Awesome Quantum Computing Experiments
Benchmarking Experimental Progress Towards Fault-Tolerant Quantum Computation
June 4, 2026
Introduction
Overview
- Reviews experimental advancements towards fault-tolerant quantum computation (FTQC)
- Analyzes physical metrics: coherence times, entanglement error, qubit count
- Surveys QEC codes with parameters \([[n, k, d]]\)
- Open-source repository for community updates
Key Challenges
- Simultaneous progress in qubit performance and QEC needed
- Platforms: trapped ions, superconducting circuits, neutral atoms, NV centers, semiconductors
- Exponential trends in improvements over two decades
Physical Qubit Benchmarks
Platforms and Encoding
- Trapped ions: hyperfine or optical transitions
- Superconducting circuits: artificial atoms (transmons, fluxoniums)
- Neutral atoms: hyperfine ground states, Rydberg interactions
- NV centers: electron/nuclear spins in diamond
- Semiconductors: electron/hole spins in quantum dots
Coherence Times
Exponential improvements: doubling times from 0.9y (Se T1) to 5.9y (NA T1)
- T1: energy relaxation time, T2: dephasing time
- Mixing different superconducting flavors (transmons, resonators)
Entangled State Error
Error halving times: 1.2y (semiconductor spins) to 2.6y (superconducting circuits)
- Measures fidelity of multi-qubit operations
- Recent results below 1% across platforms
Qubit Count
Doubling times: 1.4y (neutral atoms) to 5.0y (ion traps)
- Tracks scaling to larger systems
- Largest: neutral atoms (6100 qubits), superconducting (127 qubits)
Utility Regime
Defining Utility Scale
- Working definition:
- Physical Qubit Count (PQC) of 10,000
- Entanglement Error (EE) below 0.1% (factor of 10 below surface code threshold)
- Projections based on exponential trends in EE halving and PQC doubling times
Caveats:
- Simplification
- Assumes trends continue
- True utility depends on additional factors like logical fidelity and gate speeds
Projected Timelines by Platform
- Projections estimate when platforms might reach utility thresholds
- Neutral atoms lead due to rapid scaling
| Platform | Best EE | Best EE Year | EE Halving Time (years) | Projected EE Utility Year | Best PQC | Best PQC Year | PQC Doubling Time (years) | Projected PQC Utility Year |
|---|---|---|---|---|---|---|---|---|
| Ion traps | 0.0003 | 2024 | 2.4 | 2024 | 100 | 2023 | 5.0 | 2056 |
| Neutral atoms | 0.0020 | 2023 | 2.3 | 2025 | 6,100 | 2024 | 1.4 | 2024 |
| Superconducting circuits | 0.0030 | 2020 | 2.6 | 2024 | 127 | 2023 | 2.6 | 2039 |
| Semiconductor spins | 0.0035 | 2022 | 1.2 | 2024 | 6 | 2022 | 2.6 | 2050 |
QEC Code Implementation
Concept of QEC
- Encode logical qubits in larger Hilbert space
- \([[n, k, d]]\) Parameters:
- \(n\) physical qubits,
- \(k\) logical qubits,
- \(d\) distance
- Detect and correct up to \(\lfloor (d-1)/2 \rfloor\) errors
Experiments Considered
- Codes:
- repetition, surface, color, Bacon-Shor
- Platforms:
- ions, superconducting, neutral atoms, photons, NV, NMR
- From 1998 (!!) demonstrations to 2024 milestones
Cumulative Experiment Counts
Accelerating growth, led by superconducting circuits & Recent rise in neutral atoms
QEC Cumulative
Breakdown by platform and code family
QEC Timeline
- Early: repetition codes in NMR
- Recent: surface codes with d=7, below threshold
\([[n, k, d]]\) Distribution
- Focus on low d (≤3) with n<50
- Progress to higher d: surface/color codes up to d=7
- Mostly k=1, some k=2/3
QEC Performance Benchmarks
- Suppression factor \(\Lambda\):
- up to 2.15 for surface codes
- Break-even:
- logical lifetime 2.4x physical
- Below threshold:
- 0.143% logical error per cycle
Conclusion
Conclusion
Summary
- Orders-of-magnitude improvements in physical metrics
- Persistent challenges: overhead, logical gates, decoding
Future Directions
- Incorporate finer metrics: cycle times, \(\Lambda\) factors
- Community contributions to repository
- Path to utility-scale FTQC
Check it out
Appendices
Experiment Counts per Year
- Non-cumulative view
- Peak activity in recent years
