Surface Code (Distance-3)

Overview

The surface code is a topological quantum error correcting code and the leading candidate for practical fault-tolerant quantum computing. It uses a 2D lattice of qubits with only nearest-neighbor interactions and has an error threshold of ~1% — achievable with today's hardware. Google demonstrated below-threshold operation in 2023.

This implementation is a minimal distance-3 surface code: 9 data qubits + 8 ancilla = 17 total qubits.

Why Surface Code?

2D Lattice Layout (Distance-3)

CODECopy
Data qubits (d): vertices    Ancillae: between vertices

    d0 ──── d1 ──── d2
    │  X_0  │  X_1  │
    d3 ──── d4 ──── d5
    │  X_2  │  X_3  │
    d6 ──── d7 ──── d8

X stabilizers (stars):   measure product of X on 4 adjacent data qubits
Z stabilizers (plaquettes): measure product of Z on 4 adjacent data qubits

Stabilizers

X Stabilizers (Stars) — Detect Z Errors

Z Stabilizers (Plaquettes) — Detect X Errors

Key Property: Central Qubit

Qubit d4 participates in all 8 stabilizers (4 X-type + 4 Z-type). An X error on d4 flips all 4 Z stabilizers (syndrome = 1111). This unique signature allows unambiguous identification.

Error Detection

Logical Operators

• Logical X: Chain of X operators spanning top to bottom (e.g., X_0 X_3 X_6)
• Logical Z: Chain of Z operators spanning left to right (e.g., Z_0 Z_1 Z_2)
• Logical measurement: Determined by boundary conditions

Error Threshold

The surface code can correct errors if the physical error rate satisfies:

p_physical < p_threshold ~ 1%

Below threshold, the logical error rate scales as:

p_logical ~ (p_physical / p_threshold)^((d+1)/2)

Each increase in distance d provides exponential suppression of logical errors.

Scaling with Distance

Decoding

The surface code requires a classical decoder to interpret syndromes and choose corrections:

• MWPM (Minimum-Weight Perfect Matching): Standard, near-optimal, O(n^3)
• Union-Find: Faster (almost linear time), slightly suboptimal
• Neural Network: Emerging approach, can adapt to hardware-specific noise

Running the Circuit

PYTHONCopy
from circuit import run_circuit, verify_surface_code

# No error — all syndromes should be trivial
result = run_circuit()
print(f"Trivial syndrome rate: {result['trivial_syndrome_rate']:.2%}")

# X error on qubit 4 (center) — all Z stabilizers should trigger
result = run_circuit(error_type='X', error_qubit=4)
print(f"Z syndrome: {max(result['z_syndrome_counts'], key=result['z_syndrome_counts'].get)}")

# Full verification
v = verify_surface_code()
for check in v["checks"]:
    print(f"[{'PASS' if check['passed'] else 'FAIL'}] {check['name']}")

Hardware Progress

Code Properties

Learn More

• Fowler et al. — Surface Codes Review (2012)
• Google — Below-Threshold Surface Code (2023)
• Dennis et al. — Topological Quantum Memory (2002)
• Surface Codes Tutorial (2012)