本意是这样的，给定一个整数数组，查找最小k个元素或最大k个元素。

假定有这样一组数列 { 10, 33, 2, 4, 55, 6, 12, 34, 456, 66, 43, 23, 65, 1, 345, 61, 76, 31, 43, 76 };

如和求出最小的k个值，或最大的k个值？下面我结合快速排序来进行了查找，本人只是做了一个实例，抛砖引玉，希望大家有更好的方法，代码是用C#代码实现的。

private int Partition(int[] R, int low, int high)
        {
            int temp = R[low];
             while (low < high)
            {
                while (low < high && temp <= R[high])
                {
                    high--;
                }
                R[low] = R[high];
                while (low < high && temp >= R[low])
                {
                    low++;
                }
                R[high] = R[low];
            }
            R[low] = temp;
            return low;
        }

/// <summary>
        /// 快速排序
        /// </summary>
        /// <param name="R"></param>
        /// <param name="low"></param>
        /// <param name="high"></param>
        private void QuickSort(int[] R, int low, int high,int k)
        {
          
            int pivotLoc = 0;
            if (low < high)
            {
                pivotLoc = Partition(R, low, high);
                QuickSort(R, low, pivotLoc - 1, k);
                if ( pivotLoc <= k)
                {                
                    QuickSort(R, pivotLoc + 1, high, k);
                }                                      
            }
        }

这是求k个最小值的，如果求ｋ个最大值的，只需要修改下partition方法就可以了，本质上是一致的。

/// <summary>
        /// 从大到小排序
        /// </summary>
        /// <param name="R"></param>
        /// <param name="low"></param>
        /// <param name="high"></param>
        /// <returns></returns>
        private static int Big_To_Small_Partition(int[] R, int low, int high)
        {
            int temp = R[high];
            while (low < high)
            {
                while (low < high && temp <= R[low])
                {
                    low++;
                }
                R[high] = R[low];
                while (low < high && temp > R[high])
                {
                    high--;
                }
                R[low] = R[high];
            }
            R[high] = temp;
            return high;
        }