outputs and canceled others. So some would have appeared with a higher probability than others. Over millions of tests, a spiky output pattern emerged.

The Google researchers stated that simulating those interference effects would also overwhelm Summit, a supercomputer at Oak Ridge National Lab, which has 9216 central processing units and 27,648 faster graphic processing units (GPUs). Researchers with IBM that developed Summit quickly countered that if they exploited every bit of hard drive available to the computer, it might handle the computation in a few days. Now, Pan Zhang, an analytical physicist at the Institute of Theoretical Physics at the Chinese Academy of Sciences, and colleagues have illustrated how to beat Sycamore in a paper in press at Physical Review Letters.

Following others, Zhang and colleagues recast the issue as a 3D mathematical array called a tensor network. It consisted of twenty layers, 1 for each cycle of gates, with each layer comprising fifty-three dots, one for each qubit. Lines linked the dots to represent the gates, with each gate encoded in a tensor– a 2D or 4D grid of complex numbers. Running the simulation then lowered to, essentially, multiplying all the tensors. “The advantage of the tensor network technique is we can use many GPUs to do the calculations in parallel,” Zhang says.

Zhang and colleagues also relied on a vital insight: Sycamore’s computation was far from exact, so theirs did not require to be either. Sycamore calculated the distribution of outputs with an estimated integrity of 0.2%– just sufficient to distinguish the fingerprintlike spikiness from the noise in the circuitry. So Zhang’s group traded accuracy for speed by reducing some lines in its network and eliminating the corresponding gates. Losing simply eight lines made the computation 256 times faster while maintaining a integrity of 0.37%.

The calculation

The scientists calculated the output pattern for one million of the nine quadrillion possible number strings, relying on an innovation of their own to have a truly random, representative set. The calculation took 15 hours on 512 GPUs and yielded the telltale spiky output. “It is fair to state that the Google experiment has been simulated on a conventional computer,” states Dominik Hangleiter, a quantum computer scientist at the College of Maryland, University Park. The calculation would take a few dozen seconds on a supercomputer, Zhang says– 10 billion times faster than the Google group estimated.

Scientists say that the advance underscores the pitfalls of racing a quantum computer against a conventional one. “There is an urgent need for better quantum supremacy experiments,” Aaronson says. Zhang recommends a more practical approach: “We should discover some real-world applications to demonstrate the quantum advantage.”

Still, the Google illustration was not just hype, researchers state. Sycamore needed far fewer operations and less power than a supercomputer, Zhang notes. Moreover, if Sycamore had slightly higher integrity, he says, his group’s simulation could not have kept up. As Hangleiter places it, “The Google experiment did what it was meant to do, begin this race.”

Read the original article on Science.