RIKEN cuts the qubit cost of preparing a high-rate error-correcting code, in simulation
Short answer: Researchers at RIKEN's Center for Quantum Computing have designed a cheaper way to prepare the protected starting states that a high-rate quantum error-correcting code needs, according to a RIKEN announcement dated 14 November. The work was published in Science Advances on 13 November.
The team, Naoyuki Kanomata and Hayato Goto, studied a symplectic double code that stores six logical qubits in 30 physical qubits. RIKEN says that is a code rate of 20%, against about 4% for a surface code with similar error-correcting ability, and that other codes with a 20% rate usually need more than 100 physical qubits.
High-rate codes save hardware on paper, but preparing their encoded states fault-tolerantly normally adds many extra qubits and complex error checks. RIKEN says the team reduced the number of CNOT gates in the circuit that prepares all six logical qubits in the plus state from 108 to 67, then extended an error-detection method Goto had developed for the Steane code so that dangerous errors are caught with fewer qubits.
In numerical simulations of a logical CNOT gate, the method kept logical error rates similar to a conventional approach that uses six copies of the encoded state, while cutting the average total number of physical qubits needed for preparation to half or less. The team also proposed a way to prepare arbitrary states, which RIKEN says allows any gate operation on the code.
QubitWire notes that the core idea appeared in a preprint in September 2025, that the 30-qubit block is not the full size of the preparation system, and that the earlier manuscript's simulations left out memory errors and single-qubit gate errors. A hardware test would be needed to see how much of the saving survives real device noise.
