受Raymond Chen帖子的启发，假设你有一个4x4二维数组，写一个函数使它旋转90度。Raymond链接到伪代码中的解决方案，但我想看到一些真实的东西。
[1][2][3][4]
[5][6][7][8]
[9][0][1][2]
[3][4][5][6]
就变成:
[3][9][5][1]
[4][0][6][2]
[5][1][7][3]
[6][2][8][4]
更新:Nick的答案是最直接的，但是有没有比n²更好的方法呢?如果矩阵是10000x10000呢?
当前回答
虽然旋转数据可能是必要的(也许是为了更新物理存储的表示)，但在数组访问上添加一层间接层(也许是一个接口)会变得更简单，可能更性能:
interface IReadableMatrix
{
int GetValue(int x, int y);
}
如果你的矩阵已经实现了这个接口，那么它可以通过这样一个装饰器类来旋转:
class RotatedMatrix : IReadableMatrix
{
private readonly IReadableMatrix _baseMatrix;
public RotatedMatrix(IReadableMatrix baseMatrix)
{
_baseMatrix = baseMatrix;
}
int GetValue(int x, int y)
{
// transpose x and y dimensions
return _baseMatrix(y, x);
}
}
旋转+90/-90/180度，水平/垂直翻转和缩放都可以以这种方式实现。
Performance would need to be measured in your specific scenario. However the O(n^2) operation has now been replaced with an O(1) call. It's a virtual method call which is slower than direct array access, so it depends upon how frequently the rotated array is used after rotation. If it's used once, then this approach would definitely win. If it's rotated then used in a long-running system for days, then in-place rotation might perform better. It also depends whether you can accept the up-front cost.
与所有性能问题一样，测量，测量，测量!
其他回答
这里有大量的好代码，但我只是想以几何形式展示，这样你就能更好地理解代码逻辑。以下是我的处理方法。
首先，不要把这和换位相混淆，换位是很容易的。
基本的想法是把它当作层，我们一次旋转一个层。
假设我们有一辆4x4
1 2 3 4
5 6 7 8
9 10 11 12
13 14 15 16
当我们顺时针旋转90度，我们得到
13 9 5 1
14 10 6 2
15 11 7 3
16 12 8 4
我们来分解它，首先旋转这四个角
1 4
13 16
然后我们旋转下面这个有点歪斜的菱形
2
8
9
15
然后是第二个斜菱形
3
5
12
14
这就搞定了外缘基本上我们一次做一个壳层直到
最后是中间的方块(如果是奇数则是最后一个不动的元素)
6 7
10 11
现在我们来算出每一层的指标，假设我们总是在最外层工作，我们正在做
[0,0] -> [0,n-1], [0,n-1] -> [n-1,n-1], [n-1,n-1] -> [n-1,0], and [n-1,0] -> [0,0]
[0,1] -> [1,n-1], [1,n-2] -> [n-1,n-2], [n-1,n-2] -> [n-2,0], and [n-2,0] -> [0,1]
[0,2] -> [2,n-2], [2,n-2] -> [n-1,n-3], [n-1,n-3] -> [n-3,0], and [n-3,0] -> [0,2]
等等等等 直到我们走到边缘的一半
所以总的来说模式是
[0,i] -> [i,n-i], [i,n-i] -> [n-1,n-(i+1)], [n-1,n-(i+1)] -> [n-(i+1),0], and [n-(i+1),0] to [0,i]
我的c#示例代码的伟大算法发送@dimple:
/* Author: Dudi,
* http://www.tutorialspoint.com/compile_csharp_online.php?PID=0Bw_CjBb95KQMYm5qU3VjVGNuZFU */
using System.IO;
using System;
class Program
{
static void Main()
{
Console.WriteLine("Rotating this matrix by 90+ degree:");
int[,] values=new int[3,3]{{1,2,3}, {4,5,6}, {7,8,9}};
//int[,] values=new int[4,4]{{101,102,103, 104}, {105,106, 107,108}, {109, 110, 111, 112}, {113, 114, 115, 116}};
print2dArray(ref values);
transpose2dArray(ref values);
//print2dArray(ref values);
reverse2dArray(ref values);
Console.WriteLine("Output:");
print2dArray(ref values);
}
static void print2dArray(ref int[,] matrix){
int nLen = matrix.GetLength(0);
int mLen = matrix.GetLength(1);
for(int n=0; n<nLen; n++){
for(int m=0; m<mLen; m++){
Console.Write(matrix[n,m] +"\t");
}
Console.WriteLine();
}
Console.WriteLine();
}
