[cc]18.9中位数

Numbers are randomly generated and passed to a method. Write a program to find and maintain the median value as new values are generated.

随机产生一些数传递给一个函数，写程序找出并维护这些数的中位数。

Solution：

1 足够大的数组，将元素排序。 O(1) get median.

new number comes, insert into the array, O(N) time to move bigger element to the end.

2 Heap

(max or min) , O(logN) to insert

取中位数需要先排序，时间复杂度O(nlogn)

3 基于方法2， 用两个heap，一个maxheap，value <= median,一个minheap ， value > medina.

这样，maxheap.size() > minHeap.size(), maxHeap.top()就是median

如果两个heap大小一样，median就是两个heap的top的平均数

public class MinHeapComparator implements Comparator<Integer>{
	//comparator sort integers from lowest to highest
	public int compare(Integer o1, Integer o2){
		if(o1 > o2) return 1;
		else if( o1 == o2) return 0;
		else return -1;
	}
}
public class MaxHeapComparator implements Comparator<Integer>{
	// Comparator that sorts integers from highest to lowest
	@Override
	public int compare(Integer o1, Integer o2){
		if(o1 < o2) return 1;
		else if( o1 == o2) return 0;
		else return -1;
	}
}

	public static double getMedian() {
		/* maxHeap is always at least as big as minHeap. So if maxHeap is empty, then minHeap is also. */		
		if (maxHeap.isEmpty()) return 0;
		
		if (maxHeap.size() == minHeap.size()) {
			return ((double)minHeap.peek() + (double) maxHeap.peek()) / 2;
		} else {
			/* If maxHeap and minHeap are of different sizes, then maxHeap must have one extra element. Return maxHeap¡¯s top element.*/			
			return maxHeap.peek();
		} 
	}

插入一个新数：

注意需要判断应该插入maxHeap还是minHeap，然后根据size，banalance两个heap确保maxheap至多比minheap多一个元素

//insert a new number
public void addNewNumber(int randomNum){
	//
	if(maxHeap.size() == minHeap.size()){
		//should add to max heap to make sure maxheap is bigger
		if((minHeap.peek() != null) && (randomNum > minHeap.peek())){
			//> median, should add to min heap, to make sure maxheap has more number
			//need remove one to maxheap
			//O(logN)
			maxHeap.offer(minHeap.poll()); //move top to maxHeap
			minHeap.offer(randomNum);
		}else{
			//<= median
			maxHeap.offer(randomNum);
		}
	}else{
		//maxHeap has more number
		if(randomNum < maxHeap.peek()){
			//move one to min to balance the two heap
			minHeap.offer(maxHeap.poll());
			maxHeap.offer(randomNum);
		}else{
			minHeap.offer(randomNum);
		}
	}
}