我正在研究一个问题,试图在典型的XY图(也就是网格)中找到所谓的Snake序列。 Snake序列被定义为一个数字序列,其中每个新数字(只能位于当前数字的右侧或下方)是加号或减号。例如,如果您位于图表的中心,则可以向右移动(如果该数字为+或 - 1)或向下移动(如果该数字为+或 - 1)。该问题的目标是通过图形找到最长的路径(也称为蛇序列)(请记住,您只能绘制到当前单元格值为+ - 1的新单元格的路径。)
因此,对于以下XY图,最长的蛇序列为:9, 8, 7, 6, 5, 6, 7
9, 6, 5, 2
8, 7, 6, 5
7, 3, 1, 6
1, 1, 1, 7
以下是我的代码,似乎无效。
问题:您如何解决上述问题? (我提供的代码显示了我到目前为止的内容,但它不起作用)
import java.util.ArrayList;
public class SnakeSequence {
private final int maxX = 3;
private final int maxY = 3;
private final int[][] board = new int[][]{
{1, 2, 3, 4},
{2, 1, -1, 5},
{3, 0, -1, 6},
{6, 2, 1, 7}
};
private ArrayList<Integer> findSequence(int xPos,
int yPos, ArrayList<Integer> currentPath) {
currentPath.add(board[yPos][xPos]);
ArrayList<Integer> pathRight = new ArrayList<Integer>(currentPath);
ArrayList<Integer> pathDown = new ArrayList<Integer>(currentPath);
if (xPos < maxX || yPos < maxY) {
if (yPos < maxY && (board[yPos + 1][xPos] + 1 == board[yPos][xPos] ||
board[yPos + 1][xPos] - 1 == board[yPos][xPos])) {
pathDown = findSequence(xPos, yPos + 1, currentPath);
}
if (xPos < maxX && (board[yPos][xPos + 1] + 1 == board[yPos][xPos] ||
board[yPos][xPos + 1] - 1 == board[yPos][xPos])) {
pathRight = findSequence(xPos + 1, yPos, currentPath);
}
if (pathDown.size() > pathRight.size()) {
return pathDown;
} else {
return pathRight;
}
}
return currentPath;
}
private void getSequence() {
ArrayList<Integer> currentPath = new ArrayList<Integer>();
ArrayList<Integer> result;
result = findSequence(0, 0, currentPath);
for (int i = 0; i < result.size(); i++) {
System.out.println(result.get(i));
}
}
public static void main(String[] args) {
SnakeSequence sequence = new SnakeSequence();
sequence.getSequence();
}
}
答案 0 :(得分:2)
您可以将表格想象为oriented graph,那么您的问题就是找到longest path。
幸运的是,只允许向下和向右移动,因此您的图表为acyclic,因此您可以使用critical path method等算法。
这就是图表的样子:
但是,您希望找到任何两个单元格之间的最长路径。为此,我将为每个单元格计算从该单元格开始的最长路径。它与您的行为相似,但您计算的次数更多。考虑一下:
6 -> 5
| |
v v
7 -> 6
同时5
和7
计算从6
向右下方的路径有多长,这是无用的重复计算。在最坏的情况下,这可能导致指数时间消耗,而问题可以在线性时间内解决!
