[英]Why recursion in java is so slow compared to .net?
我編寫了一種算法來查找兩點之間的所有可能路徑。 我在c＃.net和java中都編寫了算法。 該算法產生大約40,000條可能的路徑。 令我驚訝的是，C＃.net中的算法僅用了13分鍾即可完成（在i3處理器中），而Java甚至無法在4小時內完成。
遞歸在Java中可能會很慢，但是我不這么認為。
我有什么想念的嗎？
好的，這里是必要條件，迷宮是一個二維數組，其中包含數字0,1,2和3。0表示步行點，1表示障礙物，2表示起點，3表示出口點。 該程序是通過僅觸摸全0一次找到從點2到3的所有路徑。 我們無法超越1，因為它們是障礙。
public class Maze
{
//Represents each point in the Maze.
private class Node
{
private int val;
public Node(int num)
{
val = num;
switch (num)
{
case 1: IsPit = true;
break;
case 2: IsEntranceNode = true;
break;
case 3:
IsExitNode = true;
break;
}
}
public string Name { get; set; }
public bool IsPit { get; private set; }
public bool IsVisited { get; set; }
public bool IsExitNode { get; private set; }
public bool IsEntranceNode { get; private set; }
public Node UpNode { get; set; }
public Node DownNode { get; set; }
public Node LeftNode { get; set; }
public Node RightNode { get; set; }
}
#endregion
//Number of pits in the maze.
private int numberOfPits;
//Stack traces the path traversed.
private Stack<String> pathTraversed;
//Represents the maze.All Nodes are wired to its adjacent nodes.
private Node[,] Maze { get; set; }
//Number of paths found.
private int pathCount = 1;
//displays a set.
static void printSet(int[,] set)
{
Console.Write(" \t");
for (var i = 0; i < set.GetLength(1); i++)
Console.Write(i + " ");
Console.WriteLine("\n--------------------------");
for (var i = 0; i < set.GetLength(0); i++)
{
Console.Write((char)(65 + i) + " | \t");
for (var j = 0; j < set.GetLength(1); j++)
Console.Write(set[i, j] + " ");
Console.WriteLine();
}
}
static void Main(string[] args)
{
int[,] set2 = new int[7, 7] { { 2, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 3, 1, 1 }
};
Console.WriteLine("\n\nSet 2");
Console.WriteLine("=====");
printSet(set2);
Console.WriteLine("\nPath found");
Console.WriteLine("==========");
startTime = DateTime.Now;
new Maze().AddPoints(set2).FindPaths();
endTime = DateTime.Now;
difference = endTime.Subtract(startTime).TotalMinutes;
Console.WriteLine("Time taken in mins : " + difference);
Console.WriteLine("\nPress any key to exit");
Console.ReadKey();
}
//Adds points to the maze.Transforms a matrix to 2 dimensional connected nodes.
public Program AddPoints(int[,] matrix)
{
Maze = new Node[matrix.GetLength(0), matrix.GetLength(1)];
for (int i = 0; i < matrix.GetLength(0); i++)
{
for (var j = 0; j < matrix.GetLength(1); j++)
{
Node node = Maze[i, j];
if (node == null)
{
node = new Node(matrix[i, j]);
Maze[i, j] = node;
}
if (!(i - 1 < 0))
Maze[i - 1, j].DownNode = node;
if (!(i + 1 > matrix.GetLength(0) - 1))
{
var downNode = Maze[i + 1, j];
if (downNode == null)
Maze[i + 1, j] = new Node(matrix[i + 1, j]);
Maze[i + 1, j].UpNode = node;
}
if (!(j - 1 < 0))
Maze[i, j - 1].RightNode = node;
if (!(j + 1 > matrix.GetLength(1) - 1))
{
var leftNode = Maze[i, j + 1];
if (leftNode == null)
Maze[i, j + 1] = new Node(matrix[i, j + 1]);
Maze[i, j + 1].LeftNode = node;
}
node.Name = ((char)(65 + i)).ToString() + j.ToString();
if (node.IsPit)
numberOfPits++;
}
}
return this;
}
//Finds path between the start node and end node.
public void FindPaths()
{
var startNode = Maze[0, 0];
if (!startNode.IsEntranceNode)
throw new Exception("Node is not starting node.");
pathTraversed = new Stack<string>();
pathTraversed.Push(startNode.Name);
Traverse(startNode.RightNode);
Traverse(startNode.DownNode);
Traverse(startNode.LeftNode);
Traverse(startNode.UpNode);
}
//Traverses a node.
