Infleqtion’s neutral atom quantum computers are paving the way to multi-round error-corrected operation by using a state-of-the-art quantum state readout technique that we refer to as Non-Destructive State-Selective Readout (NDSSR).
As the name implies, the key feature of the technique is that it allows the readout of qubits' quantum states (for example, the states of ancilla qubits during mid-circuit measurement) without leading to the loss of those qubits. The name distinguishes the approach from the traditional “destructive readout” technique used in the neutral atom field, where one of the two qubit states is expelled from the system, and the quantum state is assessed from the presence or absence of remaining atoms.
While the neutral atom quantum computing field has seen recent impressive demonstrations toward fault-tolerant operation using destructive readout, the NDSSR technique developed by Infleqtion (and others*) is critical to extending to deep fault-tolerant circuits with multiple rounds of error correction. If you want to execute deep circuits, you’d better keep your qubits around!
Furthermore, when atoms are lost, the qubit array needs to be refreshed from a cold atom reservoir. This refreshing operation takes time, reducing the system's availability for circuit execution. So, not only does NDSSR enable deep fault-tolerant circuit execution, but it also enables faster operations!
Infleqtion’s QPU has recorded circuit execution shot rates above 10 Hz (in contrast with more typical operation rates at the level of 1 Hz using destructive readout). NDSSR also provides the advantage that it allows for detecting (and post-selecting against) some leakage-type errors in gate operations. Post-selecting for the absence of such errors boosts Infleqtion’s recently measured CZ fidelity to 99.5%±0.2%.
NDSSR characterization results show the distribution of the quantum state detection signal. In blue are the shots in which the qubit was initialized in qubit state |1>, and in orange are shots in which the qubit was prepared in state |0>. The third color comes from the overlap of the distributions (where a qubit prepared in |1> has a measured signal below threshold or a qubit prepared in |0> has a signal above threshold, which corresponds to errors due to imperfect readout or state initialization). A subsequent measurement verifies the continued presence of the atomic qubit. Typical state discrimination error is 2% with a sub-percent atom loss rate. The inset shows a single-shot camera image of a 4-qubit GHZ state in the center of a larger array of atoms prepared in state |1>. In this shot, the central qubits GHZ state was measured in state |0000> appearing dark.
*Phys. Rev. A 108, 032407 (2023) - High-fidelity low-loss state detection of alkali-metal atoms in optical tweezer traps
**Phys. Rev. X 13, 041051 (2023) - Midcircuit Measurements on a Single-Species Neutral Alkali Atom Quantum Processor
***Phys. Rev. Lett. 131, 030602 (2023) - Randomized Benchmarking Using Nondestructive Readout in a Two-Dimensional Atom Array