Scientific Papers

Phys. Rev. X 13, 041013 (2023)

Quantum computation can be achieved with noisy quantum bits, or qubits, by storing information redundantly using error-correcting codes. Understanding the structure of noise in specific qubit hardware allows for more optimized correction schemes relative to an abstract, general noise model. Furthermore, an understanding of what noise is easier to correct can guide the design of physical qubits and gate operations. Previous work has considered two types of noise: biased noise, where phase-flip errors occur more often than bit-flip errors, and erasure noise, where the location of errors is known. In this work, we combine these two into a single model called “biased erasure” and show that it is the natural model for a neutral atom qubit based on ytterbium.

To study the behavior of this structured noise, we compute something called the threshold error rate, which is the maximum rate of errors that can be corrected. With a physically motivated noise model for a particular error-correcting code, the XZZX surface code, we estimate a threshold that is approximately 8 times as high as that of a general noise model. We also design an alternate implementation of the surface code adapted from the field of photonic quantum computing and demonstrate that this has even higher thresholds and is particularly well suited to implementation with neutral atom qubits.

This work represents a significant advance in the development of error-correction architectures for neutral atom qubits. It is also an example of how the joint design of qubits, gates, and error-correcting codes can lead to significant improvements in quantum computing performance.

Key Image

Source link