此外,无法保证最长路径将从(0,0)
开始。
(一种可能)解决方案:
计算每个单元格的最长路径,从右下角到左上角。在每个细胞处......记住那个细胞最长的路径有多长,并且从那个细胞开始走路。 (我将根据它上面的细胞数来测量路径长度)。例如,对于您的格鲁吉亚唯一的8
,我们会记住[length=8, direction=right]
。
为什么如此复杂?因为如果我们知道细胞向右和向下的最长路径,现在在单元格中计算最长路径非常容易。示例(我编写):
现在2
的正确数据为[length=4, direction=down]
,因为无法从2
转到4
。
您还可以保持全球最长的路径并启动它。计算完成后,只需从那个开始通过direction
走最长路径并写下数字,位置或任何你需要的路径。
答案 1 :(得分:1)
为我的Java道歉(我主要是一个c#程序员),但这是一个解决方案。我将从算法中发现蛇的算法(实现接口ISnakeProcessor)分离出来,处理每个蛇。通过这种方式,您可以增强代码,例如,使用最大值的值来收集蛇,或者通过添加更多的ISnakeProcessor类来收集所有最长的蛇,以防有多个蛇。
import java.util.*;
import java.lang.*;
class Rextester
{
public static void main(String args[])
{
SnakeSequence sequence = new SnakeSequence();
sequence.getSequence();
}
}
interface ISnakeProcessor
{
void process(List<Pair<Integer, Integer>> snake);
}
class SnakeSequence {
private final int[][] board;
public SnakeSequence()
{
this(new int[][]{
{1, 2, 3, 4},
{2, 1, -1, 5},
{3, 0, -1, 6},
{6, 2, 1, 7}
});
}
public SnakeSequence(int[][] board)
{
this.board = board;
}
public boolean isValid(int iRow, int iCol)
{
if (iRow < 0 || iRow >= board.length)
return false;
if (iCol < 0 || iCol >= board[iRow].length)
return false;
return true;
}
private boolean continuesInRow(int iRow, int iCol)
{
if (!isValid(iRow, iCol) || !isValid(iRow+1, iCol))
return false;
int myVal = board[iRow][iCol];
if (board[iRow+1][iCol] == myVal - 1 || board[iRow+1][iCol] == myVal + 1)
return true;
return false;
}
private boolean continuesInCol(int iRow, int iCol)
{
if (!isValid(iRow, iCol) || !isValid(iRow, iCol+1))
return false;
int myVal = board[iRow][iCol];
if (board[iRow][iCol+1] == myVal - 1 || board[iRow][iCol+1] == myVal + 1)
return true;
return false;
}
private boolean isHead(int iRow, int iCol)
{
if (!isValid(iRow, iCol))
return false;
if (isValid(iRow-1, iCol) && continuesInRow(iRow-1, iCol))
return false;
if (isValid(iRow, iCol-1) && continuesInRow(iRow, iCol-1))
return false;
return true;
}
private boolean isTail(int iRow, int iCol)
{
if (!isValid(iRow, iCol))
return false;
if (continuesInRow(iRow, iCol))
return false;
if (continuesInCol(iRow, iCol))
return false;
return true;
}
private void testHead()
{
System.out.println("Dumping list of heads");
for (int iRow = 0; iRow < board.length; iRow++)
{
for (int iCol = 0; iCol < board[iRow].length; iCol++)
{
boolean head = isHead(iRow, iCol);
boolean tail = isTail(iRow, iCol);
if (head && tail)
System.out.print(" B");
else if (head)
System.out.print(" H");
else if (tail)
System.out.print(" T");
else
System.out.print(" -");
}
System.out.println("");
}
}
private void walkSnake(ISnakeProcessor processor, int iRow, int iCol, ArrayList<Pair<Integer, Integer>> snake)
{
snake.add(new Pair<Integer, Integer>(iRow, iCol));
boolean isTail = true;
if (continuesInRow(iRow, iCol))
{
walkSnake(processor, iRow+1, iCol, snake);
isTail = false;
}
if (continuesInCol(iRow, iCol))
{
walkSnake(processor, iRow, iCol+1, snake);
isTail = false;
}
if (isTail)
{
processor.process(snake);
}
snake.remove(snake.size() - 1);
}
private void walkSnakes(ISnakeProcessor processor)
{
ArrayList<Pair<Integer, Integer>> snake = new ArrayList<Pair<Integer, Integer>>();
for (int iRow = 0; iRow < board.length; iRow++)
