一、思想
1.1 基本概念
- 加权无向图的生成树：一棵含有其所有顶点的无环连通子图。
- 最小生成树（MST）：一棵权值最小（树中所有边的权值之和）的生成树。
1.2 算法原理
1.2.1 切分定理
- 切分定义：图的一种切分是将图的所有顶点分为两个非空且不重合的两个集合。横切边是一条连接两个属于不同集合的顶点的边。
- 切分定理：在一幅加权图中，给定任意的切分，它的横切边中的权重最小者必然属于图的最小生成树。
1.2.2 算法原理
切分定理是解决最小生成树问题的所有算法的基础。切分定理再结合贪心算法思想，就可以最终落地实现最小生成树。
Prim算法原理：
一开始树中只有一个顶点，向它添加v-1条边，每次总是将下一条连接 “树中的顶点” 与 “不在树中的顶点” 且权重最小的边，加入树中。如下图，当我们将顶点v添加到树中时，可能使得w到最小生成树的距离更近了（然后遍历顶点v的领接链表即可）。
核心：
使用一个索引优先队列，保存每个非树顶点w的一条边（将它与树中顶点连接起来的权重最小的边）。优先队列（小顶堆）的最小键即是权重最小的横切边的权重，而和它相关联的顶点V就是下一个将被添加到树中的顶点。
二、实现
2.1 无向边
1 package study.algorithm.graph; 2 3 import study.algorithm.base.StdOut; 4 5 /*** 6 * @Description 无向边 7 * @author denny.zhang 8 * @date 2020/5/25 10:34 上午 9 */ 10 public class Edge implements Comparable<Edge> { 11 12 /** 13 * 一个顶点 14 */ 15 private final int v; 16 /** 17 * 另一个顶点 18 */ 19 private final int w; 20 /** 21 * 权重 22 */ 23 private final double weight; 24 25 /** 26 * Initializes an edge between vertices {@code v} and {@code w} of 27 * the given {@code weight}. 28 * 29 * @param v one vertex 30 * @param w the other vertex 31 * @param weight the weight of this edge 32 * @throws IllegalArgumentException if either {@code v} or {@code w} 33 * is a negative integer 34 * @throws IllegalArgumentException if {@code weight} is {@code NaN} 35 */ 36 public Edge(int v, int w, double weight) { 37 if (v < 0) throw new IllegalArgumentException("vertex index must be a nonnegative integer"); 38 if (w < 0) throw new IllegalArgumentException("vertex index must be a nonnegative integer"); 39 if (Double.isNaN(weight)) throw new IllegalArgumentException("Weight is NaN"); 40 this.v = v; 41 this.w = w; 42 this.weight = weight; 43 } 44 45 /** 46 * Returns the weight of this edge. 47 * 48 * @return the weight of this edge 49 */ 50 public double weight() { 51 return weight; 52 } 53 54 /** 55 * 返回边的任意一个顶点 56 * 57 * @return either endpoint of this edge 58 */ 59 public int either() { 60 return v; 61 } 62 63 /** 64 * 返回边的另一个顶点 65 * 66 * @param vertex one endpoint of this edge 67 * @return the other endpoint of this edge 68 * @throws IllegalArgumentException if the vertex is not one of the 69 * endpoints of this edge 70 */ 71 public int other(int vertex) { 72 if (vertex == v) return w; 73 else if (vertex == w) return v; 74 else throw new IllegalArgumentException("Illegal endpoint"); 75 } 76 77 /** 78 * Compares two edges by weight. 79 * Note that {@code compareTo()} is not consistent with {@code equals()}, 80 * which uses the reference equality implementation inherited from {@code Object}. 81 * 82 * @param that the other edge 83 * @return a negative integer, zero, or positive integer depending on whether 84 * the weight of this is less than, equal to, or greater than the 85 * argument edge 86 */ 87 @Override 88 public int compareTo(Edge that) { 89 return Double.compare(this.weight, that.weight); 90 } 91 92 /** 93 * Returns a string representation of this edge. 94 * 95 * @return a string representation of this edge 96 */ 97 public String toString() { 98 return String.format("%d-%d %.5f", v, w, weight); 99 } 100 101 /** 102 * Unit tests the {@code Edge} data type. 103 * 104 * @param args the command-line arguments 105 */ 106 public static void main(String[] args) { 107 Edge e = new Edge(12, 34, 5.67); 108 StdOut.println(e); 109 StdOut.println("任意一个顶点="+e.either()); 110 StdOut.println("另一个顶点="+e.other(12)); 111 } 112 }
如上图，初始化时构造了一个邻接表。Bag<Edge>[] adj，如下图。每条边有2个顶点，所以插入adj[]中2次。
