Weltnachrichten – AU – Light-based quantum computer outperforms the fastest classic supercomputers

Setting up lasers and mirrors « solved » a problem that is far too complicated for even the largest traditional computer system

For the first time, a quantum computer made of photons – light particles – has outperformed even the fastest classic supercomputers.

Physicists led by Chao-Yang Lu and Jian-Wei Pan from the University of Science and Technology in China (USTC) in Shanghai used their quantum computer called Jiŭzhāng to perform a technique called Gaussian boson sampling. The result, published in the journal Science, was 76 detected photons – far beyond the previous recording of five detected photons and the capabilities of classic supercomputers.

In contrast to a conventional computer made of silicon processors, Jiŭzhāng is a sophisticated tabletop setup consisting of lasers, mirrors, prisms and photon detectors. It’s not a universal computer that could one day send emails or store files, but it shows the potential of the quantum computer.

Last year, Google hit the headlines when its Sycamore quantum computer took about three minutes to do what a supercomputer took three days (or 10. 000 years, depending on your estimation method). . In their work, the USTC team estimated that the Sunway TaihuLight, the third most powerful supercomputer in the world, would be an amazing 2. 5 billion years to do the same calculation as Jiŭzhāng.

This is only the second demonstration of the quantum primacy. This term describes the point at which a quantum computer exponentially outperforms a classical one and effectively does what would otherwise be essentially computationally impossible. It’s not just a proof of the principle; There is also some evidence that Gaussian boson sampling could have practical applications, such as solving specific problems in quantum chemistry and mathematics. In a broader sense, the ability to control photons as qubits is a prerequisite for any large quantum internet. (A qubit is a quantum bit, analogous to the bits that are used to represent information in classical arithmetic. )

« It wasn’t obvious that this was going to happen, » says Scott Aaronson, a theoretical computer scientist at the University of Texas at Austin, who first outlined the basics of boson sampling in 2011 with then student Alex Arkhipov. For many years, boson sampling experiments got stuck with around three to five detected photons, which, according to Aaronson, is “a hell of a long way” from the quantum prime. « It’s hard to make it bigger, » he says. “Hats off to you. ”

In recent years, quantum computing has gone from being obscure to being a billion dollar company known for its potential impact on national security, the global economy, and the fundamentals of physics and computer science. In 2019 the U. . S.. . The National Quantum Initiative Act was signed into law to invest more than $ 1. 2 billion in quantum technology in the next 10 years. The field has also generated a lot of hype, with unrealistic schedules and bombastic claims about quantum computers that are making classical computers completely redundant.

This recent demonstration of the potential of quantum computers by the USTC group is crucial as it is very different from Google’s approach. Sycamore uses superconducting metal loops to form qubits; In Jiŭzhāng the photons are themselves the qubits. The independent confirmation that quantum computing principles can also lead to primacy on completely different hardware gives us the certainty that useful quantum simulators and a fault-tolerant quantum computer will become feasible in the long term, says Lu.

Why do quantum computers have enormous potential? Consider the famous double slit experiment in which a photon is fired at a barrier with two slits, A and B.. The photon does not go through A or through B.. . Instead, the double slit experiment shows that the photon exists in a “superposition” or combination of possibilities to have passed through both A and B.. In theory, the use of quantum properties like overlay enables quantum computers to achieve exponential accelerations over their classical counterparts when applied to certain specific problems.

In the early 2000s, physicists were interested in using the quantum properties of photons to create a quantum computer. This is in part because photons can act as qubits at room temperature, eliminating the costly task of cooling your own system to a few Kelvin (approximately -455 degrees Fahrenheit) as with other quantum computing schemes. However, it quickly became clear that building a universal photonic quantum computer was not possible. Millions of lasers and other optical devices would be required to even build a functioning quantum computer. As a result, the quantum prime seemed unattainable with photons.

In 2011, Aaronson and Arkhipov introduced the concept of boson sampling and showed how this can be done with a limited quantum computer consisting of just a few lasers, mirrors, prisms and photon detectors. Suddenly there was a way for photonic quantum computers to show that they could be faster than classical computers.

The setup for sampling bosons is analogous to the toy called the bean machine. It’s just a board with pegs covered with a clear sheet of glass. Balls are dropped into the rows of pins from above. On the way down, they jump off the pins and off each other until they land in slots at the bottom. The simulation of the distribution of balls in slots is relatively simple on a classic computer.