static void transpose2dArray(ref int[,] matrix){
int nLen = matrix.GetLength(0);
int mLen = matrix.GetLength(1);
for(int n=0; n<nLen; n++){
for(int m=0; m<mLen; m++){
if(n>m){
int tmp = matrix[n,m];
matrix[n,m] = matrix[m,n];
matrix[m,n] = tmp;
}
}
}
}
static void reverse2dArray(ref int[,] matrix){
int nLen = matrix.GetLength(0);
int mLen = matrix.GetLength(1);
for(int n=0; n<nLen; n++){
for(int m=0; m<mLen/2; m++){
int tmp = matrix[n,m];
matrix[n,m] = matrix[n, mLen-1-m];
matrix[n,mLen-1-m] = tmp;
}
}
}
}
/*
Rotating this matrix by 90+ degree:
1 2 3
4 5 6
7 8 9
Output:
7 4 1
8 5 2
9 6 3
*/
虽然旋转数据可能是必要的(也许是为了更新物理存储的表示)，但在数组访问上添加一层间接层(也许是一个接口)会变得更简单，可能更性能:
interface IReadableMatrix
{
int GetValue(int x, int y);
}
如果你的矩阵已经实现了这个接口，那么它可以通过这样一个装饰器类来旋转:
class RotatedMatrix : IReadableMatrix
{
private readonly IReadableMatrix _baseMatrix;
public RotatedMatrix(IReadableMatrix baseMatrix)
{
_baseMatrix = baseMatrix;
}
int GetValue(int x, int y)
{
// transpose x and y dimensions
return _baseMatrix(y, x);
}
}
旋转+90/-90/180度，水平/垂直翻转和缩放都可以以这种方式实现。
Performance would need to be measured in your specific scenario. However the O(n^2) operation has now been replaced with an O(1) call. It's a virtual method call which is slower than direct array access, so it depends upon how frequently the rotated array is used after rotation. If it's used once, then this approach would definitely win. If it's rotated then used in a long-running system for days, then in-place rotation might perform better. It also depends whether you can accept the up-front cost.
与所有性能问题一样，测量，测量，测量!
Python:
rotated = list(zip(*original[::-1]))
和逆时针方向:
rotated_ccw = list(zip(*original))[::-1]
这是如何工作的:
Zip (*original)将通过将列表中的对应项堆叠到新的列表中来交换2d数组的轴。(*操作符告诉函数将包含的列表分布到参数中)
>>> list(zip(*[[1,2,3],[4,5,6],[7,8,9]]))
[[1,4,7],[2,5,8],[3,6,9]]
语句[::-1]反转数组元素(请参阅扩展切片或这个问题):
>>> [[1,2,3],[4,5,6],[7,8,9]][::-1]
[[7,8,9],[4,5,6],[1,2,3]]
最后，将两者结合就得到了旋转变换。
改变[::-1]的位置将使列表在矩阵的不同层次上颠倒。
这是我的实现，在C, O(1)内存复杂度，原地旋转，顺时针90度:
#include <stdio.h>
#define M_SIZE 5
static void initMatrix();
static void printMatrix();
static void rotateMatrix();
static int m[M_SIZE][M_SIZE];
int main(void){
initMatrix();
printMatrix();
rotateMatrix();
printMatrix();
return 0;
}
static void initMatrix(){
int i, j;
for(i = 0; i < M_SIZE; i++){
for(j = 0; j < M_SIZE; j++){
m[i][j] = M_SIZE*i + j + 1;
}
}
}
static void printMatrix(){
int i, j;
printf("Matrix\n");
for(i = 0; i < M_SIZE; i++){
for(j = 0; j < M_SIZE; j++){
printf("%02d ", m[i][j]);
}
printf("\n");
}
printf("\n");
}
static void rotateMatrix(){
int r, c;
for(r = 0; r < M_SIZE/2; r++){
for(c = r; c < M_SIZE - r - 1; c++){
int tmp = m[r][c];
m[r][c] = m[M_SIZE - c - 1][r];
m[M_SIZE - c - 1][r] = m[M_SIZE - r - 1][M_SIZE - c - 1];
m[M_SIZE - r - 1][M_SIZE - c - 1] = m[c][M_SIZE - r - 1];
m[c][M_SIZE - r - 1] = tmp;
}
}
}
