Insider Brief
- Google Quantum AI’s surface code demonstrated significant error reduction in quantum computing, but IBM’s competing QLDPC code promises similar results with far fewer qubits.
- IBM’s QLDPC code leverages a unique connectivity strategy to minimize qubit overhead, sparking debate over the future of error correction methods in the quantum industry.
- The competition between error correction codes highlights the interplay between quantum hardware limitations and software innovations, with researchers exploring alternatives like ultracold-atom qubits for greater flexibility.
A significant advance by Google Quantum AI in quantum error correction, using a surface code approach, may have competition in a competing method that its advocates suggest offers greater efficiency and scalability. Researchers in the field are divided, however, over which approach will define the future of practical quantum computing, New Scientist is reporting.
Quantum computers, which hold the promise of solving complex problems in materials science, chemistry, and logistics, are extremely sensitive devices and, therefore, are plagued by errors. These errors also grow as the machines scale up, making error correction essential for practical use.
Researchers at Google Quantum AI recently demonstrated that their quantum processor, Willow, could mitigate this issue using the surface code, a mathematical framework that groups physical qubits into “logical qubits.” This grouping protects calculations from errors without negatively impacting performance.
The Google Quantum AI team members recently made headlines when they reported in a study that they were able to scale from a 3×3 grid to 5×5 and then to 7×7 grids of physical qubits reduced errors by a factor of two each time.
The method they used — called a surface code — has long been the dominant strategy for quantum error correction. It arranges qubits in interwoven grids, with data qubits performing calculations and ancillary qubits monitoring for errors. While effective, it requires a significant number of qubits to operate, which has limited its utility, according to New Scientist.
IBM’s Competing Approach
In 2023, IBM introduced a rival method called QLDPC (quantum low-density parity-check) code. Unlike the surface code, QLDPC connects each qubit to six others, allowing them to monitor each other’s errors. According to IBM researchers, this method could achieve the same error-correction capabilities as the surface code but with far fewer qubits. For example, on paper, where the surface code might require 4,000 qubits, QLDPC could deliver equivalent performance with just 288 qubits.
“With QLDPC, that lower qubit overhead is hard to compete with,” said Joe Fitzsimons of Horizon Quantum, a quantum computing startup, told New Scientist.
IBM has tailored its quantum chips to support the connectivity demands of QLDPC. While adding these connections poses engineering challenges, IBM has reported that the changes do not compromise the reliability of its chips.
Oliver Dial, an IBM researcher, emphasized the importance of tailoring codes to the capabilities of specific hardware during a presentation at the Q2B conference in December.
Balancing Hardware and Software
The competition between the surface code and the theoretical QLDPC highlights a broader challenge in quantum computing: the interplay between hardware and software. Superconducting qubits, like those used by Google and IBM, are limited in how they can be connected, making some error-correction methods more practical than others.
However, alternative technologies, such as qubits made from ultracold atoms, could provide greater flexibility.
“Maybe someone somewhere is working on a type of surface code that is really great, but right now there is competition [to the surface code],” said Yuval Boger of QuEra Computing, a U.S.-based quantum startup, as reported by New Scientist.
The QuEra team previously worked with ultracold-atom qubits to achieve one of the largest groups of logical qubits, exploring various codes to optimize their usefulness.
The Case for the Surface Code
Despite the excitement around QLDPC, the surface code remains a strong contender, Google’s team countered. Its theoretical framework is well understood, having been studied for more than two decades. It also offers a balance between performance and hardware requirements, making it particularly suitable for the superconducting qubits used in Google’s Willow processor.
“The surface code is well understood, with a well-studied theoretical framework. It offers a balance between performance and required qubit connectivity,” said Sergio Boixo of Google Quantum AI, as reported in New Scientist.
Google, however, is not resting on its laurels. Boixo confirmed that the team is exploring alternative error-correction codes alongside the surface code.