6 stable releases

Uses old Rust 2015

3.0.0 Nov 29, 2018
2.0.0 Mar 18, 2018
1.1.0 Mar 18, 2018
1.0.2 Mar 11, 2018
1.0.1 Mar 6, 2018

#1378 in Algorithms

Download history 42/week @ 2023-10-29 25/week @ 2023-11-05 21/week @ 2023-11-12 31/week @ 2023-11-19 44/week @ 2023-11-26 28/week @ 2023-12-03 39/week @ 2023-12-10 26/week @ 2023-12-17 33/week @ 2023-12-24 18/week @ 2023-12-31 24/week @ 2024-01-07 30/week @ 2024-01-14 25/week @ 2024-01-21 26/week @ 2024-01-28 14/week @ 2024-02-04 41/week @ 2024-02-11

115 downloads per month
Used in sliding_puzzle

MIT license

11KB
236 lines

Lehmer

Travis

Lehmer is a Rust crate for converting between permutation vectors, Lehmer codes and their decimal representations. It is designed to run as quickly as possible and as a result, doesn't do any error handling.

This implementation is based on the 'Mapping Permutations to Integers' section of this paper. It doesn't implement the linear time speedup as the speed gains are only ~10% and require precomputing a lookup table.

Usage

extern crate lehmer;

use lehmer::Lehmer;

fn main() {
    // Compute the Lehmer code for a permutation:
    let lehmer = Lehmer::from_permutation(&[1, 0, 4, 3, 2]);

    assert_eq!(vec![1, 0, 2, 1, 0], lehmer.code);
    assert_eq!(29, lehmer.to_decimal());

    // Compute the Lehmer code for a decimal (requires the permutation length)
    let another = Lehmer::from_decimal(29, 5);

    assert_eq!(vec![1, 0, 2, 1, 0], another.code);
    assert_eq!(vec![1, 0, 4, 3, 2], another.to_permutation());

    // Compute the maximum decimal value for a permutation of five elements
    let max = Lehmer::max_value(5);
    assert_eq!(119, max);
}

Lehmer supports permutations up to 20 elements in length. The behaviour is not specified for permutations longer than this and will likely cause a panic. No unsafe operations are used, though.

Additionally, permutations must be vectors containing sequential integers starting from 0 (in any order), e.g. [1, 0, 4, 3, 2]. Lehmer will either panic or produce incorrect results for other vectors.

Benchmarks

Benchmarks can be run with cargo bench:

test benchmark_from_decimal     ... bench:         264 ns/iter (+/- 10)
test benchmark_from_permutation ... bench:          83 ns/iter (+/- 9)
test benchmark_max_value        ... bench:          13 ns/iter (+/- 1)
test benchmark_to_decimal       ... bench:          40 ns/iter (+/- 2)
test benchmark_to_permutation   ... bench:         142 ns/iter (+/- 9)

e.g. Lehmer::from_permutation runs at ~13 million iterations per second.

Tests

Tests can be run with cargo test. Unit tests are in files next to their modules and integration tests are in tests/.

Contributions

Contributions are welcome. Please test/benchmark your changes and open a PR.

No runtime deps