for (int iCol = 0; iCol < board[iRow].length; iCol++)
if (isHead(iRow, iCol))
walkSnake(processor, iRow, iCol, snake);
}
class LongestSnakeFinder implements ISnakeProcessor
{
private final SnakeSequence parent;
ArrayList<Pair<Integer, Integer>> longest = new ArrayList<Pair<Integer, Integer>>();
public LongestSnakeFinder(SnakeSequence parent)
{
this.parent = parent;
}
public void process(List<Pair<Integer, Integer>> snake)
{
if (snake.size() > longest.size())
{
longest.clear();
longest.addAll(snake);
}
}
public void dumpLongest()
{
System.out.format("The first encountered longest snake has length %d:\n", longest.size());
for (int i = 0; i < longest.size(); i++)
{
int iRow = longest.get(i).getFirst();
int iCol = longest.get(i).getSecond();
System.out.format(" (%d,%d): %d\n", iRow, iCol, parent.getValue(iRow, iCol));
}
}
}
public int getNRows() { return board.length; }
public int getNCols(int iRow) { return board[iRow].length; }
public int getValue(int iRow, int iCol) { return board[iRow][iCol]; }
public void getSequence() {
testHead();
LongestSnakeFinder finder = new LongestSnakeFinder(this);
walkSnakes(finder);
finder.dumpLongest();
}
}
class Pair<F, S> {
private F first; //first member of pair
private S second; //second member of pair
public Pair(F first, S second) {
this.first = first;
this.second = second;
}
public F getFirst() {
return first;
}
public S getSecond() {
return second;
}
}
此处运行示例:http://rextester.com/AKUFNL43897 更新 - 清理了一点代码。新示例在此处运行:http://rextester.com/AVOAIY11573
而且,输出:
Dumping list of heads
H - - -
- - B -
T - T -
B H T T
The first encountered longest snake has length 7:
(0,0): 1
(0,1): 2
(0,2): 3
(0,3): 4
(1,3): 5
(2,3): 6
(3,3): 7
这是你想要的吗?
答案 2 :(得分:0)
这是一种纠正解决方案并避免在每一步中复制路径的简单方法
import java.util.ArrayList;
import java.util.Collections;
public class SnakeSequence {
private final int maxX = 3;
private final int maxY = 3;
private final int[][] board = new int[][]{
{1, 2, 3, 4},
{2, 1, -1, 5},
{3, 0, -1, 6},
{6, 2, 1, 7}
};
private ArrayList<Integer> findSequence(int xPos,
int yPos) {
ArrayList<Integer> pathRight = new ArrayList<Integer>();
ArrayList<Integer> pathDown = new ArrayList<Integer>();
if (yPos < maxY && (board[yPos + 1][xPos] + 1 == board[yPos][xPos] ||
board[yPos + 1][xPos] - 1 == board[yPos][xPos])) {
pathDown = findSequence(xPos, yPos + 1);
}
if (xPos < maxX && (board[yPos][xPos + 1] + 1 == board[yPos][xPos] ||
board[yPos][xPos + 1] - 1 == board[yPos][xPos])) {
pathRight = findSequence(xPos + 1, yPos);
}
ArrayList<Integer> ans;
if (pathDown.size() > pathRight.size()) {
ans = pathDown;
} else {
ans = pathRight;
}
ans.add(board[yPos][xPos]);
return ans;
}
private void getSequence() {
ArrayList<Integer> result;
result = findSequence(0, 0);
Collections.reverse(result);
for (int i = 0; i < result.size(); i++) {
System.out.println(result.get(i));
}
}
public static void main(String[] args) {
SnakeSequence sequence = new SnakeSequence();
sequence.getSequence();
}
}
但是这样,对于大型数组,它可以更快地工作,因为每次在递归期间访问相同的数字时都不会重新计算最长的路径。实际上,在这个版本中,每个号码最多访问两次。它通过为每个节点保存最佳解决方案来实现。单独存储路径和长度不允许在不需要时复制路径。
import java.util.ArrayList;
import java.util.Collections;
public class SnakeSequence {
private final int maxX = 3;