2.2.边加权无向图
1 package study.algorithm.graph; 2 3 import study.algorithm.base.*; 4 5 import java.util.NoSuchElementException; 6 7 /*** 8 * @Description 边权重无向图 9 * @author denny.zhang 10 * @date 2020/5/25 10:50 上午 11 */ 12 public class EdgeWeightedGraph { 13 private static final String NEWLINE = System.getProperty("line.separator"); 14 15 /** 16 * 顶点数 17 */ 18 private final int V; 19 /** 20 * 边数 21 */ 22 private int E; 23 /** 24 * 顶点邻接表，每个元素Bag代表：由某个顶点关联的边数组,按顶点顺序排列 25 */ 26 private Bag<Edge>[] adj; 27 28 /** 29 * Initializes an empty edge-weighted graph with {@code V} vertices and 0 edges. 30 * 31 * @param V the number of vertices 32 * @throws IllegalArgumentException if {@code V < 0} 33 */ 34 public EdgeWeightedGraph(int V) { 35 if (V < 0) throw new IllegalArgumentException("Number of vertices must be nonnegative"); 36 this.V = V; 37 this.E = 0; 38 adj = (Bag<Edge>[]) new Bag[V]; 39 for (int v = 0; v < V; v++) { 40 adj[v] = new Bag<Edge>(); 41 } 42 } 43 44 /** 45 * Initializes a random edge-weighted graph with {@code V} vertices and <em>E</em> edges. 46 * 47 * @param V the number of vertices 48 * @param E the number of edges 49 * @throws IllegalArgumentException if {@code V < 0} 50 * @throws IllegalArgumentException if {@code E < 0} 51 */ 52 public EdgeWeightedGraph(int V, int E) { 53 this(V); 54 if (E < 0) throw new IllegalArgumentException("Number of edges must be nonnegative"); 55 for (int i = 0; i < E; i++) { 56 int v = StdRandom.uniform(V); 57 int w = StdRandom.uniform(V); 58 double weight = Math.round(100 * StdRandom.uniform()) / 100.0; 59 Edge e = new Edge(v, w, weight); 60 addEdge(e); 61 } 62 } 63 64 /** 65 * Initializes an edge-weighted graph from an input stream. 66 * The format is the number of vertices <em>V</em>, 67 * followed by the number of edges <em>E</em>, 68 * followed by <em>E</em> pairs of vertices and edge weights, 69 * with each entry separated by whitespace. 70 * 71 * @param in the input stream 72 * @throws IllegalArgumentException if {@code in} is {@code null} 73 * @throws IllegalArgumentException if the endpoints of any edge are not in prescribed range 74 * @throws IllegalArgumentException if the number of vertices or edges is negative 75 */ 76 public EdgeWeightedGraph(In in) { 77 if (in == null) throw new IllegalArgumentException("argument is null"); 78 79 try { 80 // 顶点数 81 V = in.readInt(); 82 // 邻接表 83 adj = (Bag<Edge>[]) new Bag[V]; 84 // 初始化邻接表，一个顶点对应一条链表 85 for (int v = 0; v < V; v++) { 86 adj[v] = new Bag<Edge>(); 87 } 88 // 边数 89 int E = in.readInt(); 90 if (E < 0) throw new IllegalArgumentException("Number of edges must be nonnegative"); 91 // 遍历每一条边 92 for (int i = 0; i < E; i++) { 93 // 一个顶点 94 int v = in.readInt(); 95 // 另一个顶点 96 int w = in.readInt(); 97 validateVertex(v); 98 validateVertex(w); 99 // 权重 100 double weight = in.readDouble(); 101 // 构造边 102 Edge e = new Edge(v, w, weight); 103 // 添加边 104 addEdge(e); 105 } 106 } 107 catch (NoSuchElementException e) { 108 throw new IllegalArgumentException("invalid input format in EdgeWeightedGraph constructor", e); 109 } 110 111 } 112 113 /** 114 * Initializes a new edge-weighted graph that is a deep copy of {@code G}. 