競技プログラミングで使うグラフアルゴリズム Rust

1. ダイクストラ法

use std::collections::BinaryHeap;
const INF: usize = 1 << 61;

struct Graph {
    edges: Vec<Vec<(usize, i64)>>, // adjacent list
    n: usize
}

impl Graph {
    fn new(n: usize) -> Self {
        Graph {
            edges: vec![Vec::new(); n],
            n: n
        }
    }

    fn add_costs(&mut self, from: usize, to: usize, cost: i64, directed: bool) {
        if directed {
            self.edges[from].push((to, cost));
        } else {
            self.edges[from].push((to, cost));
            self.edges[to].push((from, cost));
        }

    }

    fn dijkstra_search(&self, start: usize) -> (Vec<i64>, Vec<usize>) {
        let mut dist: Vec<i64> = vec![INF; self.n]; // Dijkstraなのでi64ではなくusizeの方がいいかも
        let mut visited: Vec<bool> = vec![false; self.n];
        let mut prev: Vec<usize> = vec![INF as usize; self.n]; //nodeを格納するのにusizeでないの気持ち悪い
        dist[start] = 0;
        let mut que: BinaryHeap<(i64, usize)> = BinaryHeap::new();
        que.push((0, start));
        while let Some((_, v)) = que.pop() {
            if visited[v] {
                continue
            }
            visited[v] = true;
            for edge in &self.edges[v] {
                let (u, cost) = *edge;
                if visited[u] {
                    continue
                }
                if dist[u] > dist[v] + cost {
                    dist[u] = dist[v] + cost;
                    prev[u] = v;
                    que.push(( -dist[u], u));
                }
            }
        }
        return (dist, prev)
    }

    fn dijkstra_shortest_path(&self, start: usize, goal: usize) -> Vec<usize> {
        let (dist, prev) = self.dijkstra_search(start);
        let mut shortest_path: Vec<usize>= vec![];
        let mut node = goal;
        while node != INF {
            shortest_path.push(node);
            node = prev[node];
        }
        shortest_path.reverse();
        return shortest_path
    }
}

2.1. ベルマンフォード法 近接リスト

const INF: i64 = 1 << 61;


// adjacent list
struct Graph {
    edges: Vec<(usize, usize, i64)>, // adjacent list
    n: usize
}

impl Graph {
    fn new(n: usize) -> Self {
        Graph {
            edges: vec![],
            n: n
        }
    }

    fn add_costs(&mut self, from: usize, to: usize, cost: i64, directed: bool) {
        if directed {
            self.edges.push((from, to, cost));
        } else {
            self.edges.push((from, to, cost));
            self.edges.push((to, from, cost));
        }
    }

    fn bellmanford_search(&self, start: usize) -> Vec<i64> {
        let mut dist: Vec<i64> = vec![INF; self.n];
        dist[start] = 0;
        loop {
            let mut update: bool = false;
            for edge in &self.edges {
                let (from, to, cost) = *edge;
                if (dist[from] != INF) & (dist[to] > dist[from] + cost) {
                    dist[to] = dist[from] + cost;
                    update = true;
                }
            }
            if !update {
                break;
            }
        }
        return dist
    }

    fn bellmanford_search_negative_cycle(&self, start: usize) -> (bool, Vec<i64>) {
        // (true, dist) なら負の閉路を含む
        let mut dist: Vec<i64> = vec![INF; self.n];
        dist[start] = 0;
        for i in 0..self.n {
            for &(from, to, cost) in &self.edges{
                if (dist[from] != INF) & (dist[to] > dist[from] + cost) {
                    dist[to] = dist[from] + cost;
                    if i == self.n - 1{
                        return (true, dist);
                    }
                }
            }
        }
        return (false, dist);
    }
}

2.2. ベルマンフォード法　近接行列

//adjacent matrix
struct Graph {
    edges: Vec<Vec<(usize, i64)>>, // adjacent matrix
    n: usize
}

impl Graph {
    fn new(n: usize) -> Self {
        Graph {
            edges: vec![Vec::new(); n],
            n: n
        }
    }

    fn add_costs(&mut self, from: usize, to: usize, cost: i64, directed: bool) {
        if directed {
            self.edges[from].push((to, cost));
        } else {
            self.edges[from].push((to, cost));
            self.edges[to].push((from, cost));
        }
    }

    fn bellmanford_search(&self, start: usize) -> Vec<i64> {
        let mut dist: Vec<i64> = vec![INF; self.n];
        dist[start] = 0;
        let mut update: bool = true;
        while update {
            update = false;
            for from in 0..self.n {
                for edge in &self.edges[from] {
                    let (to, cost) = *edge;
                    if dist[to] > dist[from] + cost {
                        dist[to] = dist[from] + cost;
                        update = true;
                    }
                }
            }
        }
        return dist
    }

    fn bellmanford_search_negative_cycle(&self, start: usize) -> (bool, Vec<i64>) {
        // (true, dist) なら負の閉路を含む
        let mut dist: Vec<i64> = vec![INF; self.n];
        dist[start] = 0;
        let mut counter = 0;
        let mut update: bool = true;
        while update {
            update = false;
            for from in 0..self.n {
                for edge in &self.edges[from] {
                    let (to, cost) = *edge;
                    if (dist[from] != INF) & (dist[to] > dist[from] + cost) {
                        dist[to] = dist[from] + cost;
                        update = true;
                        counter += 1;
                    }
                }
            }
            if counter >= self.n * self.n {
                return (true, dist);
            }
        }
        return (false, dist);
    }
}

3. ワーシャルフロイド法

const INF: i64 = 1 << 61;
use std::cmp::min;


struct GraphWarshalFloyd {
    edges: Vec<Vec<i64>>,
    n: usize
}

impl GraphWarshalFloyd {
    fn new(n: usize) -> Self {
        GraphWarshalFloyd {
            edges: vec![vec![INF; n]; n],
            n: n
        }
    }

    fn add_costs(&mut self, from: usize, to: usize, cost: i64, directed: bool) {
        if directed {
            self.edges[from][to] = cost;
        } else {
            self.edges[from][to] = cost;
            self.edges[to][from] = cost;
        }
    }

    fn warshalfloyd(&self) -> (bool, Vec<Vec<i64>>) {
        // (true, dist) なら負の閉路を含む
        let mut dist: Vec<Vec<i64>> = vec![vec![INF; self.n]; self.n];
        for i in 0..self.n {
            for j in 0..self.n {
                dist[i][j] = self.edges[i][j]
            }
        }
        for i in 0..self.n {
            for j in 0..self.n {
                for k in 0..self.n {
                    if (dist[j][i] != INF) & (dist[i][k] != INF) {
                        dist[j][k] = min(dist[j][k], dist[j][i]+dist[i][k])
                    }
                }
            }
        }
        let mut has_negative_cycle = false;
        for i in 0..self.n {
            if dist[i][i] < 0 {
                has_negative_cycle = true;
            }
            dist[i][i] = 0;
        }
        return (has_negative_cycle, dist)
    }
}