A team from Tokyo University of Science in Japan has achieved a significant breakthrough in quantum error correction technology: they have developed an efficient and scalable quantum low-density parity check (LDPC) error-correcting code that maintains extremely high stability in systems containing hundreds of thousands of logical qubits, approaching theoretical limits. This achievement provides key technical support for large-scale fault-tolerant quantum computing and is expected to promote the practical application of quantum computers in fields such as quantum chemistry, cryptanalysis, and complex optimization.
Currently, quantum computers can manipulate dozens of qubits, but solving practical problems often requires millions or even more stable and reliable logical qubits. Because quantum states are extremely fragile and susceptible to environmental perturbations, errors can accumulate rapidly as the system scales. Therefore, efficient error correction mechanisms are essential to maintain computational accuracy. However, existing quantum error correction methods generally suffer from high resource consumption and low efficiency. They typically require a large number of physical qubits to encode a small number of logical qubits, severely limiting the system’s scalability. This bottleneck has been the subject of ongoing research from the U.S. National Institute of Standards and Technology (NIST).
A deeper challenge lies in the low coding rate and limited potential for performance improvement of many existing error-correcting codes. Furthermore, performance often stagnates in high-precision operating regions, leaving a significant gap between the theoretically achievable optimal error correction limit and the hash bound. Furthermore, most schemes require complex post-processing after primary decoding, further increasing the computational burden.
The team successfully overcame these challenges. The Latest News has learned that they proposed a novel construction method, first designing prototype LDPC codes with excellent error-correcting properties. They then introduced techniques based on affine permutations to enhance the diversity of the code structure and effectively avoid the short-cycle problem that can hinder decoding performance. Unlike traditional LDPC codes defined over binary finite fields, this new scheme utilizes non-binary finite fields, allowing each code unit to carry more information and thus improving overall error correction capabilities.
The team then converted these prototype codes into a CSS-type quantum error-correcting code and, combined with an improved sum-product algorithm, developed an efficient joint decoding strategy. This method can simultaneously handle two fundamental quantum errors: bit flips and phase flips, whereas most previous schemes could only correct them one by one, resulting in lower efficiency.
Large-scale numerical simulations have verified that this new error-correcting code can achieve a frame error rate of 10⁻４ in a system containing hundreds of thousands of logical qubits, a performance very close to the hash limit. Importantly, the computational complexity required for decoding is proportional to the number of physical qubits, meaning that as the system scales up, the increase in resource overhead is linearly controllable, demonstrating good engineering feasibility.
