Quantum Computer Recycles Its Atomic Qubits
Trapped neutral atoms are an attractive platform for quantum computing, as large arrays of atomic qubits can be arranged and manipulated to perform gate operations. However, the loss of useable atoms—either from escape or from disturbance—can be a limitation for long computations with repeated measurements. Researchers at Atom Computing, a company in California, have devised a “reset or reload” protocol that mitigates atom losses [1]. The method was successfully employed during a computation consisting of 41 cycles of qubit measurements.
All current quantum computers require error correction, which involves measuring certain qubits at intermediate steps of a computation. Reusing these qubits would avoid needing a prohibitively high overhead in qubit numbers, says team member Matthew Norcia. But in the case of atoms, the process of resetting measured qubits risks disturbing unmeasured ones.
To overcome this challenge, the researchers have developed a way to shield unmeasured atoms from the resetting process. They use targeted laser beams to immunize the unmeasured atoms against excitation by shifting their resonances. They then turn on a second set of lasers that cool the measured atoms and reinitialize them, enabling them to join the unmeasured atoms in the next computational step.
The protocol can also identify empty trap sites. To refill these spots, the researchers had previously built an optical conveyor belt, which shunts atoms from a reservoir to the computing region. Thanks to a refinement, this system can now reload atoms without agitating other atoms. As a demonstration of their new protocol, Norcia and colleagues performed computations with trapped ytterbium atoms and showed that stored information remained uncorrupted after several resetting cycles.
–Michael Schirber
Michael Schirber is the Deputy Editor of Physics Magazine.
References
- J. A. Muniz et al., “Repeated ancilla reuse for logical computation on a neutral atom quantum computer,” Phys. Rev. X 15, 041040 (2025).