Close

Google's Quantum Tech Milestone Excites Scientists and Spurs Rivals

Google's quantum supremacy demonstration is a grand physics experiment underwritten by Silicon Valley money

7 min read
Sundar Pichai with one of Google's quantum computers in the Santa Barbara lab.
Sundar Pichai with one of Google's quantum computers in the Santa Barbara lab.
Photo: Google

Quantum computing can already seem like the realm of big business these days, with tech giants such as Google, IBM, and Intel developing quantum tech hardware. But even as rivals reacted to Google’s announcement of having shown quantum computing’s advantage over the most powerful supercomputer, scientists have welcomed the demonstration as providing crucial experimental evidence to back up theoretical research in quantum physics.

Google’s quantum supremacy demonstration showed how its “Sycamore” device could correctly verify sampling results from the quantum equivalent of a random number generator. That niche computational task took just 200 seconds on Google’s quantum computing device, whereas the same test computation would take days or even years on the world’s most powerful supercomputer. Many researchers have hailed the demonstration as an early milestone in the marathon effort to create practical quantum computers. But their most immediate interest is of a scientific nature; intrigue about commercial opportunities will come farther up the road.

“It's not just a milestone on the way to achieving a scalable quantum computer,” says Umesh Vazirani, a computer scientist and co-director of the Berkeley Quantum Computation Center at the University of California in Berkeley. “From my viewpoint, it’s also an experiment in fundamental physics.”

By leveraging the rules of quantum mechanics that determine the behavior of particles at the smallest scales of the universe, scientists and engineers have figured out how to encode information in quantum bits (qubits) that can exist in many different possible states versus modern computing’s binary bits. That could enable quantum computing devices to perform certain computational operations exponentially faster than the “classical computers” in use today. Google’s quantum supremacy experiment seems to have decisively demonstrated such an “exponential speedup” advantage, as detailed in the 23 October 2019 issue of the journal Nature

Of course, companies such as Google, IBM, Intel and others are not developing quantum tech purely in pursuit of scientific discovery. But their interest in quantum computing’s commercial possibilities has helped create what Vazirani describes as an “accelerated timetable” for quantum information science as companies team up with academic labs or poach research talent to advance their own efforts. Universities have similarly stepped up their quantum programs, investing more resources and expanding the number of experimental research groups.

National governments have also begun pouring billions of taxpayer dollars into related programs with an eye toward harnessing quantum computing and quantum science for the sake of both national security and innovation. For countries such as the United States and China, investment in quantum tech can help solidify technological leadership or leapfrog a rival’s traditional technological advantages. Given the competitive global landscape, it’s easy to see why even a family member of a U.S. president would openly celebrate Google’s quantum supremacy demonstration—even if researchers have just begun to scrutinize the results.

“For something like this, we have to evaluate it for some time before the verdict is in,” Vazirani says. “It may be a large or small step, but in the larger scheme the important thing is that it’s a first step.”

How Google Built Its Quantum Tech

The Sycamore processor The Sycamore processor. Photo: Erik Lucero/Google

Google’s quantum supremacy success, albeit preliminary, came from a 54-qubit array called “Sycamore” that uses qubits made of superconducting metal loops kept chilled at subzero temperatures. But creating a practical quantum computer is not just a matter of cramming as many qubits as possible into a device. Researchers also have to ensure that the qubits can remain coupled with each other and maintain their fragile quantum states long enough to perform useful computations. 

The company originally had two teams working in parallel on the 54-qubit Sycamore and a larger 72-qubit device called “Bristlecone.” But a new adjustable coupler that helps keep neighboring qubits connected proved so successful in the Sycamore design that Google’s researchers soon decided to focus on that device. That decision paved the way for the quantum supremacy demonstration with Sycamore that took place in June.

“Because you can build an adjustable coupler with Sycamore, we thought there would be a profound advantage in using that,” said John Martinis, chief scientist for quantum hardware with the Google Quantum AI team based in Santa Barbara, Calif., during a press conference on Wednesday.

During the next year, Google’s team plans to conduct another quantum supremacy experiment using the same Sycamore device architecture—this time, demonstrating error correction on single-qubit and two-qubit errors.

One of the qubits on the Sycamore device was out of commission because of a non-functioning control line that Google discovered during testing in December 2018. But the team realized that they could move ahead with the 53 functioning qubits to carry out the quantum supremacy experiment.

When it came time for the quantum supremacy demonstration, the Sycamore device handily tackled the task of random quantum circuit sampling in just 200 seconds. Google’s estimates suggest that the same test computation would have probably taken 10,000 years on the world’s most powerful computer, the Summit supercomputer developed by IBM that is located at Oak Ridge National Laboratory in Tennessee.

Quantum Computing Rivals React

Much of the quantum computing community has already been abuzz about Google’s quantum supremacy demonstration for weeks because of a premature leak of an early draft of the research paper. And just days before Google’s official announcement, IBM uploaded a paper claiming to show how the Summit supercomputer could pull off the quantum circuit sampling in just 2.5 days rather than thousands of years.

To give the supercomputer a boost in matching the quantum computing device, IBM’s team proposed storing some of the necessary calculation results in secondary disk storage rather than relying only on the primary RAM (random access memory) storage. IBM also published a blog post that described the term “quantum supremacy” as being overhyped, and warned that it gives the misleading impression that quantum computers will someday replace all classical computers.

