1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
//! Crate of algorithms for solving the traveling salesman problem.
//!
//! # Example
//! ```
//! use std::time;
//!
//! use tsp_rs::Tour;
//! use tsp_rs::point::Point;
//!
//! let tour: Vec<Point> = vec![
//!     Point::new(0., 0.),
//!     Point::new(0., 1.),
//!     Point::new(1., 0.),
//!     Point::new(1., 1.),
//! ];
//!
//! let mut tour = Tour::from(&tour);
//!
//! tour.optimize_kopt(std::time::Duration::from_secs(1));
//! ```
//!
//! _Disclaimer:_
//!
//! This is not a work of art, nor is it perfect (or even good?) Rust.
//! This was written alongside my first reading of the Rust book (https://doc.rust-lang.org/book/)
//! while trying to learn the language.
mod kopt;
mod nn;
pub mod point;

use std::borrow::Borrow;
use std::collections::HashSet;
use std::time;

use rand::Rng;

/// Trait used by all algorithms to calculate the cost of moving along an edge
///
/// # Examples
/// An example implementation is found on `tsp::point::Point`, that implements
/// standard euclidean distance as its metric.
pub trait Metrizable {
    fn cost(&self, other: &Self) -> f64;
}

/// Represents a solution to the tsp for the items T
#[derive(Debug, Clone, PartialEq)]
pub struct Tour<T: Metrizable> {
    pub path: Vec<T>,
}

impl<T: Metrizable + Clone + Borrow<T>> Tour<T> {
    /// Returns a new, empty Tour<T>
    ///
    /// # Example
    /// ```
    /// use tsp_rs::Tour;
    /// use tsp_rs::point::Point;
    ///
    /// let tour: Tour<Point> = Tour::new();
    /// ```
    pub fn new() -> Tour<T> {
        Tour {
            path: Vec::new() as Vec<T>,
        }
    }

    /// Returns a tour from `nodes: Vec<T>` passed in where the tour
    /// is nodes[0] -> nodes[1] -> ... nodes[nodes.len() - 1] -> nodes[0]
    ///
    /// # Example
    /// ```
    /// use tsp_rs::Tour;
    /// use tsp_rs::point::Point;
    ///
    /// let nodes = vec![
    ///     Point::new(0., 0.),
    ///     Point::new(1., 0.),
    ///     Point::new(1., 1.),
    ///     Point::new(0., 1.),
    /// ];
    ///
    /// let tour = Tour::from(&nodes);
    /// ```
    pub fn from(nodes: &Vec<T>) -> Tour<T>
    where
        T: Clone,
    {
        Tour {
            path: (*nodes).clone(),
        }
    }

    /// Returns the length of a tour.
    ///
    /// # Example
    /// let tour = Tour::from(&some_points);
    /// let total_cost = tour.tour_len();
    pub fn tour_len(&self) -> f64 {
        if self.path.len() <= 0 {
            return 0.;
        }

        let mut sum = 0.;
        let mut prev = self.path.last().unwrap();
        for curr in &self.path {
            sum += prev.cost(&curr);
            prev = &curr;
        }
        sum
    }

    /// Improves the tour in place using the 2opt heuristic (with 3opt kicks if it gets stuck)
    ///
    /// # Examples
    ///
    /// ```
    /// use std::time;
    ///
    /// use tsp_rs::Tour;
    /// use tsp_rs::point::Point;
    ///
    /// let nodes = vec![
    ///     Point::new(0., 0.),
    ///     Point::new(1., 0.),
    ///     Point::new(1., 1.),
    ///     Point::new(0., 1.),
    /// ];
    ///
    /// let mut tour = Tour::from(&nodes);
    ///
    /// tour.optimize_kopt(time::Duration::from_secs(1));
    /// ```
    pub fn optimize_kopt(&mut self, timeout: time::Duration) {
        self.optimize_nn();
        let start_time = time::Instant::now();
        let max_iter_withouth_impr = self.path.len() ^ 2;
        let mut iter_without_impr = 0;
        let mut best_tour_length = std::f64::MAX;
        let mut best_tour: Vec<T> = Vec::new();
        loop {
            match kopt::k_opt(2, self) {
                Some(_) => {
                    iter_without_impr = 0;
                }
                None => {
                    iter_without_impr += 1;
                    if iter_without_impr > max_iter_withouth_impr {
                        let current_tour_length = self.tour_len();
                        if current_tour_length < best_tour_length {
                            best_tour = self.path.clone();
                            best_tour_length = current_tour_length;
                        }
                        kopt::k_opt(4, self); // kick
                        iter_without_impr = 0;
                    }
                }
            }
            if start_time.elapsed() > timeout {
                break;
            }
        }
        let current_tour_length = self.tour_len();
        if current_tour_length < best_tour_length {
            best_tour = self.path.clone();
        }
        self.path = best_tour;
    }

    /// Constructs a tour inplace using the nearest neighbor heuristic
    ///
    /// # Examples
    ///
    /// ```
    /// use tsp_rs::Tour;
    /// use tsp_rs::point::Point;
    ///
    /// let nodes = vec![
    ///     Point::new(0., 0.),
    ///     Point::new(1., 0.),
    ///     Point::new(1., 1.),
    ///     Point::new(0., 1.),
    /// ];
    ///
    /// let mut tour = Tour::from(&nodes);
    ///
    /// tour.optimize_nn();
    /// ```
    pub fn optimize_nn(&mut self)
    where
        T: Metrizable + Clone,
    {
        let mut path = Vec::new();
        let nodes = index_path(&self.path);
        let mut visited = HashSet::new();

        let start_index: usize = rand::thread_rng().gen_range(0, nodes.len());
        let mut curr = &nodes[start_index].value.clone();
        path.push(curr.clone());
        visited.insert(nodes[start_index].index);

        loop {
            match nn::nearest_neighbor(curr, &nodes, &mut visited) {
                Some(next) => {
                    path.push(next.clone());
                    curr = &next;
                }
                None => {
                    if visited.len() == nodes.len() {
                        break;
                    }
                }
            };
        }

        self.path = path;
    }
}

#[derive(Clone)]
pub(crate) struct IndexedT<T> {
    pub index: usize,
    pub value: T,
}

#[inline]
pub(crate) fn index_path<T>(path: &Vec<T>) -> Vec<IndexedT<&T>> {
    path.iter()
        .enumerate()
        .map(|(index, value)| IndexedT { index, value })
        .collect()
}