| name | pauli-propagation-error-mitigation |
| description | Hybrid quantum-classical error mitigation framework embedding Pauli propagation with noise-canceling inverse channels. Use when mitigating errors in observable estimation on NISQ processors, combining classical simulation (Pauli propagation) with quantum hardware measurements, or extending observable estimation beyond classical/quantum limits alone. Activation: Pauli propagation, noise-canceling observable, hybrid error mitigation, quantum-centric supercomputing, observable estimation, inverse channel noise cancellation, truncated Pauli paths, quantum-classical co-simulation, 56-qubit benchmark, noise-aware Pauli truncation |
Pauli Propagation Error Mitigation
Methodology from arXiv:2606.20441 for hybrid error mitigation combining classical Pauli propagation with quantum noise cancellation.
Core Insight
Classical Pauli propagation and quantum hardware have complementary limitations:
- Pauli propagation: accuracy limited by exponential growth of operator paths; requires truncation
- Quantum hardware: accuracy limited by error rates and sampling overhead
By embedding Pauli propagation within a noise-canceling framework, both limitations are mitigated simultaneously.
Framework Architecture
Step 1: Define target observable O to be estimated
Step 2: Classically propagate O backwards through noise-canceling inverse channels
- This produces a modified observable O' = N^{-1}(O)
- The inverse channel N^{-1} counteracts hardware noise
Step 3: Truncate the expanded Pauli paths at a chosen order
Step 4: Measure O' directly on the quantum processor
Step 5: The measurement outcome approximates <psi|O|psi> with reduced error
Two Truncation Strategies
Strategy 1: Path-count truncation
Keep only the K largest-magnitude Pauli paths after propagation. Best for observables with few dominant terms.
Strategy 2: Weight-threshold truncation
Keep all paths above magnitude epsilon. Best for distributed-weight observables.
Key Results
- Benchmarked on 56 superconducting qubits (IBM hardware)
- Lower truncation errors with fewer classical resources vs. traditional Pauli propagation
- Reduced quantum sampling overhead vs. pure hardware approach
- Trade-offs between the two truncation strategies characterized numerically
Usage Pattern
Pattern 1: Shallow Circuit Observable Estimation
For circuits with depth < 20: propagate observable through 2-3 layers of inverse channels. Truncation at order 3-4 typically sufficient.
Pattern 2: Deep Circuit with Local Observables
For circuits with depth > 20: focus propagation on local observable support. Use weight-threshold truncation to capture dominant error channels.
Pattern 3: Global Observables
For global observables (e.g., magnetization): use path-count truncation with K ~ O(n log n) paths to balance accuracy vs. classical cost.
Implementation Steps
- Specify circuit and noise model (depolarizing, amplitude damping, etc.)
- Choose target observable (local operator, Hamiltonian term, etc.)
- Compute inverse channel N^{-1} from noise model
- Back-propagate observable: O' = N^{-1} compose U_dagger compose O compose U
- Truncate Pauli expansion of O' at chosen threshold
- Execute measurement of O' on quantum hardware
- Post-process: combine measurement outcomes to estimate
Error Handling
- Truncation error too large: increase truncation order or use path-count truncation with higher K
- Sampling overhead too large: reduce number of retained Pauli terms or use importance sampling
- Inverse channel unstable: regularize N^{-1} with small identity term (Tikhonov regularization)
References
- arXiv:2606.20441 - Computing noise-canceling observables via Pauli propagation