Despite IBM disputing Google’s quantum supremacy claim, several independent researchers pointed out that the Sycamore device’s performance time of 200 seconds still represents a significant speedup in comparison to the Summit supercomputer handling the same task in 2.5 days. (Google’s researchers also said they welcome IBM’s effort and similar technical challenges as part of the broader research community’s efforts to validate their results.)

“This is still quantum David vs classical Goliath, in the extreme,” says Greg Kuperberg, a mathematician at the University of California, Davis, in comments shared with several journalists. “To answer those 53 qubits, IBM still used entire days of computer time with the world's fastest supercomputer, a 200-petaflop machine with hundreds of thousands of processing cores and trillions of high-speed transistors.”

Intel, another quantum computing rival, issued a more congratulatory press release that praised Google’s quantum supremacy demonstration. But Intel also took the opportunity to suggest that quantum computing architecture based on spin qubits may have an advantage over Google’s superconducting qubit approach when scaling up the size of quantum computing devices. Unlike Google, Intel has split its efforts fairly evenly between exploring the two different quantum computing architectures.

A more oblique response came from a Chinese research team that uploaded its own paper to coincide with Google’s quantum supremacy announcement. The Chinese experiment seemed to demonstrate another possible path toward achieving quantum supremacy by using a “boson sampling” approach based on the interaction and measurement of photon particles.

“Google’s supercomputing effort is doing very well and they’ve put a lot of resources into it. It’s paid off, but doesn’t mean the race is over.”

Any advance among all the competing quantum computing efforts represents good news for the field overall, said Scott Aaronson, director of the Quantum Information Center at the University of Texas at Austin, in a blog post about Google’s demonstration and the various responses. After all, no research group has demonstrated a fully error-corrected and fault-tolerant quantum computer—one of the next big milestones that would also help enable truly practical quantum computing.

Building on Quantum Supremacy

Google already has its eyes on that next milestone because its Sycamore device was designed to incorporate error correction techniques such as surface code. During the next year, Google’s team plans to conduct another quantum supremacy experiment using the same Sycamore device architecture—this time, demonstrating error correction on single-qubit and two-qubit errors. 

“We’re very excited because what we [achieved] here is most of the way [to the starting line for doing] an error correction experiment,” Martinis said.

Another milestone is to demonstrate error correction on a 1,000-qubit device in the next several years. That would still fall short of the estimated 100 million physical qubits required for quantum computing to potentially crack the complex digital codes that safeguard computer security and the Internet.(That’s the long-term expectation for quantum computing’s capabilities that often attracts the most conspiracy theories surrounding such research.) But, says Martinis, “We have time, in our view, to think about this.” 

Google has built several Sycamore quantum processors in its lab and expressed confidence that it has a reliable process for building more quantum computing devices beyond Sycamore. The company also plans to let Google engineers start remotely running computational operations on the quantum computing devices through a cloud computing interface starting in 2020. And it says it will follow that up by offering similar quantum computing access to outside researchers.

There may also be the possibility of leveraging the random quantum circuit used in Google’s recent experiment for commercial applications. The Google team said it plans to investigate Scott Aaronson’s recent hypothesis that a random quantum circuit of this type could enable a certified randomness protocol. Such a protocol that might prove useful for cryptocurrencies and other cryptographic applications.

But other independent researchers remain more cautious about the random quantum circuit’s applications. Kuperberg described himself as “skeptical of real-world applications of a quantum supremacy test” and compared the quantum supremacy milestone to having built an impressive kite on a string. In his view, it would seem strange to begin asking whether the kite could deliver packages or carry human passengers.

“There are many ways to certify randomness in the real world, and I have no thought at all that the world needs this very fancy type of certification,” Kuperberg says. “Still, I could be wrong about that and I respect certified randomness as a topic of discussion.”

For now, many academic researchers anticipate taking the time to dig into Google’s 63-page paper supplement that contains many of the technical details underlying the quantum supremacy experiment. They’re also looking forward to future quantum computing experiments and to getting more hands-on time with quantum computing devices. Those are the steps that could move the overall field away from theoretical work and toward the gathering of empirical proof for quantum computing and quantum physics.

“Google’s supercomputing effort is doing very well and they’ve put a lot of resources into it,” says Aram Harrow, a theoretical physicist at MIT. “It’s paid off, but doesn’t mean the race is over.”

The Conversation (0)

The Future of Deep Learning Is Photonic

Computing with light could slash the energy needs of neural networks

10 min read

This computer rendering depicts the pattern on a photonic chip that the author and his colleagues have devised for performing neural-network calculations using light.

Alexander Sludds
DarkBlue1

Think of the many tasks to which computers are being applied that in the not-so-distant past required human intuition. Computers routinely identify objects in images, transcribe speech, translate between languages, diagnose medical conditions, play complex games, and drive cars.

The technique that has empowered these stunning developments is called deep learning, a term that refers to mathematical models known as artificial neural networks. Deep learning is a subfield of machine learning, a branch of computer science based on fitting complex models to data.

Keep Reading ↓ Show less