gryf

Graph data structure library aspiring to be convenient, versatile, correct and performant.

Status: The library is not released on crates.io yet. It is incomplete, lacks documentation and contains bugs. Breaking changes are expected. Design contributions are very welcome!

gryf = { git = "https://github.com/pnevyk/gryf.git" }

Example

use gryf::{algo::ShortestPaths, graph::Graph};

fn main() {
    // Default storage is adjacency list, but that can be simply changed by
    // using `Graph::new_undirected_in`.
    let mut graph = Graph::new_undirected();

    let prague = graph.add_vertex("Prague");
    let bratislava = graph.add_vertex("Bratislava");
    let vienna = graph.add_vertex("Vienna");
    let munich = graph.add_vertex("Munich");
    let nuremberg = graph.add_vertex("Nuremberg");
    let florence = graph.add_vertex("Florence");
    let rome = graph.add_vertex("Rome");

    graph.extend_with_edges([
        (prague, bratislava, 328u32),
        (prague, nuremberg, 293),
        (bratislava, vienna, 79),
        (nuremberg, munich, 170),
        (vienna, munich, 402),
        (munich, florence, 646),
        (florence, rome, 278),
    ]);

    // As the edge weights are unsigned and there is a specific goal, Dijktra's
    // algorithm is applied. For signed edges, Bellman-Ford would be used.
    let shortest_paths = ShortestPaths::on(&graph).goal(prague).run(rome).unwrap();
    let distance = shortest_paths[prague];
    let path = shortest_paths
        .reconstruct(prague)
        .map(|v| graph[v])
        .collect::<Vec<_>>()
        .join(" - ");

    println!("{distance} km from Prague through {path}");
    // 1387 km from Prague through Nuremberg - Munich - Florence - Rome
}

Overview

The main goal of gryf is to be

• convenient -- "make the common case straightforward and natural¹",
• versatile -- "offer simplicity as well as flexibility and strive for a good balance if in conflict",
• correct -- "use extensive fuzzing and property-based testing to increase confidence about correctness", and
• performant -- "write the code with performance and memory efficiency in mind".

The algorithms are organized into the problems they solve. For specifying the options of an algorithm the builder pattern is utilized. Graphs can use different storage under the hood.

¹Failing in this should be considered a bug and reported.

Details

For more details, see the design doc.

Problems instead of algorithms

For users without much experience or knowledge in graph theory and algorithms, it may not be obvious which algorithm should (or even can) be used to solve the given problem at hand. Instead, gryf organizes the algorithms into the problem they solve (e.g., ShortestPaths) instead of exposing the algorithms directly (dijkstra, bellman_ford).

This brings a number of benefits, among which the most important are:

• It is convenient for the user, especially if they are a beginner. It allows them not to care about details if they don't want to care.
• Having a specific type instead of a generic one such as Vec or HashMap gives the opportunity to provide additional functionality (like path reconstruction for shortest paths or "is perfect?" query on matching).
• Not specifying the algorithm enables the use of automatic algorithm selection, which makes the decision based on the properties of the input graph.

let shortest_paths = ShortestPaths::on(&graph).run(rome).unwrap();

Builder pattern for algorithms

Specifying arguments for algorithms is done using the builder pattern. This avoids the need of passing dummy values (like None) to parameters that are not useful for the use case. On the other hand, it allows tweaking the algorithm with many optional arguments. Moreover, new optional parameters can be added in a backward-compatible way. A lot of care is taken to make the error feedback from the compiler helpful and obvious.

let shortest_paths = ShortestPaths::on(&graph)
    .edge_weight_fn(|e| e.distance)
    .goal(prague)
    .run(rome)
    .unwrap();

Separation of graph storage and semantics

In gryf, high-level semantics provided by user-facing types are strictly separated from the underlying storage/representation. The graph data can be stored in a common representation (e.g., adjacency list or adjacency matrix), but it can as well be stored in or represented by a custom, problem-tailored implementation, as long as it implements provided interfaces.

On top of a storage, there is an encapsulation with clear semantics. The most general is a generic graph, but restricted forms include simple graph (without parallel edges), path, bipartite graph and so on. Among the advantages of restrictive encapsulations are:

• The type of graph clearly communicates the intention and structure.
• The API is limited such that it is impossible to violate the rules of the user-desired class of graph.
• The guaranteed properties of a restricted graph can be utilized in choosing a more efficient algorithm.

use gryf::storage::AdjMatrix;

let mut graph = Graph::new_undirected_in(AdjMatrix::default());

Alternatives (and inspiration)

• petgraph
• prepona
• pathfinding
• graph
• graphlib
• graphific

See the differences between them and gryf in this comparison repository.

License

Dual-licensed under MIT and UNLICENSE. Feel free to use it, contribute or spread the word.