115 * 116 * @param G the edge-weighted graph to copy 117 */ 118 public EdgeWeightedGraph(EdgeWeightedGraph G) { 119 this(G.V()); 120 this.E = G.E(); 121 for (int v = 0; v < G.V(); v++) { 122 // reverse so that adjacency list is in same order as original 123 Stack<Edge> reverse = new Stack<Edge>(); 124 for (Edge e : G.adj[v]) { 125 reverse.push(e); 126 } 127 for (Edge e : reverse) { 128 adj[v].add(e); 129 } 130 } 131 } 132 133 134 /** 135 * Returns the number of vertices in this edge-weighted graph. 136 * 137 * @return the number of vertices in this edge-weighted graph 138 */ 139 public int V() { 140 return V; 141 } 142 143 /** 144 * Returns the number of edges in this edge-weighted graph. 145 * 146 * @return the number of edges in this edge-weighted graph 147 */ 148 public int E() { 149 return E; 150 } 151 152 // throw an IllegalArgumentException unless {@code 0 <= v < V} 153 private void validateVertex(int v) { 154 if (v < 0 || v >= V) 155 throw new IllegalArgumentException("vertex " + v + " is not between 0 and " + (V-1)); 156 } 157 158 /** 159 * Adds the undirected edge {@code e} to this edge-weighted graph. 160 * 161 * @param e the edge 162 * @throws IllegalArgumentException unless both endpoints are between {@code 0} and {@code V-1} 163 */ 164 public void addEdge(Edge e) { 165 // 一个顶点 166 int v = e.either(); 167 // 另一个顶点 168 int w = e.other(v); 169 validateVertex(v); 170 validateVertex(w); 171 // 追加进顶点v的邻接链表 172 adj[v].add(e); 173 // 追加进顶点w的领接链表 174 adj[w].add(e); 175 // 边数++ 176 E++; 177 } 178 179 /** 180 * 顶点V关联的全部边 181 * 182 * @param v the vertex 183 * @return the edges incident on vertex {@code v} as an Iterable 184 * @throws IllegalArgumentException unless {@code 0 <= v < V} 185 */ 186 public Iterable<Edge> adj(int v) { 187 validateVertex(v); 188 return adj[v]; 189 } 190 191 /** 192 * Returns the degree of vertex {@code v}. 193 * 194 * @param v the vertex 195 * @return the degree of vertex {@code v} 196 * @throws IllegalArgumentException unless {@code 0 <= v < V} 197 */ 198 public int degree(int v) { 199 validateVertex(v); 200 return adj[v].size(); 201 } 202 203 /** 204 * Returns all edges in this edge-weighted graph. 205 * To iterate over the edges in this edge-weighted graph, use foreach notation: 206 * {@code for (Edge e : G.edges())}. 207 * 208 * @return all edges in this edge-weighted graph, as an iterable 209 */ 210 public Iterable<Edge> edges() { 211 Bag<Edge> list = new Bag<Edge>(); 212 for (int v = 0; v < V; v++) { 213 int selfLoops = 0; 214 for (Edge e : adj(v)) { 215 if (e.other(v) > v) { 216 list.add(e); 217 } 218 // add only one copy of each self loop (self loops will be consecutive) 219 else if (e.other(v) == v) { 220 if (selfLoops % 2 == 0) list.add(e); 221 selfLoops++; 222 } 223 } 224 } 225 return list; 226 } 227 228 /** 229 * Returns a string representation of the edge-weighted graph. 230 * This method takes time proportional to <em>E</em> + <em>V</em>. 231 * 232 * @return the number of vertices <em>V</em>, followed by the number of edges <em>E</em>, 233 * followed by the <em>V</em> adjacency lists of edges 234 */ 235 public String toString() { 236 StringBuilder s = new StringBuilder(); 237 s.append(V + " " + E + NEWLINE); 238 for (int v = 0; v < V; v++) { 239 s.append(v + ": "); 240 for (Edge e : adj[v]) { 241 s.append(e + " "); 242 } 243 s.append(NEWLINE); 244 } 245 return s.toString(); 246 } 247 248 /** 249 * Unit tests the {@code EdgeWeightedGraph} data type. 250 * 251 * @param args the command-line arguments 252 */ 253 public static void main(String[] args) { 254 In in = new In(args[0]); 255 EdgeWeightedGraph G = new EdgeWeightedGraph(in); 256 StdOut.println(G); 257 } 258 259 }