Photons are used instead of spheres in boson scanning, and the pins are replaced by mirrors and prisms. Photons from the lasers bounce off mirrors and through prisms until they land in a « slit » to be detected. In contrast to the classical spheres, the quantum properties of the photon lead to an exponentially increasing number of possible distributions.

The problem that boson sampling solves is essentially « What is the distribution of photons like? » Boson sampling is a quantum computer that dissolves itself by distributing photons. In the meantime, a classic computer has to determine the distribution of photons by calculating what is called the « permanent » of a matrix. For an input of two photons, this is just a short calculation using a two-by-two array. However, as the number of photonic inputs and detectors increases, the size of the array increases, which increases the computational difficulty of the problem exponentially.

Last year, the USTC group demonstrated a boson scan with 14 detected photons – difficult to calculate for a laptop but easy for a supercomputer. To scale to the quantum prime, they used a slightly different protocol, Gaussian boson sampling.

According to Christine Silberhorn, a quantum optics expert at the University of Paderborn in Germany and one of the co-developers of Gaussian boson sampling, the technique was developed to avoid the unreliable single photons used in Aaronson and Arkhipov’s « vanilla » boson are sampling.

« I really wanted to make it practical, » she says. “It’s a scheme that is specific to what you can do experimentally. ”

Even so, she admits that the USTC setup is shockingly complicated. Jiŭzhāng begins with a laser that is split in such a way that it hits 25 crystals of potassium titanyl phosphate. After each crystal is hit, it reliably spits out two photons in opposite directions. The photons are then sent through 100 entrances where they pass through a track made up of 300 prisms and 75 mirrors. Eventually, the photons land in 100 slits where they are detected. On average over 200 second runs, the USTC group detected about 43 photons per run. In one run they observed 76 photons – more than enough to justify their claim of the quantum primacy.

It is difficult to estimate how much time it will take a supercomputer to solve a distribution with 76 detected photons – in large part because it is not exactly feasible to spend 2. A supercomputer has been running for 5 billion years to check it directly. Instead, the researchers extrapolate the time that is required for the classic calculation of a lower number of detected photons. Dissolving after 50 photons would take two days at best, which would take a supercomputer two days, which is far slower than the 200-second run time of Ji Sekundenzhāng.

Boson sampling schemes have suffered for years with low photon counts because they are incredibly difficult to scale. In order to maintain the sensitive quantum arrangement, the photons do not have to remain distinguishable. Imagine a horse race where all horses have to be released from the starting gate and finished at exactly the same time. Unfortunately, photons are much more unreliable than horses.

If photons travel a 22-meter path in Jiŭzhāng, their positions cannot differ by more than 25 nanometers. That’s the equivalent of 100 horses driving 100 kilometers and crossing the finish line with a hair’s breadth, says Lu.

The USTC quantum computer takes its name Jiŭzhāng from Jiŭzhāng Suànshù or « The nine chapters on the mathematical art », an ancient Chinese text with an effect comparable to Euclid’s elements.

Quantum computers also have many twists ahead of them. According to Lu, it is not a matter of course to outperform classical computers, but there will be constant competition to see whether classical algorithms and computers can catch up or whether quantum computers retain the priority they have taken.

Things are unlikely to be static. At the end of October, researchers at the Canadian quantum computer start-up Xanadu found an algorithm that reduced the classical simulation time for some boson sampling experiments by the square. In other words, if 50 detected photons were previously enough for the quantum prime, you would need 100 now.

For theoretical computer scientists like Aaronson, the result is exciting, as it provides further evidence against the extended Church-Turing thesis, according to which any physical system can be efficiently simulated on a classic computer.

« If we think of the universe as a computer in the broadest sense, what kind of computer is it? » Aaronson says. “Is it a classic computer? Or is it a quantum computer? “

So far, like the computers we want to make, the universe seems stubborn.

Daniel Garisto is a freelance science journalist reporting on advances in physics and other natural sciences. His writing has been published in Nature News, Science News, Undark, and elsewhere.

Scientific American Space & Physics is a summary of the most important stories about the universe and beyond

Quantum computers, quantum mechanics, computers, quantum superiority, Pan Jianwei, China, boson scanning