S-CLIFF: Random Clifford Benchmarking - Rationale

Experiment ID: S-CLIFF Workstream: S (Shadows) Status: Planned (Phase 1) Target: Nov 2025

Overview

S-CLIFF tests classical shadows on random Clifford circuits (5 qubits, depth-limited) to validate performance on non-stabilizer states. Compares shadow-based estimation vs. direct fidelity estimation (DFE) for ≥50 Pauli observables.

Scientific Rationale

1. Non-GHZ States: Test shadows beyond simple entangled states
2. Many Observables: Estimate ≥50 Paulis simultaneously (high-value demonstration)
3. DFE Comparison: Quantify shadows vs. standard benchmarking approach
4. Phase 2 Prep: Random states are precursor to chemistry/optimization problems

Expected Outcomes

• Circuit: Random Clifford, 5 qubits, depth ≤ 10
• Observables: ≥50 Pauli strings (full stabilizer group sampling)
• Shadow Budget: 500-1000
• SSR vs. DFE: ≥ 2× for many-observable regime

Relevant Literature

• Huang et al. (2021): Derandomization for Pauli observable estimation
• Emerson et al. (2005): Randomized benchmarking protocols

Part of Phase 1 Research Plan

Purpose: Validates shadows for general quantum states (not just GHZ) Timeline: Nov 2025 Priority: Medium (Phase 1 optional)