2.3 索引小值优先队列
1 package study.algorithm.base; 2 3 import java.util.Iterator; 4 import java.util.NoSuchElementException; 5 6 /** 7 * 索引（顶点）最小优先级队列 8 * 9 * @param <Key> 10 */ 11 public class IndexMinPQ<Key extends Comparable<Key>> implements Iterable<Integer> { 12 /** 13 * 元素数量上限 14 */ 15 private int maxN; 16 /** 17 * 元素数量 18 */ 19 private int n; 20 /** 21 * 索引二叉堆（数组中每个元素都是顶点，顶点v，对应keys[v]）：数组从pq[0]代表原点其它顶点从pq[1]开始插入 22 */ 23 private int[] pq; 24 /** 25 * 标记索引为i的元素在二叉堆中的位置。pq的反转数组（qp[index]=i）：qp[pq[i]] = pq[qp[i]] = i 26 */ 27 private int[] qp; 28 29 /** 30 * 元素有序数组（按照pq的索引赋值） 31 */ 32 public Key[] keys; 33 34 /** 35 * 初始化一个空索引优先队列，索引范围:0 ~ maxN-1 36 * 37 * @param maxN the keys on this priority queue are index from {@code 0} 38 * {@code maxN - 1} 39 * @throws IllegalArgumentException if {@code maxN < 0} 40 */ 41 public IndexMinPQ(int maxN) { 42 if (maxN < 0) throw new IllegalArgumentException(); 43 this.maxN = maxN; 44 // 初始有0个元素 45 n = 0; 46 // 初始化键数组长度为maxN + 1 47 keys = (Key[]) new Comparable[maxN + 1]; 48 // 初始化"键值对"数组长度为maxN + 1 49 pq = new int[maxN + 1]; 50 // 初始化"值键对"数组长度为maxN + 1 51 qp = new int[maxN + 1]; 52 // 遍历给"值键对"数组赋值-1，后续只要!=-1,即包含i 53 for (int i = 0; i <= maxN; i++) 54 qp[i] = -1; 55 } 56 57 /** 58 * Returns true if this priority queue is empty. 59 * 60 * @return {@code true} if this priority queue is empty; 61 * {@code false} otherwise 62 */ 63 public boolean isEmpty() { 64 return n == 0; 65 } 66 67 /** 68 * Is {@code i} an index on this priority queue? 69 * 70 * @param i an index 71 * @return {@code true} if {@code i} is an index on this priority queue; 72 * {@code false} otherwise 73 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 74 */ 75 public boolean contains(int i) { 76 validateIndex(i); 77 return qp[i] != -1; 78 } 79 80 /** 81 * Returns the number of keys on this priority queue. 82 * 83 * @return the number of keys on this priority queue 84 */ 85 public int size() { 86 return n; 87 } 88 89 /** 90 * 插入一个元素，将元素key关联索引i 91 * 92 * @param i an index 93 * @param key the key to associate with index {@code i} 94 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 95 * @throws IllegalArgumentException if there already is an item associated 96 * with index {@code i} 97 */ 98 public void insert(int i, Key key) { 99 validateIndex(i); 100 if (contains(i)) throw new IllegalArgumentException("index is already in the priority queue"); 101 // 元素个数+1 102 n++; 103 // 索引为i的二叉堆位置为n 104 qp[i] = n; 105 // 二叉堆底部插入新元素，值=i 106 pq[n] = i; 107 // 索引i对应的元素赋值 108 keys[i] = key; 109 // 二叉堆中，上浮最后一个元素（小值上浮） 110 swim(n); 111 } 112 113 /** 114 * 返回最小元素的索引 115 * 116 * @return an index associated with a minimum key 117 * @throws NoSuchElementException if this priority queue is empty 118 */ 119 public int minIndex() { 120 if (n == 0) throw new NoSuchElementException("Priority queue underflow"); 121 return pq[1]; 122 } 123 124 /** 125 * 返回最小元素（key） 126 * 127 * @return a minimum key 128 * @throws NoSuchElementException if this priority queue is empty 129 */ 130 public Key minKey() { 131 if (n == 0) throw new NoSuchElementException("Priority queue underflow"); 132 return keys[pq[1]]; 133 } 134 135 /** 136 * 删除最小值key,并返回最小值（优先队列索引） 137 * 138 * @return an index associated with a minimum key 139 * @throws NoSuchElementException if this priority queue is empty 140 */ 141 public int delMin() { 142 if (n == 0) throw new NoSuchElementException("Priority queue underflow"); 143 // pq[1]即为索引最小值 144 int min = pq[1]; 145 // 交换第一个元素和最后一个元素 146 exch(1, n--); 147 // 把新换来的第一个元素下沉 148 sink(1); 149 // 校验下沉后，最后一个元素是最小值 150 assert min == pq[n+1]; 151 // 恢复初始值，-1即代表该元素已删除 152 qp[min] = -1; // delete 153 // 方便垃圾回收 154 keys[min] = null; 155 // 最后一个元素（索引）赋值-1 156 pq[n+1] = -1; // not needed 157 return min; 158 } 159 160 /** 161 * Returns the key associated with index {@code i}. 