本质上,与findKth是一样的,有三种算法:
1.堆,适用于大数据的场景,只需要Ok的空间。
public List<Integer> topKByHeap(int[] array, int k){
PriorityQueue<Integer> minHeap = new PriorityQueue<>();
for(int i : array){
if(minHeap.size() < k)
minHeap.add(i);
else if(minHeap.peek() < i){
minHeap.poll();
minHeap.add(i);
}
}
List<Integer> r = new ArrayList<>();
r.addAll(minHeap);
return r;
}
2.随机的partition。
public List<Integer> topKByPartition(int[] array, int k){
int low = 0, high = array.length - 1;
while(low <= high){
int m = randomPartition(array, low, high);
if(m == array.length - k)
return getResult(array, m);
else if(m > array.length - k)
high = m - 1;
else
low = m + 1;
}
return null;
}
private List<Integer> getResult(int[] array, int start){
List<Integer> r = new ArrayList<>();
for(int i = start; i < array.length; i++){
r.add(array[i]);
}
return r;
}
private int randomPartition(int[] array, int start, int end){
int random = random(start, end);
swap(array, start, random);
return partition(array, start, end);
}
private int partition(int[] array, int start, int end){
int pivot = array[start];
int last = start;
for(int i = start + 1; i <= end; i++){
if(array[i] <= pivot){
last++;
swap(array, last, i);
}
}
swap(array, last, start);
return last;
}
private void swap(int[] array, int i, int j){
int t = array[i];
array[i] = array[j];
array[j] = t;
}
private int random(int min, int max){
Random r = new Random();
return r.nextInt(max - min + 1) + min;
}
这里需要注意partition的写法和random数字生成。
3.中位数的中位数。