What do I use for a max-heap implementation in Python?

问题

Python includes the heapq module for min-heaps, but I need a max heap. What should I use for a max-heap implementation in Python?

回答1:

The easiest way is to invert the value of the keys and use heapq. For example, turn 1000.0 into -1000.0 and 5.0 into -5.0.

回答2:

You can use

import heapq
listForTree = [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]    
heapq.heapify(listForTree)             # for a min heap
heapq._heapify_max(listForTree)        # for a maxheap!!

If you then want to pop elements, use:

heapq.heappop(minheap)      # pop from minheap
heapq._heappop_max(maxheap) # pop from maxheap

回答3:

The solution is to negate your values when you store them in the heap, or invert your object comparison like so:

import heapq

class MaxHeapObj(object):
  def __init__(self,val): self.val = val
  def __lt__(self,other): return self.val > other.val
  def __eq__(self,other): return self.val == other.val
  def __str__(self): return str(self.val)

Example of a max-heap:

maxh = []
heapq.heappush(maxh,MaxHeapInt(x))
x = maxh[0].val # fetch max value
x = heapq.heappop(maxh).val # pop max value

But you have to remember to wrap and unwrap your values, which requires knowing if you are dealing with a min- or max-heap.

MinHeap, MaxHeap classes

Adding classes for MinHeap and MaxHeap objects can simplify your code:

class MinHeap(object):
  def __init__(self): self.h = []
  def heappush(self,x): heapq.heappush(self.h,x)
  def heappop(self): return heapq.heappop(self.h)
  def __getitem__(self,i): return self.h[i]
  def __len__(self): return len(self.h)

class MaxHeap(MinHeap):
  def heappush(self,x): heapq.heappush(self.h,MaxHeapObj(x))
  def heappop(self): return heapq.heappop(self.h).val
  def __getitem__(self,i): return self.h[i].val

Example usage:

minh = MinHeap()
maxh = MaxHeap()
# add some values
minh.heappush(12)
maxh.heappush(12)
minh.heappush(4)
maxh.heappush(4)
# fetch "top" values
print(minh[0],maxh[0]) # "4 12"
# fetch and remove "top" values
print(minh.heappop(),maxh.heappop()) # "4 12"

回答4:

The easiest and ideal solution

Multiply the values by -1

There you go. All the highest numbers are now the lowest and vice versa.

Just remember that when you pop an element to multiply it with -1 in order to get the original value again.

回答5:

If you are inserting keys that are comparable but not int-like, you could potentially override the comparison operators on them (i.e. <= become > and > becomes <=). Otherwise, you can override heapq._siftup in the heapq module (it's all just Python code, in the end).

回答6:

I implemented a max heap version of heapq and submitted it to PyPI. (Very slight change of heapq module CPython code.)

https://pypi.python.org/pypi/heapq_max/

https://github.com/he-zhe/heapq_max

Installation

pip install heapq_max

Usage

tl;dr: same as heapq module except adding ‘_max’ to all functions.

heap_max = []                           # creates an empty heap
heappush_max(heap_max, item)            # pushes a new item on the heap
item = heappop_max(heap_max)            # pops the largest item from the heap
item = heap_max[0]                      # largest item on the heap without popping it
heapify_max(x)                          # transforms list into a heap, in-place, in linear time
item = heapreplace_max(heap_max, item)  # pops and returns largest item, and
                                    # adds new item; the heap size is unchanged

回答7:

Allowing you to chose an arbitrary amount of largest or smallest items

import heapq
heap = [23, 7, -4, 18, 23, 42, 37, 2, 8, 2, 23, 7, -4, 18, 23, 42, 37, 2]
heapq.heapify(heap)
print(heapq.nlargest(3, heap))  # [42, 42, 37]
print(heapq.nsmallest(3, heap)) # [-4, -4, 2]

回答8:

Extending the int class and overriding __lt__ is one of the ways.

import queue
class MyInt(int):
    def __lt__(self, other):
        return self > other

def main():
    q = queue.PriorityQueue()
    q.put(MyInt(10))
    q.put(MyInt(5))
    q.put(MyInt(1))
    while not q.empty():
        print (q.get())


if __name__ == "__main__":
    main()

回答9:

I have created a heap wrapper that inverts the values to create a max-heap, as well as a wrapper class for a min-heap to make the library more OOP-like. Here is the gist. There are three classes; Heap (abstract class), HeapMin, and HeapMax.

Methods:

isempty() -> bool; obvious
getroot() -> int; returns min/max
push() -> None; equivalent to heapq.heappush
pop() -> int; equivalent to heapq.heappop
view_min()/view_max() -> int; alias for getroot()
pushpop() -> int; equivalent to heapq.pushpop

回答10:

In case if you would like to get the largest K element using max heap, you can do the following trick:

nums= [3,2,1,5,6,4]
k = 2  #k being the kth largest element you want to get
heapq.heapify(nums) 
temp = heapq.nlargest(k, nums)
return temp[-1]

来源：https://stackoverflow.com/questions/2501457/what-do-i-use-for-a-max-heap-implementation-in-python