Located between the mountains and the coast of Goleta, California, is a modest office on the side of a building next to the highway. It could belong to any Southern California company; Workers sit in gray cubicles under fluorescent lights and there is a counter for the employees' bicycles and surfboards. But at these desks, physicists and computer scientists are working on developing a computer like you've never seen before. Behind a series of double doors, cylindrical machines hold computer chips at temperatures colder than the vacuum of space.
Here, Google's scientists have struggled to develop a computer processor that can solve a problem that is too difficult for the best supercomputers in the world. Today they announced that they had succeeded: their Sycamore quantum computer was capable of solving a problem in 200 seconds that a supercomputer estimated it would take 1
"One of the criticisms that we have heard many times is that we cooked this invented benchmark problem –  does not make any sense," said Hartmut Neven, a Google engineering director in a silver puff coat reminiscent of a space suit, told journalists today at a press event. "That's why we like to compare it to a Sputnik moment. Sputnik has not done much either. All it did was circle the earth. But it was the beginning of the space age. "
Today, Google gave journalists a first look at the device and showed how it could complete the experiment.
While classical computers use transistors to represent data in zeros and ones Quantum computers represent data using artificial atoms called qubits. Instead of just using the rules of logic, these qubits interact via the strange mathematics of quantum mechanics. T They assume zero or one and, like classical computers, produce long sequences of binary code, but during the calculation they can assume states between zero and one that determine how likely it is for you to take zero or one the final measurement ,
Each qubit consists of a tiny plus-sign loop of superconducting wire. These systems not only allow current to flow without resistance, but it is almost as if the entire unit acts like a single electron. Each plus sign touches four more plus signs in grid form.
The chip (which looks like an ordinary processor chip to the inexperienced viewer) is housed in a housing at the bottom of a structure in the shape of a top. Down wedding cake, kept in a vacuum chamber. The environment gets colder with each stage until the operating temperature is 15 millikelvin. A jumble of wires sends tiny microwave pulses to the qubit, causing it to assume excited states measured by another tiny component attached to the plus sign.
Google's scientists have already designed the quantum superiority experiment in 2016. Use these quantum sensors to randomize. Measure the same circuit a thousand to a million times, and certain sequences of zeroes and ones become more likely than others by an effect called quantum interference. Have a supercomputer simulate the quantum computer and try to make a similar probability distribution of those strings. With each additional qubit (and with each additional operation), it becomes much harder for the supercomputer to keep up. The scientists of Google probably felt that they had hit the supercomputer with 53 of the 54 qubits from Sycamore (one of which did not work). To confirm the correctness of the response, the complexity of the circuit must be slightly reduced, checked by the supercomputer and then extrapolated.
With the help of physicist Scott Aaronson from the University of Texas, they even found use for this quantum superiority experiment. It spins random bits, and randomness is important in areas such as cryptography and lottery. But what if it's not really coincidental – what if someone can secretly guess the supposed random number? With this test Google can check for you if a normal computer did not invent these random numbers.
Despite the technological advances, the computer is error-prone. Any interaction with the outside world can cause the qubit to give the wrong values. However, the experiment has shown that the number of errors increases predictably as more qubits are added. The layout, especially the grid of the plus signs, should be compatible with a future in which these problems can be anticipated and circumvented.
"We have shown that we understand these mistakes," said Google scientist Marissa Giustina. "This is a crucial part of the breakthrough in engineering and physics." (For your information, Giustina was the only scientist in the room.)
I also had to program the computer. Similar to IBM's Q Experience you use a regular computer interface to pull pulse-generating qubit-value-changing operations on each qubit, for example, B. music notes on the staff. Oscilloscopes showed the shape of the impulses that I sent to the qubits. I watched how the odds of each qubit delivering a zero or a one changed with each additional operation.
Many scientists have already criticized that classical computers can actually perform the superiority experiment in shorter time or that the right classical algorithm has not yet been found. Neven responded to IBM's claim that a classic supercomputer would take 2.5 days and not 10,000 years to perform the superiority experiment:
"Since we published the quantum supremacy proposal, there has been a steady stream of improvements over it classic page, which has become a benchmark of classic supercomputers, "he said. He explained that researchers at NASA, the Oak Ridge National Lab and other countries are working to improve on classic computational algorithms so that the Google device can compete with the latest supercomputers.
On a scientific level, Google has so far demonstrated a large, nested quantum system more complex than previously shown. In the field of computing, we have broken new ground: quantum computers are now devices that may be able to do something that a classic computer can not do .
Giustina said, "We've reached a new computational scope that no other tool can match."