每日一题(55) - 最小的k个数

题目来自剑指Offer

题目

思路：求最小的k个数，建立一个容量为k的大根堆。插入元素时，堆不满直接插入，堆满且数小插入。

使用堆的原因：适用于海量数据。使用堆不用把海量数据一次性放入内存，也不用打乱源数据的顺序。

代码：借助STL的堆

#include <iostream>
#include <algorithm>
#include <vector>
#include <assert.h>
using namespace std;

void OutPut(vector<int>& Heap)
{
	vector<int>::iterator itCur = Heap.begin();
	while(itCur != Heap.end())
	{
		cout<<*itCur<<" ";
		itCur++;
	}
}

void FindTopKMin(int nArr[],int nLen,int nTopK)
{
	assert(nLen > 0 && nLen >= nTopK);
	vector<int> Heap(nArr,nArr + nTopK);
	make_heap(Heap.begin(),Heap.end());//默认为大根堆
	//TopK比较
	for (int i = nTopK;i < nLen;i++)
	{
		if (nArr[i] < Heap[0])
		{
			Heap[0] = nArr[i];
			make_heap(Heap.begin(),Heap.end());
		}
	}
	//输出结果
	sort_heap(Heap.begin(),Heap.end());
	OutPut(Heap);
	
}


int main()
{
	int nTopK = 8;
	int nLen = 10;
	int nArr[10] = {1,3,8,5,3,20,15,9,20,18};
	FindTopKMin(nArr,nLen,nTopK);
	system("pause");
	return 1;
}

代码：自己手写堆

#include <iostream>
#include <assert.h>
using namespace std;

const int TOP_K = 8; 

int nMaxSize = TOP_K;
int nCurSize = 0;
int arrHeap[TOP_K + 1];

void Adjust(int nPos)
{
	assert(nPos > 0);
	int nData = arrHeap[nPos];
	int nParentPos = nPos;
	int nChildPos = nPos << 1;
	while(nChildPos <= nCurSize)
	{
		if (nChildPos < nCurSize && arrHeap[nChildPos] < arrHeap[nChildPos + 1])
		{
			nChildPos++;
		}

		if (nData < arrHeap[nChildPos])
		{
			arrHeap[nParentPos] = arrHeap[nChildPos];
			nParentPos = nChildPos;
			nChildPos <<= 1;
		}
		else
		{
			break;
		}
	}
	assert(nParentPos > 0 && nParentPos <= nCurSize);
	arrHeap[nParentPos] = nData;
}

void Insert(int nPos,int nData)
{
	arrHeap[0] = nData;
	int nChildPos = nCurSize;
	int nParentPos = nChildPos >> 1;
	while(arrHeap[nParentPos] < nData)
	{
		arrHeap[nChildPos] = arrHeap[nParentPos];
		nChildPos = nParentPos;
		nParentPos = nParentPos >> 1;
	}
	arrHeap[nChildPos] = nData;
}

void Insert(int nData)
{
	assert(nCurSize >= 0);
	
	if (nCurSize < nMaxSize)
	{
		++nCurSize;
		Insert(nCurSize,nData);
	}
	else
	{
		assert(nCurSize == nMaxSize);
		if (arrHeap[1] > nData) //比最大值小，则插入堆
		{
			arrHeap[1] = nData;
			Adjust(1);
		}
	}
}

void Sort()
{
	assert(nCurSize > 0);
	while(nCurSize > 0)
	{
		cout<<arrHeap[1]<<" ";
		
		arrHeap[1] = arrHeap[nCurSize];
		Adjust(1);
		nCurSize--;
	}
}


int main()
{
	int nArr[10] = {1,3,8,5,3,20,15,9,20,18};
	for (int nCur = 0;nCur < 10;nCur++)
	{
		Insert(nArr[nCur]);
	}
	cout<<"元素输出:"<<endl;
	Sort();
	cout<<endl;

	system("pause");
	return 1;
}

把堆封装成类：

#include <iostream>
#include <assert.h>
using namespace std;

const int TOP_K = 8; 

class CTopKMin
{
public:
	CTopKMin(int nMaxSize = TOP_K);
	void Init();
	void Insert(int nData);
	bool IsMax();
	void Sort();
	void Output();
private:
	void Adjust(int nPos);
	void Insert(int nPos,int nData);
private:
	int* m_pHeap;
	int m_nCurSize;
	int m_nMaxSize;
};