162 * 163 * @param i the index of the key to return 164 * @return the key associated with index {@code i} 165 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 166 * @throws NoSuchElementException no key is associated with index {@code i} 167 */ 168 public Key keyOf(int i) { 169 validateIndex(i); 170 if (!contains(i)) throw new NoSuchElementException("index is not in the priority queue"); 171 else return keys[i]; 172 } 173 174 /** 175 * Change the key associated with index {@code i} to the specified value. 176 * 177 * @param i the index of the key to change 178 * @param key change the key associated with index {@code i} to this key 179 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 180 * @throws NoSuchElementException no key is associated with index {@code i} 181 */ 182 public void changeKey(int i, Key key) { 183 validateIndex(i); 184 if (!contains(i)) throw new NoSuchElementException("index is not in the priority queue"); 185 keys[i] = key; 186 swim(qp[i]); 187 sink(qp[i]); 188 } 189 190 /** 191 * Change the key associated with index {@code i} to the specified value. 192 * 193 * @param i the index of the key to change 194 * @param key change the key associated with index {@code i} to this key 195 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 196 * @deprecated Replaced by {@code changeKey(int, Key)}. 197 */ 198 @Deprecated 199 public void change(int i, Key key) { 200 changeKey(i, key); 201 } 202 203 /** 204 * 减小索引i对应的值为key 205 * 更新： 206 * 1.元素数组keys[] 207 * 2.小顶二叉堆pq[] 208 * 209 * @param i the index of the key to decrease 210 * @param key decrease the key associated with index {@code i} to this key 211 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 212 * @throws IllegalArgumentException if {@code key >= keyOf(i)} 213 * @throws NoSuchElementException no key is associated with index {@code i} 214 */ 215 public void decreaseKey(int i, Key key) { 216 validateIndex(i); 217 if (!contains(i)) throw new NoSuchElementException("index is not in the priority queue"); 218 // key 值一样，报错 219 if (keys[i].compareTo(key) == 0) 220 throw new IllegalArgumentException("Calling decreaseKey() with a key equal to the key in the priority queue"); 221 // key比当前值大，报错 222 if (keys[i].compareTo(key) < 0) 223 throw new IllegalArgumentException("Calling decreaseKey() with a key strictly greater than the key in the priority queue"); 224 // key比当前值小，把key赋值进去 225 keys[i] = key; 226 // 小值上浮（qp[i]=索引i在二叉堆pq[]中的位置） 227 swim(qp[i]); 228 } 229 230 /** 231 * Increase the key associated with index {@code i} to the specified value. 