Double heap to find the median number

Heap

Dynamic creation and deletion of heaps can be referred to http://www.java3z.com/cwbwebhome/article/article1/1362.html?id=4745, which is not mentioned here.

double heap to find the median number

Algorithm Description:

1, create two heaps (a small Gan, a large root heap), the heap size is at least half of the given number of data, rounding up;

2, the assumption that the variable mid is used to save the median, the first element, assigned to mid, that is, as the initial median;

3, sequentially traverse each subsequent data, if smaller than mid, then insert Dagen; otherwise insert small Gan;

4, if the number of data on the Dagen and the small root is 2, the mid is inserted into a heap with fewer elements, then the root node is removed from the heap with a large number of elements, and the node is assigned a value to mid;

5. Repeat steps 3 and 4 until all data traversal is complete;

At this point, mid saves a number, plus the number saved in two heaps, which makes up a collection of the given data.

If the number of elements in the two heap is equal, mid is the final median, otherwise, the root node element of the heap with more elements is the median of the mid and the average of the middle.

Algorithm implementation

The implementation of the algorithm takes an integer, so the final result takes an integer, which converts the return value to a floating-point type and is more accurate.

uint32_t Getmiddle (uint32_t*array, uint32_t size) {  inti =0, Minelem =0, Maxelem =0;  uint32_t mid = array[0];  uint32_t heapsize = (size+1)/2;  heap_t*minheap= Heap_malloc (heapsize);  heap_t*maxheap= Heap_malloc (heapsize);   for(i =1; i < size; i++)  {    if(Array[i] < mid)    {      Maxheap_insert (Maxheap, array[i]);      maxelem++;    }    Else    {      Minheap_insert (Minheap, array[i]);      minelem++;    }    if(Minelem-maxelem >1)    {      Maxheap_insert (Maxheap, mid);      MID = minheap->element[0];      Minheap_delete (MINHEAP);      maxelem++;      minelem--;    }    if(Maxelem-minelem >1)    {      Minheap_insert (Minheap, mid);      MID = maxheap->element[0];      Maxheap_delete (MAXHEAP);      minelem++;      maxelem--;    }  }  printf("\nmaxelem = %d, Minelem = %d, Pre_mid = %d\ n", Maxelem, Minelem,mid);  if(Minelem > Maxelem)  {    printf("\nmid[ %d ] = (Mid [ %d ]+minheap->element[0][ %d ])/2 = %d\ n", Mid, Mid,      minheap->element[0], (mid+minheap->element[0])/2);    Mid = (mid+minheap->element[0])/2;  }  if(Maxelem < Maxelem)  {      printf("\nmid[ %d ] = (Mid [ %d ]+maxheap->element[0][ %d ])/2 = %d\ n", Mid, Mid,      minheap->element[0], (mid+maxheap->element[0])/2);    Mid = (mid+maxheap->element[0])/2;  }      Heap_free (MINHEAP);  Heap_free (MAXHEAP);  returnMid;}

Test Program

intMainintargcChar*argv[]) {intnum =Ten; uint32_tArray[num] = {2, -, -, -, the,8,3,5,4, -}; Getmiddle (Array, num); printf"\ n");return 0;}

Test Results

Double heap to find the median number