private void Traverse(Node node)
{
if (node == null)
return;
if (node.IsEntranceNode || node.IsPit || node.IsVisited)
return;
if (node.IsExitNode)
{
pathTraversed.Push(node.Name);
if (pathTraversed.Count == Maze.Length - numberOfPits)
{
var msg = "Path " + pathCount++ + " : " + string.Join("->", pathTraversed.Reverse());
Console.WriteLine(msg);
}
pathTraversed.Pop();
return;
}
pathTraversed.Push(node.Name);
node.IsVisited = true;
Traverse(node.RightNode); //
Traverse(node.DownNode); // Move to Next Node
Traverse(node.LeftNode); //
Traverse(node.UpNode); //
if (node.Name != pathTraversed.Peek())
throw new Exception("Error in Logic.");
node.IsVisited = false;
pathTraversed.Pop();
}
}
public class PathFinder
{
public static void main(String[] args){
int[][] set1 = new int[][] { { 2, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 0, 0, 0 },
{ 0, 0, 0, 0, 3, 1, 1 }};
ObservableStack<String> observableStack = new ObservableStack<String>();
StackObserver stackObserver = new StackObserver(plotter);
Maze maze = new Maze(set1,(IPathFinderStack<String>)observableStack);
maze.findPaths();
}
}
public class Maze {
private Node[][] nodes;
IPathFinderStack<String> pathTraversed;
private int numberOfPits = 0;
private int result = 0;
public Node[][] getVector(){
return nodes;
}
public Maze(int[][] matrix ,IPathFinderStack<String> stack){
pathTraversed = stack;
addNodes(matrix);
}
public void addNodes(int[][] matrix) {
int rows = matrix.length;
int cols = matrix[0].length;
nodes = new Node[rows][cols];
for (int i = 0; i < rows; i++)
{
for (int j = 0; j < cols; j++)
{
Node node = nodes[i][j];
if (node == null){
node = new Node(matrix[i][j]);
nodes[i][j] = node;
}
if (!(i - 1 < 0))
nodes[i-1][j].setDownNode(node);
if (!((i + 1) > rows - 1))
{
Node downNode = nodes[i+1][j];
if (downNode == null)
nodes[i + 1][j] = new Node(matrix[i+1][j]);
nodes[i + 1][j].setUpNode(node);
}
if (!(j - 1 < 0))
nodes[i][j-1].setRightNode(node);
if (!(j + 1 > cols - 1)){
Node leftNode = nodes[i][j+1];
if (leftNode == null)
nodes[i][j+1] = new Node(matrix[i][j+1]);
nodes[i][j+1].setLeftNode(node);
}
String name = new String(new char[] {(char) (65 + i), (char)(48 + j)});
node.setName(name);
if (node.isPit())
this.numberOfPits ++;
}
}
}
public void findPaths(){
Node startNode = nodes[0][0];
pathTraversed.push(startNode.getName());
Traverse(startNode.getRightNode());
Traverse(startNode.getDownNode());
Traverse(startNode.getLeftNode());
Traverse(startNode.getUpNode());
}
private void Traverse(Node node){
if (node == null)
return ;
if (node.isEntrance() || node.isVisited() || node.isPit())
return ;
if (node.isExit()){
pathTraversed.push(node.getName());
if (pathTraversed.getLength() == (nodes[0].length * nodes.length) - numberOfPits){
String msg = "Path found" + (++result);
Logger.Log(msg);
Logger.Log(pathTraversed.toString());
}
pathTraversed.pop();
return;
}
pathTraversed.push(node.getName());
node.setVisited(true);
Traverse(node.getRightNode());
Traverse(node.getDownNode());
Traverse(node.getLeftNode());
Traverse(node.getUpNode());
node.setVisited(false);
pathTraversed.pop();
}
}
public class Node
{
private String name;
private int val;
private boolean isPit;
private boolean isVisited;
private boolean isEntrance;
private boolean isExit;
private Node downNode;
private Node upNode;
private Node leftNode;
private Node rightNode;
public Node(int val){
this.val = val;
this.isPit = (val == 1);
this.isEntrance = (val == 2);
this.isExit = (val == 3);
this.isVisited = false;
}
public String getName(){
return this.name;
}
public void setName (String name) {
this.name = name;
}
public boolean isPit(){
return isPit;
}
public boolean isVisited(){
return isVisited;
}
public void setVisited(boolean isVisited) {
this.isVisited = isVisited;
}
public boolean isEntrance(){
return isEntrance;
}
public boolean isExit(){
return isExit;
}
public Node getDownNode(){
return downNode;
}
public void setDownNode(Node node){
downNode = node;
}
public Node getUpNode(){
return upNode;
}
public void setUpNode(Node node){
upNode = node;
}
public Node getLeftNode(){
return leftNode;
}
public void setLeftNode(Node node){
leftNode = node;
}
public Node getRightNode(){
return rightNode;
}
public void setRightNode(Node node) {
rightNode = node;
}
}
ObservableStack是由數組組成的普通堆棧。 我可以觀察到將其連接到applet以動畫化遍歷。
我希望我提供了足夠的細節。
我編寫了不同的代碼，並獲得了更快的速度。
import java.util.ArrayList;
import java.util.List;
public class PathFinder {
public static void main(String[] args) {
int[][] set1 = new int[][]{{2, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 3, 1, 1}};
