Towards Optimal Routing Strategies for qLDPC codes on Silicon Spin Qubits

One of the major challenges of implementing high-rate quantum error correction codes, like qLDPC codes, is realizing the required long-range interactions. Spin qubit platforms can provide this connectivity by shuttling qubits next to each other whenever they need to interact. When doing this, the performance of the code depends on the shuttling strategy. We study memory experiments of the [[144,12,12]] bivariate bicycle (Gross) code on two shuttling grids for future spin qubit devices: a square grid and a square grid with diagonals. We compare three shuttling strategies, structured and random initial layouts, and a range of shuttling noise models spanning best- and worst-case scenarios. The best strategy minimizes the longest distance travelled rather than the total distance. The initial layout has only a weak effect, whereas adding diagonals to the grid reduces the logical error rate by up to two orders of magnitude which may have implications for future hardware design.