下面是一个原地旋转的数组，而不是使用一个全新的数组来保存结果。我已经停止了数组的初始化和输出。这只适用于正方形数组，但它们可以是任何大小。内存开销等于数组中一个元素的大小，因此您可以对任意大的数组进行旋转。

int a[4][4];
int n = 4;
int tmp;
for (int i = 0; i < n / 2; i++)
{
    for (int j = i; j < n - i - 1; j++)
    {
        tmp             = a[i][j];
        a[i][j]         = a[j][n-i-1];
        a[j][n-i-1]     = a[n-i-1][n-j-1];
        a[n-i-1][n-j-1] = a[n-j-1][i];
        a[n-j-1][i]     = tmp;
    }
}

在JavaScript中实现dimple的+90伪代码(例如转置然后反转每一行):

function rotate90(a){
  // transpose from http://www.codesuck.com/2012/02/transpose-javascript-array-in-one-line.html
  a = Object.keys(a[0]).map(function (c) { return a.map(function (r) { return r[c]; }); });
  // row reverse
  for (i in a){
    a[i] = a[i].reverse();
  }
  return a;
}

哦，伙计。我一直认为这是一个“我很无聊，我能思考什么”的谜题。我想出了我的原地换位码，但到了这里发现你的和我的几乎一模一样……啊,好。这里是Ruby版本。

require 'pp'
n = 10
a = []
n.times { a << (1..n).to_a }

pp a

0.upto(n/2-1) do |i|
  i.upto(n-i-2) do |j|
    tmp             = a[i][j]
    a[i][j]         = a[n-j-1][i]
    a[n-j-1][i]     = a[n-i-1][n-j-1]
    a[n-i-1][n-j-1] = a[j][n-i-1]
    a[j][n-i-1]     = tmp
  end
end

pp a

这是c#的

int[,] array = new int[4,4] {
    { 1,2,3,4 },
    { 5,6,7,8 },
    { 9,0,1,2 },
    { 3,4,5,6 }
};

int[,] rotated = RotateMatrix(array, 4);

static int[,] RotateMatrix(int[,] matrix, int n) {
    int[,] ret = new int[n, n];

    for (int i = 0; i < n; ++i) {
        for (int j = 0; j < n; ++j) {
            ret[i, j] = matrix[n - j - 1, i];
        }
    }

    return ret;
}

下面是我的Ruby版本(注意，值显示的不一样，但它仍然按照描述旋转)。

def rotate(matrix)
  result = []
  4.times { |x|
    result[x] = []
    4.times { |y|
      result[x][y] = matrix[y][3 - x]
    }
  }

  result
end

matrix = []
matrix[0] = [1,2,3,4]
matrix[1] = [5,6,7,8]
matrix[2] = [9,0,1,2]
matrix[3] = [3,4,5,6]

def print_matrix(matrix)
  4.times { |y|
    4.times { |x|
      print "#{matrix[x][y]} "
    }
    puts ""
  }
end

print_matrix(matrix)
puts ""
print_matrix(rotate(matrix))

输出:

1 5 9 3 
2 6 0 4 
3 7 1 5 
4 8 2 6 

4 3 2 1 
8 7 6 5 
2 1 0 9 
6 5 4 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!