Quantum-Scaling RL Visualization Modules

Four visualization modules for analyzing Quantum-Scaling RL Hybrid Agent performance.

Modules Overview

1. Backend Performance Comparison

Compares IBM vs Russian backends across languages with mean reward and standard deviation.

Visualizations: Bar charts with error bars, grouped bars per language

2. Reward vs Batch Size Scaling

Shows how reward scales with batch size across different model sizes.

Visualizations: Scatter plots, scaling law validation, efficiency heatmaps

3. Cross-Lingual Backend Preference

Displays backend preferences per language based on learned heuristics.

Visualizations: Pie charts, language-backend matrices, horizontal bars

4. Performance Trend Over Edit Cycles

Tracks agent improvement over time through RL retraining and heuristic updates.

Visualizations: Line plots with moving average, stacked area charts, learning curves

Quick Start

# Run demo (generates 11 visualizations)
cd agent/visualizations
python demo_all_visualizations.py

Usage Example

from Backend_Performance_Comparison import plot_backend_performance_comparison

backend_performance = {
    'ibm': [0.807, 0.785, 0.820],
    'russian': [0.825, 0.810, 0.840]
}

plot_backend_performance_comparison(backend_performance, 'output.png')

Integration

from quantum_scaling_rl_hybrid import QuantumScalingRLHybrid
from visualizations.Backend_Performance_Comparison import plot_backend_performance_comparison

agent = QuantumScalingRLHybrid()
# ... run edit cycles ...
stats = agent.get_statistics()
plot_backend_performance_comparison(stats['backend_performance'])

Dependencies

pip install matplotlib numpy

Files

Backend_Performance_Comparison.py - Backend comparison charts
Reward_vs_BatchSize_Scaling.py - Batch size scaling analysis
Cross_Lingual_Backend_Preference.py - Language preference visualization
Performance_Trend_Over_Edit_Cycles.py - Performance trend tracking
demo_all_visualizations.py - Complete demo script

Output

All visualizations are 300 DPI PNG files with professional styling, clear labels, and color-coded data.