CTopKMin::CTopKMin(int nMaxSize)
{
	m_nCurSize = 0;
	m_nMaxSize = nMaxSize;
	m_pHeap = new int[m_nMaxSize + 1];
	assert(m_nMaxSize > 0);
}

void CTopKMin::Init()
{
	m_nCurSize = 0;
}

void CTopKMin::Insert(int nData)
{
	assert(m_nCurSize >= 0);
	
	if (m_nCurSize < m_nMaxSize)
	{
		++m_nCurSize;
		Insert(m_nCurSize,nData);
	}
	else
	{
		assert(m_nCurSize == m_nMaxSize);
		if (m_pHeap[1] > nData) //比最大值小，则插入堆
		{
			m_pHeap[1] = nData;
			Adjust(1);
		}
	}
}

void CTopKMin::Insert(int nPos,int nData)
{
	m_pHeap[0] = nData;
	int nChildPos = m_nCurSize;
	int nParentPos = nChildPos >> 1;
	while(m_pHeap[nParentPos] < nData)
	{
		m_pHeap[nChildPos] = m_pHeap[nParentPos];
		nChildPos = nParentPos;
		nParentPos = nParentPos >> 1;
	}
	m_pHeap[nChildPos] = nData;
}

void CTopKMin::Adjust(int nPos)
{
	assert(nPos > 0);
	int nData = m_pHeap[nPos];
	int nParentPos = nPos;
	int nChildPos = nPos << 1;
	while(nChildPos <= m_nCurSize)
	{
		if (nChildPos < m_nCurSize && m_pHeap[nChildPos] < m_pHeap[nChildPos + 1])
		{
			nChildPos++;
		}

		if (nData < m_pHeap[nChildPos])
		{
			m_pHeap[nParentPos] = m_pHeap[nChildPos];
			nParentPos = nChildPos;
			nChildPos <<= 1;
		}
		else
		{
			break;
		}
	}
	assert(nParentPos > 0 && nParentPos <= m_nCurSize);
	m_pHeap[nParentPos] = nData;
}

bool CTopKMin::IsMax()
{
	assert(m_nCurSize > 0);
	int nChildPos = m_nCurSize;
	int nParentPos = m_nCurSize >> 1;
	while(nChildPos > 1 && m_pHeap[nChildPos] <= m_pHeap[nParentPos])
	{
		nChildPos--;
		nParentPos = nChildPos >> 1;
	}
	if (nChildPos == 1)
	{
		return true;
	}
	else
	{
		return false;
	}

};

void CTopKMin::Sort()
{
	assert(m_nCurSize > 0);
	while(m_nCurSize > 0)
	{
		cout<<m_pHeap[1]<<" ";
		
		m_pHeap[1] = m_pHeap[m_nCurSize];
		Adjust(1);
		m_nCurSize--;
	}
}

void CTopKMin::Output()
{
	for (int nCur = 1;nCur <= m_nCurSize;nCur++)
	{
		cout<<m_pHeap[nCur]<<" ";
	}
	cout<<endl;
}

int main()
{
	int nArr[10] = {1,3,8,5,3,20,15,9,20,18};
	CTopKMin topKMin;
	for (int nCur = 0;nCur < 10;nCur++)
	{
		topKMin.Insert(nArr[nCur]);
		topKMin.Output();
	}


	if (topKMin.IsMax())
	{
		cout<<"大顶堆!"<<endl;
	}
	else
	{
		cout<<"不是大顶堆!"<<endl;
	}
	cout<<"元素输出:"<<endl;
	topKMin.Sort();
	cout<<endl;

	system("pause");
	return 1;
}