第13天 Monotonic Queue and Heap 239、347

239. 滑动窗口最大值 

• 根据题意，i为nums下标，是要在[i - k + 1, i] 中选到最大值，只需要保证两点
• 1.队列头结点需要在[i - k + 1, i]范围内，不符合则要弹出队列开头
  • E.g. nums = [1,3,-1,-3,5,3,2,1] deque = [1,3,-1] 现在要更新下一个窗口，要把1去掉（1的index为0，i现在为3，3-0>=3），加-3
• 2.既然是单调，就要保证每次放进去的数字要比末尾的都大，否则也弹出队列结尾
• 因为单调，当i增长到符合第一个k范围的时候，每滑动一步都将队列头节点放入结果就行了

• Time complexity : O(N), since each element is processed exactly twice - it's index added and then removed from the deque.

• Space complexity : O(N), since O(N−k+1) is used for an output array and O(k) for a deque.

class Solution {
    public int[] maxSlidingWindow(int[] nums, int k) {
        Deque<Integer> deque = new ArrayDeque<Integer>();
        int[] ans = new int[nums.length-k+1];
        for (int i=0; i<nums.length; i++) {
            int start = i-k+1;
            while (!deque.isEmpty() && i-deque.peekFirst()>=k) 
                deque.pollFirst();
            while (!deque.isEmpty() && nums[i]>nums[deque.peekLast()]) 
                deque.pollLast();
            deque.addLast(i);
            if (start >= 0) ans[start] = nums[deque.peekFirst()];
        }
        return ans;
    }
}

347.前 K 个高频元素

• 方法1: 用小根堆，出现频率低的在前面

Queue<Integer> q = new PriorityQueue<> ((a,b)->map.get(a)-map.get(b));

• heap.size()>=k 弹出最前面的（频率最少的）

• Time complexity : O(Nlog⁡k) if k<N and O(N) in the particular case of N=k. That ensures time complexity to be better than O(Nlog⁡N).

• Space complexity : O(N+k) to store the hash map with not more N elements and a heap with k elements.

class Solution {
    public int[] topKFrequent(int[] nums, int k) {
        HashMap<Integer, Integer> map = new HashMap<>();
        for (int num: nums) {
            map.put(num, map.getOrDefault(num, 0)+1);
        }
        
        Queue<Integer> q = new PriorityQueue<> ((a,b)->map.get(a)-map.get(b));
        for (Integer key: map.keySet()) {
            q.add(key);
            if (q.size() > k) {
                q.poll();
            }
        }
        
        int[] ans = new int[k];
        int i=k;
        while (!q.isEmpty()) {
            ans[--i] = q.poll();
        }
        return ans;
    }
}