Agent Skills: Möbius Path Filter

Möbius Path Filter

Category: Theorem Dependency Analysis Type: Graph Path Classification Language: Julia Status: Production Ready Version: 1.0.0 Date: December 22, 2025

Overview

Identifies tangled geodesics in proof dependency graphs via Möbius inversion. Classifies paths by prime factorization to determine which dependencies are problematic (create cycles) vs. optimal (linear chains).

Key Functions

• enumerate_paths(adjacency): Discovers all paths in graph
• factor_number(n): Prime factorization for Möbius weights
• mobius_weight(n): Computes μ(n) ∈ {-1, 0, +1}
• filter_tangled_paths(adjacency): Identifies problem paths
• generate_filter_report(): Human-readable analysis

Mathematical Foundation

Möbius Inversion for Path Classification

μ(n) = +1   : prime paths (keep - linear chains)
μ(n) = -1   : odd-composite paths (rewrite needed)
μ(n) = 0    : squared-factors (remove - redundant)

Uses prime factorization to weight geodesic paths in dependency graph. Helps identify which theorems create circular dependencies that impede spectral gap.

Usage

using MobiusFilter

# Analyze proof dependencies
prime_paths, tangled = filter_tangled_paths(adjacency)

# Get recommendations
report = generate_filter_report(adjacency)
println(report)

Integration Points

• Diagnosis tool for Week 2 analysis phase
• Feeds into safe_rewriting_advisor for remediation
• Used by continuous-inverter for automated detection

Performance

• Execution time: ~1 second (for 5-node test graphs)
• Path enumeration: Exponential but capped by practical graph size
• Prime factorization: O(√n) per path

References

• Hardy & Wright (1979): Elementary Number Theory
• Möbius inversion theory for discrete mathematics