arXiv:2610.27556
Hybrid Lattice Surgery between Smooth Boundaries
Grace Ding, Yanzhu Chen, Sheng-Jie Huang
Abstract
We develop smooth hybrid lattice surgery for group surface codes, enabling joint logical measurements through merge and split operations along smooth boundaries. The code patches may correspond to identical or different finite groups, including non-Abelian groups. Our central result is that smooth merge and split realize a logical $K$-coset measurement, where $K$ is the subgroup defining the interface. This generalizes the joint logical parity measurement of ordinary surface codes. Using a tensor-network representation, we derive explicit measurement circuits and develop procedures for correcting nonidentity measurement outcomes. We illustrate the framework through a generalized controlled multiplication gate between group surface codes and protocols for preparing the $T$, $CZ$, $CCX$, and $CS$ magic states, producing resources encoded in one or more $\mathbb{Z}_2$ surface code patches. Our results show how smooth hybrid lattice surgery broadens the range of logical operations available for group surface codes.
AI Summary
The paper asks how lattice surgery along smooth boundaries can couple surface-code patches based on different finite groups, including non-Abelian groups, and what logical measurements these interfaces implement. It constructs merge–split operations for interfaces specified by a subgroup $K\leq G\times G'$, restricting to trivial interface cocycles. The main result is a logical $K$-coset measurement that preserves coherence within each selected coset subspace. Writing $K$ through subgroups $H,H'$, normal subgroups $N,N'$, and an isomorphism $\varphi:H/N\to H'/N'$, outcome $m'$ selects logical labels satisfying $j\in H$ and $N'k=\varphi(Nj)m'$. For $H=G$ this measures the full logical space; otherwise the construction postselects a logical restriction to $H$. Ordinary surface-code joint parity measurement is recovered as a special case. The logical action follows from contracting local interface constraints and summing ancilla labels. Tensor-network decompositions provide measurement circuits. Sequential corrections apply to arbitrary finite-group interface data; parallel corrections have sufficient conditions based on subnormal subgroup chains and nilpotent quotients. The illustrated $D_4$ interfaces require at most two boundary-restriction rounds and two plaquette-correction rounds. Applications combine smooth surgery with rough surgery, logical measurements, and existing transversal gates to obtain generalized controlled multiplication and protocols for resources labelled $T$, $CZ$, $CCX$, and $CS$ in qubit surface-code patches. Fault-tolerant decoding across changing interfaces, numerical thresholds, logical error rates, and space–time resource comparisons are left open. Protocol optimization and nontrivial-cocycle interfaces are not developed.
Central claim may not hold: For a proper subgroup H<G, the stated logical operation requires postselection onto j∈H rather than implementing a complete K-coset measurement on the full logical input space. The logical-projector derivation postselects every split ancilla onto its trivial representation, but the paper supplies no treatment of the other split outcomes while claiming that no postselection is needed when H=G. For H<G, the split leaves restricted H boundary checks rather than the original G boundary checks, without a procedure restoring the output to the original G surface-code space.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 15+7 pages, 14 figures, 1 table