Skip to content

A lightweight toolkit for analysing p2p network topologies.

License

Notifications You must be signed in to change notification settings

niklaslong/spectre

Repository files navigation

Spectre

crates.io docs.rs

In French, the spectrum of a matrix is called "spectre".

This micro-library contains a small toolkit helpful in analysing network topologies. Namely:

  • Degree, adjacency and Laplacian matrices for a graph.
  • Various centrality measures for a graph.
  • The algebraic connectivity of a graph (Fiedler).

Basic usage

The library is centered around the Graph structure which can be constructed from one or more Edge instances. Once constructed, various measurements and matrix representations of the graph can be computed.

use std::net::SocketAddr;

use spectre::edge::Edge;
use spectre::graph::Graph;

// Construct the graph instance.
let mut graph = Graph::new();

// Create some addresses to be part of a network topology.
let addrs: Vec<SocketAddr> = (0..3)
    .map(|i| format!("127.0.0.1:{i}").parse().unwrap())
    .collect();
let (a, b, c) = (addrs[0], addrs[1], addrs[2]);

// Insert some edges, note the IDs can be any type that is `Copy + Eq + Hash + Ord`.
graph.insert(Edge::new(a, b));
graph.insert(Edge::new(a, c));

// Compute some metrics on that state of the graph.
let density = graph.density();
let degree_centrality_delta = graph.degree_centrality_delta();

// Matrices can be pretty printed...
println!("{}", graph.laplacian_matrix());
// ...outputs:
//  ┌          ┐
//  │  2 -1 -1 │
//  │ -1  1  0 │
//  │ -1  0  1 │
//  └          ┘

About

A lightweight toolkit for analysing p2p network topologies.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages