最短路径问题Shortest-Paths Problem

Single-Source Shortest-Paths Problem on a weighted graph G=(V, E)

主要学习了《Introduction to Algorithms》3rd，24章

1）Dijkstra Algorithm
all edges weights are nonnegative, w(u, v) >= 0 for each edge (u, v) ∈ E
「松弛Relaxation」, approximation to the correct distance is gradually replaced by more accurate values until eventually reaching the optimum solution.
（from wiki）

DIJKSTRA(G, w, s) – Graph, weight, source vertex
weight可以用「Adjacency LinkList」「Adjacency Matrix」表征

INITIALIZE-SINGLE-SOURCE(G, s)
S = ∅       // 空集，算法意义上指源点到所有顶点的距离正无穷
Q = G.V     // Queue，未处理的目标顶点序列
while(Q ≠ ∅)    // 1...|V|
    u = EXTRACT-MIN(Q)    // 选取总距离最近的顶点，O(V) for array
    S = S ∪ u
    for each vertex v ∈ G.adj[u]  // totally E times operation
        RELAX(u, v, w)    // 更新到“新”邻接顶点的最短总距离

都以「BFS」为基底，与Prim有些像
「Prim」不断拓展并候选邻接点，比较并选取到临点最轻的边
「Dijkstra」不断拓展并候选邻接点，比较并更新总距离，选取最近的点

The running time of Dijkstra algorithm depends on how we implement the 「Min-Priority Queue」.
*「Array」, O(V^2 + E) = O(V^2)
「Min-Priority Queue with a binary min-heap」, O((V+E)lgV) = O(ElgV)
「Min-Priority Queue with a Fibonacci heap」, O(VlgV + E)

猴子也能懂：目标宏远，步步为营

2）Bellman-Ford Algorithm
slower than Dijkstra, capable of handling negative weights
「Relaxation」

https://en.wikipedia.org/wiki/Bellman%E2%80%93Ford_algorithm
Dijkstra’s algorithm uses a priority queue to greedily select the closest vertex that has not yet been processed, and performs this relaxation process on all of its outgoing edges; by contrast, the Bellman–Ford algorithm simply relaxes all the edges, and does this |V|-1 times, where |V| is the number of vertices in the graph.

（from book）

BELLMAN-FORD(G, w, s)

INITIALIZE-SINGLE-SOURCE(G, s) // 初始依然是距离所有其他顶点正无穷
for i = 1 to |G.V| - 1         // 所有顶点扫一遍，进行核心松弛计算
    for each edge(u, v) ∈ G.E  // 每次对全边做优化...第一次一定是源的邻接顶点变得可计算...第二次逐渐多起来
        RELAX(u, v, w)         // 如果此边的u.d+w(u,v)<v.d，则更新
for each edge(u, v) ∈ G.E      // 负环校验
    if v.d > u.d + w(u, v)     // 扫了辣么多次，这式子还能成立！？
        return false           // --只可能是负环（有负权边的死循环）
return true

猴子也能懂：刷的飞起，负环狗带

O(VE), number of vertices and edges respectively