Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
Although quantum computing companies and researchers have made progress in scaling the number of physical qubits, this also tends to increase the rate of errors. A main concern in this area is that adding enough qubits together to solve significant problems may also lead to error-prone results.
Researchers at Quantinuum report that have recently found a way to scale the number of qubits to increase the performance and reduce the error rate. This is no simple task because quantum computers have a higher volume of errors compared with classical computers. In addition, many error correction techniques that form a mainstay of classical computing, like a parity check, also introduce new errors in quantum computing.
Quantinuum was formed by the merger of Cambridge Quantum Computing, a leading quantum software company, and the quantum hardware division of Honeywell. Cambridge Quantum Computing had been developing better quantum algorithms and ways to translate classical computer algorithms to work on quantum computers. Meanwhile, Honeywell had been pioneering a novel quantum computing ion trap architecture that allows qubits to connect more easily than other approaches.
Honeywell’s work allowed the team to transform 20 physical qubits into two more reliable logical qubits. Although this may seem like a step backward numerically speaking, it’s a tremendous step forward since these qubits can be added together.
Researchers commonly refer to the current generation of quantum computers as part of the noisy intermediate scale quantum (NISQ) era. This work will ultimately pave the way to build fault-tolerant quantum computers that can scale to address significant problems.
Quantum twist on redundancy
Hardware errors in which a transistor spontaneously switches tend to be rare in modern semiconductor circuits, but in some cases — like running a safety-critical system exposed to radiation — engineers design error correction systems that combine three processors. A supervisory system compares the results. If an error occurs, the supervisory system can detect if the calculation does not match and can safely ignore it if it does not match the others.
Quantum computer can introduce new problems. There are more kinds of errors that need to be corrected. A relatively simple parity check in classical computing can produce new errors in quantum computing.
Quantum computers can suffer from two kinds of errors: bit flips and phase flips. In a bit flip error, the qubit flips the computational state incorrectly from zero to one and vice versa. In a phase flip error, which does not occur in a classical computer, the phase of the qubit flips state. Previous theoretical research identified a way to correct both types of errors by constructing logical qubits. Last year, Quantinuum demonstrated a practical implementation of these techniques in a quantum computer using a 5 qubit code. However, this still increased errors as the number of qubits was scaled.
In the new technique, called a color code, the researchers found a way to combine seven logical qubits into one logical qubit in coordination with 2-3 ancillary qubits used for probing. They implemented this new color code technique on top of Quantinuum’s latest computer with 20 physical qubits to create two reliable logical qubits. These new logical qubits can be efficiently scaled in a way that increases fault tolerance that was not practical with the physical qubits or even the 5 qubit approach.
Russelll Stutz, director of commercial hardware at Quantinuum, told VentureBeat this means that as they add more qubits, the probability of getting failures that ruin the entire computation decreases with a modest rise in the number of physical qubits.
One remaining challenge is the quantum error correction cycle. The simple act of probing a qubit for errors can introduce new ones. Stutz said future work will explore ways to ensure they are not adding more errors than they remove with an error correction code.
Researchers have thought about how different quantum error correction approaches might work. Although the Quantinuum approach isn’t delivering as many raw physical qubits as other approaches, these are fully connected, which opens opportunities to leverage these innovative algorithms. In many quantum architectures, each qubit is only connected to a few neighbors.
“We are now testing quantum error correction code concepts dreamed up in the late 1990s and can implement in these real systems for the first time,” Stutz said. “It is an exciting time for learning about quantum error correction.”
Stutz says this research is a significant milestone on the long road to fault-tolerant quantum computing. He feels that researchers will be able to solve many practical problems once they scale systems to 50 logical qubits with lower error rates than physical qubits.
“It is laying the groundwork,” Stutz said. “You cannot really solve an industry-relevant problem with the number of logical qubits we are dealing with right now. We are essentially building really good components that will be used in a larger computation.”
Read more: IBM touts ‘Quantum Serverless’ as it eyes path to 4,000-plus qubit