Theoretical discovery opens the door to building quantum computers with significantly reduced resources
Quantum computers of the future may be closer to reality thanks to new research from Caltech and Oratomic, a Caltech-linked start-up company. Theorists and experimentalists teamed up to develop a new approach for reducing the errors that riddle today's rudimentary quantum computers. Whereas these machines were previously thought to require millions of qubits to work properly (qubits being the quantum equivalent to 1's and 0's in classical computers), the new results indicate that a fully realized quantum computer could be built with as few as 10,000 to 20,000 qubits. The need for fewer qubits means that quantum computers could, in theory, be operational by the end of the decade.
The team proposes a new quantum error-correction architecture that is significantly more efficient than previous approaches. Quantum error correction is a process by which extra, redundant qubits are introduced to correct errors, or faults, enabling the ultimate goal in the field: fault-tolerant quantum computing.
The results exploit special properties of quantum computing platforms built out of neutral atoms, which serve as the qubits. Alternative platforms in development include superconducting circuits and trapped ions (ions are charged whereas neutral atoms are not). In a neutral atom system, laser beams known as optical tweezers are used to arrange atoms into qubit arrays. Manuel Endres, a professor of physics at Caltech, and his colleagues recently created the largest qubit array ever assembled, containing 6,100 trapped neutral atoms.
"Unlike other quantum computing platforms, neutral atom qubits can be directly connected over large distances," Endres says. "Optical tweezers can shuttle one atom to the other end of the array and directly entangle it with another atom."
This dynamic ability to move atoms is key to the researchers' ultra-efficient error-correction scheme, which they describe in a new report posted online. The study's co-first authors are Madelyn Cain, lead theoretical scientist at Oratomic, and Qian Xu, the Sherman Fairchild Postdoctoral Fellow at Caltech and now a research scientist at Oratomic. The senior authors are Endres; John Preskill, the Richard P. Feynman Professor of Theoretical Physics and the Allen V. C. Davis and Lenabelle Davis Leadership Chair of the Institute for Quantum Information and Matter (IQIM) at Caltech; Hsin-Yuan (Robert) Huang, assistant professor of theoretical physics and a William H. Hurt Scholar at Caltech who is currently on leave while serving as CTO of Oratomic; and Dolev Bluvstein, a visiting associate in physics at Caltech and CEO of Oratomic. Other authors include Oratomic's Robbie King and Lewis Picard, and Harry Levine of Oratomic and UC Berkeley.
The theoretical results involved innovating new architectures to greatly reduce the overhead of error correction.
"We've spent years learning how to leverage this remarkable ability of neutral atom computers to rearrange qubits dynamically," Cain says. "Our results now make useful quantum computation with neutral atoms appear within reach by reducing qubit counts by up to two orders of magnitude."
Xu adds, "For decades, qubit count has been viewed as the main obstacle to fault-tolerant quantum computing. I hope our work helps shift that perspective."
The report stresses that the team's findings mean that fault-tolerant quantum computers could be on the horizon. Previously, experts in quantum computing thought that such an accurate machine would take another 10 or even 20 years to build.
"I've been working on fault-tolerant quantum computing longer than some of my coauthors have been alive," Preskill says. "Now at last we're getting close."
Huang says, "I always considered theoretical research on the usefulness of large-scale quantum algorithms to only be of interest in the distant future. Our new study made me realize they might come true in the next few years."
Importantly, the accelerated timeline indicates that the security of digital communications—which includes everyday financial transactions and many other forms of private messaging—could be vulnerable to data breaches sooner than expected. Today's computers protect data using encryption schemes, such as RSA (Rivest-Shamir-Adleman) and ECC (elliptic curve cryptography). In these classical schemes, data are encrypted using hard mathematical problems that are infeasible for current computers to solve.
Quantum computers will have the ability to break both encryption schemes thanks to an algorithm developed by Peter Shor (BS '81) in 1994, now a professor of applied mathematics at MIT. To protect against this scenario, organizations around the world have been migrating to new encryption schemes capable of resisting quantum computer attacks. The authors emphasize that the rapid progress toward practical quantum computing underscores the importance of safe and timely migration to these new cryptographic standards.
