Lower-depth local encoding circuits for the surface code

Introduction

The surface code remains the workhorse of fault‑tolerant quantum computing thanks to its high error threshold and purely two‑dimensional locality. A practical obstacle, however, is the encoding step: turning an arbitrary physical state into a logical surface‑code state. Existing constructions that respect 2‑D nearest‑neighbour connectivity need circuits of depth $2d$ to prepare a distance‑$d$ code. In a freshly published article in Quantum (Nov 2026), Jahan Claes proposes a family of depth‑$d$ encoding circuits for the rotated surface code, cutting the required depth in half and proving that this is optimal for inductively built encoders.

How the new encoder works

The construction proceeds by inductive growth: starting from a small base code, the circuit adds two layers of qubits at a time, expanding a distance‑$d$ code to distance $(d+2)$. Each growth step uses only local two‑qubit gates arranged on a square lattice, preserving the hardware‑friendly connectivity constraints of most quantum processors. By carefully ordering the CNOTs and exploiting the stabiliser structure of the rotated surface code, the author shows that the total depth scales as $d+O(1)$, matching a proven lower bound for any such inductive scheme.

Why depth matters

Circuit depth directly translates to exposure to decoherence and cumulative gate error. Halving the depth reduces the time the qubits spend in a vulnerable, unprotected state, which can dramatically improve the overall logical error rate when the encoded state is later used for computation or magic‑state distillation. Moreover, shallower circuits are easier to fit within the limited coherence windows of near‑term superconducting or trapped‑ion devices, making the result immediately relevant for experimental groups working on surface‑code demonstrations.

Context and comparison

Earlier works on local unitary encoders (e.g., Higgott et al. 2021) achieved depth proportional to $2d$, while more recent proposals for unitary encoders in other topological codes (e.g., Tsai & Puri 2026) explored alternative gate sets but did not improve the asymptotic depth. Claes’s inductive method therefore represents the first provably optimal reduction in depth for the rotated surface code under the strict locality constraint.

Practical outlook

Implementing the new encoder will require only standard two‑qubit gates and a square‑grid layout, both of which are already available on leading superconducting and neutral‑atom platforms. Researchers can benchmark the circuit on existing hardware by preparing a simple logical state (e.g., $|0_L\rangle$) and measuring stabiliser violations to verify the expected error‑suppression benefits. The paper also opens a pathway to explore whether similar inductive techniques can lower depths for other topological codes, such as the toric code or color codes.

Closing thoughts

By delivering a depth‑$d$ encoding circuit that is both local and optimal, this work narrows a key gap between theoretical fault‑tolerance and experimental feasibility. As quantum processors scale up, such engineering‑level improvements will be essential for turning the surface code from a theoretical construct into a practical error‑correcting layer.