Babysitter barren-plateau-analyzer

Analysis skill for detecting and mitigating barren plateaus in variational circuits

install

source · Clone the upstream repo

git clone https://github.com/a5c-ai/babysitter

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/a5c-ai/babysitter "$T" && mkdir -p ~/.claude/skills && cp -r "$T/library/specializations/domains/science/quantum-computing/skills/barren-plateau-analyzer" ~/.claude/skills/a5c-ai-babysitter-barren-plateau-analyzer && rm -rf "$T"

manifest: library/specializations/domains/science/quantum-computing/skills/barren-plateau-analyzer/SKILL.md

Barren Plateau Analyzer

Purpose

Provides expert guidance on analyzing and mitigating barren plateaus in variational quantum circuits, ensuring trainability of quantum machine learning models.

Capabilities

Gradient variance estimation
Cost function landscape analysis
Expressibility vs. trainability tradeoff
Initialization strategy evaluation
Local cost function design
Layer-wise training strategies
Entanglement-induced BP detection
Noise-induced BP analysis

Usage Guidelines

Variance Estimation: Sample gradient variance across parameter space
Scaling Analysis: Evaluate gradient scaling with qubit number
Architecture Modification: Redesign circuits to avoid BP regions
Initialization: Use structured initialization to avoid plateaus
Training Strategy: Apply layer-wise or identity-initialized training

Tools/Libraries

PennyLane
Qiskit
JAX
NumPy
Matplotlib