The rising sun may have ducked behind a cloud if the quantum computing era hadn't started. A milestone known as quantum supremacy was achieved by the researchers when their quantum computer performed in 200 seconds. Scientists in China have done a computation in a short time. They say that a computer could beat the other one.
A computer scientist at the University of Texas, Austin believes that if they had access to a big enough supercomputer, they could have done the task in a matter of seconds. A mathematician at the University of California, Davis says that the advance takes a bit of the shine off of the claim. Getting to the summit is more exciting.
In an email to the team, the principal scientist said that the edge might not last very long. He said that in their paper they said that classical algorithms would get better. We don't believe that the classical approach can keep up with quantum circuits in the future.
It was designed to be difficult for a conventional computer but easy for a quantum computer, which can manipulate qubits that can be set to 0, 1, or any combination of 0 and 1 at the same time. It is possible for any number from 0 to 2 53 to be written on the 53 qubits.
When the qubits were set to 0, the researchers applied a random but fixed set of gates and read out the qubits. The gates and quantum waves created interference that reinforced some outputs and canceled others. Some should have appeared more likely. The output pattern emerged over millions of trials.
The Oak Ridge National Laboratory has a supercomputer called Summit that has 9216 central processing units and 27,700 faster graphic processing units. IBM researchers countered that if they exploited every bit of hard drive available to the computer it could handle the computation in a few days. Pan Zhang, a statistical physicist at the Institute of Theoretical Physics at the Chinese Academy of Sciences, and his colleauges have shown how to beat Sycamore in a paper.
The problem was referred to as a 3D mathematical array by the group. Each layer contained 53 dots, one for each qubit. A 2D or 4D grid of complex numbers is represented by lines connecting the dots. The simulation was reduced to calculating all the coefficients. The advantage of the tensor network method is that it allows us to use a lot of graphics processing units.
The key insight was that Sycamore's computation was far from precise, so theirs didn't need to be either. The distribution of outputs was calculated using an estimated fidelity of 0.2. The team traded accuracy for speed by cutting some lines. It took just eight lines to make the computation 256 times quicker.
The researchers used an innovation of their own to calculate the output pattern for 1 million strings. The computation took 15 hours to complete. Dominik Hangleiter is a quantum computer scientist at the University of Maryland, College Park. It would take a few dozen seconds for the computation to take 10 billion times faster than the team thought.
There are pitfalls to racing a quantum computer against a conventional one. There needs to be better quantum supremacy experiments. The quantum advantage should be demonstrated by some real-world applications.
Researchers say that the demonstration was not just hype. It required less power than a computer. His team's simulation couldn't have kept up if it had been slightly higher fidelity. Hangleiter says that the experiment did what it was supposed to do.