Experiments
This directory contains experimental designs, execution guides, and results for physics-based experiments on the StarForth VM.
Experiment Types
- factorial-doe/ - Factorial design of experiments for configuration analysis
- heartbeat-doe/ - Heartbeat subsystem performance experiments
- james-law/ - James Law window scaling validation experiments
- physics-optimization/ - Physics engine optimization experiments
Experiment Lifecycle
- Design - Define hypothesis, factors, response variables
- Protocol - Write execution guide (document methodology)
- Execute - Run experiments, collect data
- Analyze - Statistical analysis, validate results
- Document - Write summary, move to
06-research/if publishing
Key Results
Deterministic Behavior: StarForth achieves 0% algorithmic variance across 90 experimental runs, demonstrating formally proven deterministic behavior of the physics-driven adaptive runtime.
Design of Experiments Methodology
StarForth uses rigorous experimental methodology:
- Factorial designs for multi-factor analysis
- Heartbeat-driven data collection
- Statistical validation (ANOVA, Levene's test)
- Reproducible experimental protocols
See Also
- Design of Experiments schema:
../06-research/doe-metrics-schema.md - Results for publication:
../06-research/results-for-publication.md - Quality validation:
../04-quality/validation/