long start = System.nanoTime();
Maze maze = new Maze(set1);
List<String> paths = new ArrayList<String>();
maze.findPaths(0, 0, new StringBuilder(), paths);
System.out.printf("Paths found %,d%n", paths.size());
long time = System.nanoTime() - start;
System.out.printf("Took %,.1f seconds%n", time / 1e9);
}
}
class Node {
final int num;
private boolean traversed;
private final String name;
Node(int num, String name) {
this.num = num;
this.name = name;
}
public boolean isTraversed() {
return traversed;
}
public void setTraversed(boolean traversed) {
this.traversed = traversed;
}
public String getName() {
return name;
}
public boolean isExit() {
return num == 3;
}
public boolean isPit() {
return num == 1;
}
}
class Maze {
private final Node[][] nodes;
private final int rows, cols;
private final int nonPits;
public Maze(int[][] matrix) {
rows = matrix.length;
cols = matrix[0].length;
nodes = new Node[rows][cols];
int nonPits = 0;
for (int i = 0; i < rows; i++)
for (int j = 0; j < cols; j++) {
nodes[i][j] = new Node(matrix[i][j], "" + (char) ('A' + i) + (char) ('0' + j));
if (!nodes[i][j].isPit())
nonPits++;
}
this.nonPits = nonPits;
}
public void findPaths(int x, int y, StringBuilder path, List<String> paths) {
if (x < 0 || x >= cols) return;
if (y < 0 || y >= rows) return;
Node node = nodes[y][x];
if (node.isTraversed() || node.isPit()) return;
node.setTraversed(true);
int length = path.length();
path.append(node.getName());
if (node.isExit()) {
if (path.length() == nonPits*2) {
paths.add(path.toString());
if (paths.size() % 10000 == 0)
System.out.printf("... found %,d paths%n", paths.size());
}
} else {
findPaths(x - 1, y, path, paths);
findPaths(x + 1, y, path, paths);
findPaths(x, y - 1, path, paths);
findPaths(x, y + 1, path, paths);
}
path.setLength(length);
node.setTraversed(false);
}
}
版畫
Paths found 40,616
Took 81.1 seconds
記錄路徑花費了大量時間。 如果僅計算路徑（需要相同數量的遞歸）
import java.util.ArrayList;
import java.util.LinkedHashSet;
import java.util.List;
import java.util.Set;
public class PathFinder {
public static void main(String[] args) {
int[][] set1 = new int[][]{{2, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 0, 0, 0},
{0, 0, 0, 0, 3, 1, 1}};
long start = System.nanoTime();
Maze maze = new Maze(set1);
long[] count = {0L};
maze.findPaths(0, 0, 1, count);
System.out.printf("Paths found %,d%n", count[0]);
long time = System.nanoTime() - start;
System.out.printf("Took %,.1f seconds%n", time / 1e9);
}
}
class Node {
final int num;
private boolean traversed;
private final String name;
Node(int num, String name) {
this.num = num;
this.name = name;
}
public boolean isTraversed() {
return traversed;
}
public void setTraversed(boolean traversed) {
this.traversed = traversed;
}
public String getName() {
return name;
}
public boolean isExit() {
return num == 3;
}
public boolean isPit() {
return num == 1;
}
}
class Maze {
private final Node[][] nodes;
private final int rows, cols;
private final int nonPits;
public Maze(int[][] matrix) {
rows = matrix.length;
cols = matrix[0].length;
nodes = new Node[rows][cols];
int nonPits = 0;
for (int i = 0; i < rows; i++)
for (int j = 0; j < cols; j++) {
nodes[i][j] = new Node(matrix[i][j], "" + (char) ('A' + i) + (char) ('0' + j));
if (!nodes[i][j].isPit())
nonPits++;
}
this.nonPits = nonPits;
}
public void findPaths(int x, int y, int depth, long[] count) {
if (x < 0 || x >= cols) return;
if (y < 0 || y >= rows) return;
Node node = nodes[y][x];
if (node.isTraversed() || node.isPit()) return;
node.setTraversed(true);
if (node.isExit()) {
if (depth == nonPits) {
count[0]++;
if (count[0] % 10000 == 0)
System.out.printf("... found %,d paths%n", count[0]);
}
} else {
findPaths(x - 1, y, depth + 1, count);
findPaths(x + 1, y, depth + 1, count);
findPaths(x, y - 1, depth + 1, count);
findPaths(x, y + 1, depth + 1, count);
}
node.setTraversed(false);
}
}
版畫
Paths found 40,616
Took 58.5 seconds
與PrincipalSearcher相比，為什么DirectorySearcher這么慢？
[英]Why is DirectorySearcher so slow when compared to PrincipalSearcher?
與 ObjectContext 相比，為什么在 EF 4.1 中插入實體如此緩慢？
[英]Why is inserting entities in EF 4.1 so slow compared to ObjectContext?
為什么與c＃字典插入相比，std :: map插入這么慢？
[英]Why is the std::map insertion so slow compared to c# Dictionary insertion?
為什么在 .NET 4.6 中更改 WPF DataGrid ItemsSource 如此之慢？
[英]Why is changing WPF DataGrid ItemsSource so extremely slow in .NET 4.6?
ASP.net MVC：為什么 AreaRegistration.RegisterAllAreas() 這么慢？
[英]ASP.net MVC: why is AreaRegistration.RegisterAllAreas() so slow?
聲明:本站的技術帖子網頁，遵循CC BY-SA 4.0協議，如果您需要轉載，請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.