#dlx #dancing-links #polyomino

exact-cover

(WIP) Asynchronous exact cover solver library using Knuth's dancing links algorithm

2 releases

0.1.1 Feb 2, 2022
0.1.0 Feb 1, 2022

#1131 in Algorithms

39 downloads per month

MIT license

42KB
817 lines

Asynchronous exact cover solver library using Knuth's dancing links (DLX) algorithm.

⚠️ This library is working in progress and the API is highly likely to be changed.

Concept

Many puzzle-like problems, such as polyomino packing, Sudoku, N-queens problem, etc. can be modeled as exact cover problems. This library provides an efficient solver to the generic exact cover problem and its generalizations, so that you can model your own problem as an exact cover problem, solve it, and further anaylize the solutions by code.

Basic example

use exact_cover::{Problem, Solver, SolverEvent};

fn main() {
    let mut prob = Problem::default();
    prob.add_constraints(1..=3);
    prob.add_subset("A", vec![1, 2, 3]);
    prob.add_subset("B", vec![1]);
    prob.add_subset("C", vec![2]);
    prob.add_subset("D", vec![3]);
    prob.add_subset("E", vec![1, 2]);
    prob.add_subset("F", vec![2, 3]);

    let mut solver = Solver::new(prob);
    let mut solutions = vec![];
    solver.run();

    for event in solver {
        if let SolverEvent::SolutionFound(sol) = event {
            solutions.push(sol);
        }
    }

    println!("{:?}", solutions); // [["A"], ["B", "C", "D"], ["B", "F"], ["E", "D"]]
}

Asynchronous API

⚠️ The feature is not available yet, but it is one of the primary goals of this project.

Solving a complex exact cover problem takes a long time. Users don't want to wait for the solving process to end without knowing how far it has progressed or how much time is left. This library provides an asynchronous API and various features to help with this issue.

  • Thanks to the asynchronous API, your program doesn't have to wait for the solver until it finds the next solution.
  • You can fetch the estimated progress of the solving process, anytime you want.
  • When the search space is too large and the solving process is not going to end in centuries, you can abort the solver.
  • You can pause the solving process and save the solver state to resume later.

Dependencies

~2MB
~31K SLoC