Quantum computers are based on the laws of quantum physics—laws that govern the behaviors of subatomic particles such as electrons and photons. In the quantum realm, particles exhibit properties foreign to the classical realm we live in, including superposition, in which a particle exists in two places at once, and entanglement, in which particles remain intimately connected even after being separated by long distances.
Because nature is quantum at its most fundamental level, quantum computers are posited to have the power to unlock scientific mysteries, including quantum gravity and room-temperature superconductivity, as well as other problems in chemistry, medicine, sustainability, machine learning, and more.
The qubits at the hearts of these machines become both superimposed and entangled; however, these quantum states are delicate and prone to collapse. When this happens during a calculation, the information stored by the qubits is damaged, leading to errors. To address this problem, researchers have come up with error-correcting methods, like those used by classical computers, in which redundant qubits are used to check for the presence of errors. But error correction is trickier with quantum computers: Today's most common protocols often require about 1,000 or so physical qubits to act together as one single "logical" qubit—the qubit carrying out a desired calculation.
A working quantum computer will require at least 1,000 logical qubits in total, but if each logical qubit is made of 1,000 physical qubits, the whole computer would require 1 million qubits. Scaling up a quantum machine to that large of a size would be extremely challenging, so researchers have been working on ways to scale down the number of physical qubits required for each logical qubit.
The new study describes how this can be achieved using neutral-atom arrays. In other error-correction schemes, such as those using so-called surface codes, qubits arranged in two dimensions are limited to connections to their direct neighbors. In neutral atom arrays, the qubits can be connected to many other qubits that are far away, enabling what scientists call high-rate codes. In such protocols, each physical qubit can participate in many logical qubits instead of just one.
The new scheme means that each logical qubit could be encoded with as few as five or so physical qubits, as opposed to the 1,000 needed with other techniques.
"It's actually very surprising how well this works. It's what we call ultra-efficient error correction," Endres says.
While the results are theoretical, neutral atom quantum systems have rapidly advanced experimentally in recent years, with researchers demonstrating early error-corrected operations and arrays exceeding 6,000 atomic qubits. Significant engineering challenges remain to combine these capabilities into scalable systems, but the new research suggests that neutral atom architectures could ultimately run quantum algorithms powerful enough to impact modern encryption. More broadly, as these systems scale to thousands of logical qubits performing millions of operations, they are expected to enable a wide range of applications with major scientific and economic impact.
"Fault-tolerant quantum computing with neutral atoms is a rapidly emerging topic, and it was clear there are many understudied opportunities for finding shortcuts," Bluvstein says. "We gathered some of the world's top experts in the topic at Caltech to put all the pieces together. What we came up with—a clear road map to building a quantum computer—came faster than we expected."
The next steps are to take larger arrays like those of Endres and his group and scale them up to even larger numbers while demonstrating low error rates, a process that will require additional technological advances.
The scientists founded Oratomic, with Bluvstein as CEO, with the goal to build the world's first utility-scale fault-tolerant quantum computers. Oratomic will work in close collaboration with Caltech's Advanced Quantum Computing Mission, an on-campus interdisciplinary effort, which will continue to study the fundamental science of quantum information processing. In the longer term, the Caltech team plans to have quantum "supercomputers" on campus for solving scientific problems.
"Now it's time to build the machines," Bluvstein says.
The study "Shor's algorithm is possible with as few as 10,000 reconfigurable atomic qubits" was conducted at Caltech and Oratomic. Research at Caltech was funded by Caltech. Manuel Endres and John Preskill also acknowledge support from the Institute for Quantum Information and Matter, a National Science Foundation (NSF) Physics Frontiers Center. Manuel Endres also acknowledges support from the NSF Quantum Leap Challenge Institutes program Challenge Institute for Quantum Computation. Qian Xu also acknowledges funding by the Walter Burke Institute for Theoretical Physics at Caltech.