232 * 233 * @param i the index of the key to increase 234 * @param key increase the key associated with index {@code i} to this key 235 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 236 * @throws IllegalArgumentException if {@code key <= keyOf(i)} 237 * @throws NoSuchElementException no key is associated with index {@code i} 238 */ 239 public void increaseKey(int i, Key key) { 240 validateIndex(i); 241 if (!contains(i)) throw new NoSuchElementException("index is not in the priority queue"); 242 if (keys[i].compareTo(key) == 0) 243 throw new IllegalArgumentException("Calling increaseKey() with a key equal to the key in the priority queue"); 244 if (keys[i].compareTo(key) > 0) 245 throw new IllegalArgumentException("Calling increaseKey() with a key strictly less than the key in the priority queue"); 246 keys[i] = key; 247 sink(qp[i]); 248 } 249 250 /** 251 * Remove the key associated with index {@code i}. 252 * 253 * @param i the index of the key to remove 254 * @throws IllegalArgumentException unless {@code 0 <= i < maxN} 255 * @throws NoSuchElementException no key is associated with index {@code i} 256 */ 257 public void delete(int i) { 258 validateIndex(i); 259 if (!contains(i)) throw new NoSuchElementException("index is not in the priority queue"); 260 int index = qp[i]; 261 exch(index, n--); 262 swim(index); 263 sink(index); 264 keys[i] = null; 265 qp[i] = -1; 266 } 267 268 // throw an IllegalArgumentException if i is an invalid index 269 private void validateIndex(int i) { 270 if (i < 0) throw new IllegalArgumentException("index is negative: " + i); 271 if (i >= maxN) throw new IllegalArgumentException("index >= capacity: " + i); 272 } 273 274 /*************************************************************************** 275 * General helper functions. 276 ***************************************************************************/ 277 private boolean greater(int i, int j) { 278 return keys[pq[i]].compareTo(keys[pq[j]]) > 0; 279 } 280 281 private void exch(int i, int j) { 282 int swap = pq[i]; 283 pq[i] = pq[j]; 284 pq[j] = swap; 285 qp[pq[i]] = i; 286 qp[pq[j]] = j; 287 } 288 289 290 /*************************************************************************** 291 * Heap helper functions. 上浮 292 ***************************************************************************/ 293 private void swim(int k) { 294 // 如果父节点值比当前节点值大，交换，父节点作为当前节点，轮询。即小值上浮。 295 while (k > 1 && greater(k/2, k)) { 296 exch(k, k/2); 297 k = k/2; 298 } 299 } 300 //下沉 301 private void sink(int k) { 302 while (2*k <= n) { 303 int j = 2*k; 304 if (j < n && greater(j, j+1)) j++; 305 if (!greater(k, j)) break; 306 exch(k, j); 307 k = j; 308 } 309 } 310 311 312 /*************************************************************************** 313 * Iterators. 314 ***************************************************************************/ 315 316 /** 317 * Returns an iterator that iterates over the keys on the 318 * priority queue in ascending order. 319 * The iterator doesn't implement {@code remove()} since it's optional. 320 * 321 * @return an iterator that iterates over the keys in ascending order 322 */ 323 @Override 324 public Iterator<Integer> iterator() { return new HeapIterator(); } 325 326 private class HeapIterator implements Iterator<Integer> { 327 // create a new pq 328 private IndexMinPQ<Key> copy; 329 330 // add all elements to copy of heap 331 // takes linear time since already in heap order so no keys move 332 public HeapIterator() { 333 copy = new IndexMinPQ<Key>(pq.length - 1); 334 for (int i = 1; i <= n; i++) 335 copy.insert(pq[i], keys[pq[i]]); 336 } 337 338 @Override 339 public boolean hasNext() { return !copy.isEmpty(); } 340 @Override 341 public void remove() { throw new UnsupportedOperationException(); } 342 343 @Override 344 public Integer next() { 345 if (!hasNext()) throw new NoSuchElementException(); 346 return copy.delMin(); 347 } 348 } 349 350 351 /** 352 * Unit tests the {@code IndexMinPQ} data type. 353 * 354 * @param args the command-line arguments 355 */ 356 public static void main(String[] args) { 357 // insert a bunch of strings 358 String[] strings = { "it", "was", "the", "best", "of", "times", "it", "was", "the", "worst" }; 359 360 IndexMinPQ<String> pq = new IndexMinPQ<String>(strings.length); 361 for (int i = 0; i < strings.length; i++) { 362 pq.insert(i, strings[i]); 363 } 364 365 // delete and print each key 366 while (!pq.isEmpty()) { 367 int i = pq.delMin(); 368 StdOut.println(i + " " + strings[i]); 369 } 370 StdOut.println(); 371 372 // reinsert the same strings 373 for (int i = 0; i < strings.length; i++) { 374 pq.insert(i, strings[i]); 375 } 376 377 // print each key using the iterator 378 for (int i : pq) { 379 StdOut.println(i + " " + strings[i]); 380 } 381 while (!pq.isEmpty()) { 382 pq.delMin(); 383 } 384 385 } 386 }
2.4 Prim算法(基于二叉堆)
1 package study.algorithm.graph; 2 3 import study.algorithm.base.*; 4 5 import java.util.Arrays; 6 7 /*** 8 * @Description 使用Prim算法计算一棵最小生成树 27 * 28 * @author denny.zhang 29 * @date 2020/5/26 9:50 上午 30 */ 31 public class PrimMST { 32 private static final double FLOATING_POINT_EPSILON = 1E-12; 33 34 /** 35 * 顶点索引，树顶点到非树顶点的最短边(距离树最近的边) 36 */ 37 private Edge[] edgeTo; 38 /** 39 * 顶点索引，最短边的权重 40 */ 41 private double[] distTo; 42 /** 43 * 顶点索引，标记顶点是否在最小生成树中 44 */ 45 private boolean[] marked; 46 /** 47 * 有效的横切边（索引最小优先队列，索引为顶点v,值pq[v]=edgeTo[v].weight()=distTo[v]） 48 */ 49 private IndexMinPQ<Double> pq; 50 51 /** 52 * 计算一个边加权图的最小生成树 53 * @param G the edge-weighted graph 54 */ 55 public PrimMST(EdgeWeightedGraph G) { 56 // 初始化3个顶点索引数组 57 edgeTo = new Edge[G.V()]; 58 distTo = new double[G.V()]; 59 marked = new boolean[G.V()]; 60 // 初始化：顶点索引最小优先队列 61 pq = new IndexMinPQ<Double>(G.V()); 62 for (int v = 0; v < G.V(); v++) { 63 // 初始化为无穷大 64 distTo[v] = Double.POSITIVE_INFINITY; 65 } 66 // 遍历顶点数 67 for (int v = 0; v < G.V(); v++) 68 { 69 // 如果顶点0开始能全进树，那就一次搞定 70 StdOut.println("v="+ v+",marked[v]="+marked[v]); 71 // 如果没进树 72 if (!marked[v]) { 73 StdOut.println("v="+v+"，执行prim"); 74 // 最小生成树 75 prim(G, v); 76 } 77 } 78 79 // 校验，可省略 80 assert check(G); 81 } 82 83 84 /** 85 * 从顶点s开始生成图G 86 * @param G 87 * @param s 88 */ 89 private void prim(EdgeWeightedGraph G, int s) { 90 // 顶点s的权重=0 91 distTo[s] = 0.0; 92 // 顶点s进队列，key为索引，value为边权重 93 pq.insert(s, distTo[s]); 94 StdOut.println("顶点s进队列: s="+ s+",distTo[s]="+distTo[s]); 95 StdOut.println("pq="+ Arrays.toString(pq.keys)); 96 97 // 循环 98 while (!pq.isEmpty()) { 99 // 取出最小权重边的顶点（最近的顶点） 100 int v = pq.delMin(); 101 // 添加到树中 102 scan(G, v); 103 } 104 } 105 106 /** 107 * 将顶点V添加到树中，更新数据 108 * @param G 109 * @param v 110 */ 111 private void scan(EdgeWeightedGraph G, int v) { 112 StdOut.println("v="+ v+",进树"); 113 // 标记 进树 114 marked[v] = true; 115 // 遍历顶点v的邻接边 116 for (Edge e : G.adj(v)) { 117 StdOut.println("遍历顶点v的邻接边：v="+ v+",e="+e.toString()); 118 // 另一个顶点 119 int w =
版权声明：本文来源CSDN，感谢博主原创文章，遵循 CC 4.0 by-sa 版权协议，转载请附上原文出处链接和本声明。
原文链接：https://blog.csdn.net/FAw67J7/article/details/106485978
站方申明：本站部分内容来自社区用户分享，若涉及侵权，请联系站方删除。
原文链接：https://blog.csdn.net/FAw67J7/article/details/106485978
站方申明：本站部分内容来自社区用户分享，若涉及侵权，请联系站方删除。
- 发表于 2020-06-28 02:46:28
- 阅读 ( 848 )
- 分类：算法
