Fault Tolerance
The ability to perform reliable computation despite faulty components. The goal of large-scale quantum computing.
Fault tolerance means a system can continue working correctly even when individual components fail. In quantum computing, it means performing accurate computation despite noisy qubits and gates.
The Problem
Quantum hardware is noisy:
- Gates have ~0.1-1% error rates
- Qubits lose coherence over time
- Measurements can be wrong
Without fault tolerance, errors accumulate faster than computation proceeds.
The Solution: Error Correction Done Right
Quantum error correction encodes information protectively, but the correction process itself uses noisy gates!
Fault tolerance ensures errors during correction don’t spread uncontrollably:
- Errors in one qubit don’t propagate to many
- Syndrome extraction doesn’t amplify errors
- The whole system is self-consistent
Fault-Tolerant Design Principles
1. Transversal Gates
Apply gates qubit-by-qubit across code blocks:
Block 1: ──U──U──U──
Block 2: ──U──U──U──
Errors stay localized.
2. Flag Qubits
Detect when errors might have spread during syndrome measurement.
3. Code Concatenation
Encode logical qubits within logical qubits:
Physical → Level 1 logical → Level 2 logical → ...
Each level suppresses errors further.
4. Careful Circuit Design
Every syndrome extraction circuit must be fault-tolerant.
Threshold Theorem
The foundation of fault tolerance:
If the physical error rate is below a threshold , logical error rates can be made arbitrarily small by using enough physical qubits.
where is related to code distance.
Threshold Values
| Code | Approximate Threshold |
|---|---|
| Surface code | ~1% |
| Steane code | ~0.01% |
| Concatenated codes | ~0.001% |
Surface code’s high threshold is why it’s favored.
Requirements for Fault Tolerance
| Requirement | Why |
|---|---|
| Error rate below threshold | Foundation of everything |
| Fast classical processing | Decode syndromes in real-time |
| Scalable architecture | Need millions of qubits |
| Mid-circuit measurement | Measure syndromes during computation |
| Low-latency feedback | Correct errors before they spread |
Current Status
| Milestone | Status |
|---|---|
| Demonstrate error correction | Done |
| Below-threshold error rates | Achieved for some gates |
| Logical qubit better than physical | Recent demonstrations |
| Useful fault-tolerant computation | Years away |
The Path Forward
Today: NISQ (noisy, no error correction)
↓
Next: Early fault tolerance (limited logical qubits)
↓
Goal: Full fault tolerance (arbitrary computation)
See also: Quantum Error Correction, Logical Qubit, Fault-Tolerant Quantum Computing