B-T01: RB/XEB Benchmarking - Rationale

Experiment ID: B-T01 Workstream: B (Benchmarking) Status: Planned (Phase 1) Target: Nov 2025

Overview

B-T01 demonstrates classical shadows for quantum benchmarking tasks: randomized benchmarking (RB) and cross-entropy benchmarking (XEB). Compares shadow-based fidelity/entropy estimation vs. standard protocols.

Scientific Rationale

1. Benchmarking Application: RB and XEB are standard NISQ device characterization tools
2. Shadow-Benchmarking: Estimate fidelity, purity, entropy from same shadow dataset
3. Sample Efficiency: Quantify shot savings vs. direct fidelity estimation
4. Phase 2 Foundation: Validates B-T02 (Shadow-Benchmarking full workflow)

Why Shadows for Benchmarking?

Advantage: Traditional benchmarking requires many measurements for: - Fidelity estimation (state tomography or DFE) - Purity estimation (shadow norm) - Entropy estimation (von Neumann entropy)

Shadows enable: All three metrics from single dataset

Expected Outcomes

• RB: 1-3 qubit randomized benchmarking, depth-limited sequences
• XEB: Depth-limited random circuits, compare to IBM calibration data
• Sample Efficiency: ≥ 2× vs. direct methods
• Manifest Integration: Log benchmarking results into standard provenance format

Relevant Literature

• Emerson et al. (2005): Randomized benchmarking protocol
• Arute et al. (2019): XEB for quantum supremacy (Google Sycamore)
• Huang et al. (2022): Shadow-based fidelity estimation

Part of Phase 1 Research Plan

Purpose: Extends shadows to benchmarking workstream Timeline: Nov 2025 Priority: MEDIUM (Phase 1 optional, valuable for Phase 2)