private final int maxY = 3;
private final int[][] board = new int[][]{
{1, 2, 3, 4},
{2, 3, -1, 5},
{3, 2, -1, 6},
{6, 1, 2, 3}
};
int[][] pathLength;
ArrayList<ArrayList<ArrayList<Integer>>> paths;
private ArrayList<Integer> findSequence(int xPos,
int yPos) {
if(pathLength[yPos][xPos] >= 0)
{
ArrayList<Integer> ans = new ArrayList<Integer>();
int length = pathLength[yPos][xPos];
ArrayList<Integer> path = paths.get(yPos).get(xPos);
for(int i = 0; i < length; i++)
ans.add(path.get(i));
return ans;
}
ArrayList<Integer> pathRight = new ArrayList<Integer>();
ArrayList<Integer> pathDown = new ArrayList<Integer>();
if (yPos < maxY && (board[yPos + 1][xPos] + 1 == board[yPos][xPos] ||
board[yPos + 1][xPos] - 1 == board[yPos][xPos])) {
pathDown = findSequence(xPos, yPos + 1);
}
if (xPos < maxX && (board[yPos][xPos + 1] + 1 == board[yPos][xPos] ||
board[yPos][xPos + 1] - 1 == board[yPos][xPos])) {
pathRight = findSequence(xPos + 1, yPos);
}
ArrayList<Integer> ans;
if (pathDown.size() > pathRight.size()) {
ans = pathDown;
} else {
ans = pathRight;
}
ans.add(board[yPos][xPos]);
paths.get(yPos).set(xPos,ans);
pathLength[yPos][xPos] = ans.size();
return ans;
}
private void getSequence() {
ArrayList<Integer> result;
pathLength = new int[maxX + 1][maxY + 1];
paths = new ArrayList<ArrayList<ArrayList<Integer>>>();
for(int y = 0; y <= maxY; y++)
{
ArrayList<ArrayList<Integer>> line = new ArrayList<ArrayList<Integer>>();
for(int x = 0; x <= maxX; x++)
{
line.add(null);
pathLength[y][x] = -1;
}
paths.add(line);
}
result = findSequence(0, 0);
Collections.reverse(result);
for (int i = 0; i < result.size(); i++) {
System.out.println(result.get(i));
}
}
public static void main(String[] args) {
SnakeSequence sequence = new SnakeSequence();
sequence.getSequence();
}
}
答案 3 :(得分:0)
简单递归解决方案:
import java.util.ArrayList;
import java.util.List;
public class MaximumLengthSnakeSequence {
static int max = -1;
static List<Integer> maxListTemp = new ArrayList<>();
public static void main(String args[]) {
int count = 0;
int n = 4;
int m = 4;
int mat[][] = { { 9, 6, 5, 2 }, { 8, 7, 6, 5 }, { 7, 3, 1, 6 }, { 1, 1, 1, 7 }, };
List<Integer> maxList = new ArrayList<>();
for (int i = 0; i < n; i++) {
for (int j = 0; j < m; j++) {
List<Integer> list = new ArrayList<>();
list.add(mat[i][j]);
List<Integer> testList = recur(i, j, count, mat, n, m, list);
if (maxList.size() < testList.size()) {
maxList = new ArrayList<>(testList);
}
maxListTemp.clear();
}
}
System.out.println("max is " + maxList);
}
static List<Integer> recur(int i, int j, int count, int mat[][], int n, int m, List<Integer> list) {
int curData = mat[i][j];
int rightData = 0;
int downData = 0;
if (j + 1 < n && i < m) {
rightData = mat[i][j + 1];
if (Math.abs(curData - rightData) == 1) {
list.add(rightData);
recur(i, j + 1, count + 1, mat, n, m, list);
list.remove(list.size() - 1);
}
}
if (count > max) {
max = count;
}
if (maxListTemp.size() < list.size()) {
maxListTemp = new ArrayList<>(list);
}
if (i + 1 < m && j < n) {
downData = mat[i + 1][j];
if (Math.abs(curData - downData) == 1) {
list.add(downData);
recur(i + 1, j, count + 1, mat, n, m, list);
list.remove(list.size() - 1);
}
}
return maxListTemp;
}
}