Quantum error correction on hexagonal lattice
Within the Willow structure, every bodily qubit is linked to its 4 nearest neighbors, forming a sq. lattice. This association of connections permits gating between adjoining qubits, but in addition introduces design constraints, resembling the extra wire overhead required to regulate couplers between qubits. Reaching error correction on a hexagonal lattice permits every qubit to attach with solely three neighboring qubits as a substitute of 4, simplifying the design and manufacturing course of of those giant chips and bettering {hardware} efficiency.
To realize error correction with as few as three couplers per qubit, we make the most of a dynamic circuit that includes two various kinds of error correction cycles. Each cycle varieties make the most of three couplers per qubit, and one coupler is utilized twice throughout the cycle. The result’s a quantum error correction circuit with dynamically overlapping detection areas. This circuit can nonetheless be used to triangulate errors, however solely three combiners per qubit are required.
We evaluated this three-coupler error correction circuit on a Willow processor with sq. connections. To measure the hex wire, we turned off all unused couplers to simulate hex connection efficiency. We discovered that because the code distance scaled from 3 to five, the logic error fee improved by an element of two.15, corresponding to the efficiency of conventional static circuits working on the identical {hardware} we introduced in final yr’s Milestone experiment.
Our findings show the feasibility of establishing hexagonal qubit lattices for quantum error correction, a design area that we totally explored in simulations. Adopting a hexagonal lattice can considerably scale back the complexity of the optimization algorithm for choosing qubits and gate frequencies. We present that this simplification improves the simulated error suppression issue by 15% and unlocks new capabilities by designing a processor with three couplers per qubit as a substitute of 4.
