Knowledge Base

Experimental retrospectives and parameter sweeps documenting the evidence behind PyFracVAL’s algorithmic design decisions.

• Pipeline Baseline: Status of Stages and Variants
• Comparison of CCA Sticking Methods
• Cluster Pairing and Geometric Frustration in CCA Sticking
• Matching-Based CCA Pairing: a Negative Result
• Backtracking CCA Pairing
• Statistical Overlap-Failure Census
• Drop-a-Few-Particles Rescue
• Catalog Overlap Leak: success=True Clusters With Severe Residual Overlap
• Df/kf/σ Stability Boundary Near the Hard Regime
• Stability Boundary After Backtracking (Sweep v2)
• Structural Validation via the Correlation Function f(r)
• Predicting the Feasibility Boundary
• Profiling and Performance
• Failure Statistics: the Structured Event Log
• Full-Grid Stability Sweep and Runtime Model
• GPU Acceleration Evaluation: JAX vs. Numba