Crate spectral_fleet 

spectral-fleet

Spectral graph theory applied to AI agent fleet analysis and optimization.

A fleet is modeled as a graph where nodes are agents and edges are communication/dependency channels. Spectral analysis of the graph Laplacian reveals hidden structure:

• Number of clusters: count of zero eigenvalues
• Connectivity: spectral gap
• Bottleneck: Fiedler value (algebraic connectivity)

Modules

• fleet_graph: Fleet as a directed weighted graph
• laplacian: Graph Laplacian and eigenvalue decomposition
• clustering: Spectral clustering of agents
• bottleneck: Bottleneck detection and bypass suggestions
• reorganization: Fleet reorganization optimization
• dynamics: Temporal evolution and phase transitions
• embedding: Spectral embedding for visualization

Re-exports

pub use fleet_graph::AgentNode;

pub use fleet_graph::CommEdge;

pub use fleet_graph::FleetGraph;

pub use laplacian::Spectrum;

pub use clustering::FleetCluster;

pub use reorganization::Reorganization;

Modules

bottleneck

Bottleneck detection in fleet communication graphs.

clustering

Spectral clustering of fleet agents.

dynamics

Fleet graph dynamics: how the fleet evolves over time.

embedding

Spectral embedding: map agents to low-dimensional space.

fleet_graph

Fleet graph representation: agents as nodes, communication channels as edges.

laplacian

Graph Laplacian computation and eigenvalue decomposition.

reorganization

Fleet reorganization suggestions.