arXiv:2610.21357
Stitched surface codes: planar concatenation through GHZ corridors
Mohamed A. Shalby, Alexandre Bourassa, Amir Karamlou, Leigh Martin, Alexis Morvan, Nicholas Noll, Craig Gidney, Leonid P. Pryadko
Abstract
We present a fault-tolerant quantum-memory architecture that concatenates four distance-t rotated surface-code patches with the [[4,2,2]] outer code, encoding two logical qubits at code distance d = 2t in a physical footprint of N < 2(2t+1)^2 qubits on a square lattice with nearest-neighbor connectivity. The outer-code stabilizer generators XXXX and ZZZZ, each of physical weight 4t, are measured in situ via verified Greenberger-Horne-Zeilinger (GHZ) ancillae distributed through narrow corridors between adjacent patches, without lattice surgery or long-range gates. The outer-code gadgets introduce hyperedge fault mechanisms, but by design each decomposes into graphlike components, so the complete syndrome is decoded by a single minimum-weight-perfect-matching pass. Circuit-level simulations show logical error rates that scale as p^t, consistent with distance-2t protection. Finally, we implement the construction experimentally on a 105-qubit Willow superconducting processor in both the ZZ and XX memory bases. With the correlated-matching decoder, increasing the code distance from d = 4 to d = 6 reduces the block logical error rate by a factor of 1.52. The construction thus provides a near-term testbed for concatenated architectures beyond the standard surface code.
AI Summary
Can concatenated surface-code memories reduce qubit overhead while retaining planar nearest-neighbor connectivity on current hardware? This paper combines four distance-$t$ rotated surface-code patches with a $[[4,2,2]]$ outer code, encoding two logical qubits at code distance $d=2t$ using $N=2d^2+4d-3$ physical qubits. Verified GHZ ancillae in narrow corridors measure the weight-$4t$ outer stabilizers without surgery or long-range gates. Detector definitions permit hyperedge faults to decompose into graphlike components for joint matching-based decoding, optionally with correlated reweighting. Circuit-level Monte Carlo simulations show logical-error scaling proportional to $p^t$. Circuit-distance searches support $d_{\rm circ}=\min(2t,r_v+3)$ for verification depth $r_v\geq2$ over tested sizes; exhaustive searches certify distances through seven, while larger results are upper bounds. Under SI1000 noise, the simulated qubit-footprint advantage at matched logical error begins around $N=157$, $237$, and $333$ for $p=10^{-4}$, $5\times10^{-4}$, and $10^{-3}$, respectively. Memory experiments on a 105-qubit superconducting processor implement $d=4$ and $d=6$ layouts in both logical measurement bases. Correlated-matching decoding and a four-state Markov fit give basis-averaged block error probabilities per surface-code cycle of $5.836(12)\%$ and $3.850(7)\%$, a suppression factor of $1.516(4)$. The experimental $d=6$ circuit has circuit distance five, and the experiment does not demonstrate a footprint advantage over independent surface codes. Corridor noise makes savings operating-point dependent. Corridor parity extraction is substantially slower than lattice surgery: one simulated comparison requires about 80 versus five cycles to reach error $10^{-3}$. Logical operations remain undeveloped.
Error found: The nominal distance-six hardware code has circuit distance five, so its experimental protection should not be interpreted as a demonstration of full circuit distance six, although this limitation is disclosed in the body. The claim that verification depth r_v=2t-3 achieves circuit distance 2t is not established generally by the reported distance searches, because results above seven are only upper bounds.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 18 pages, 16 figures, 4 tables