问题:如何在Python中实现最大堆?
Python包括用于最小堆的heapq模块,但是我需要一个最大堆。在Python中最大堆实现应使用什么?
回答 0
最简单的方法是反转键的值并使用heapq。例如,将1000.0转换为-1000.0,将5.0转换为-5.0。
回答 1
您可以使用
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!!
如果然后要弹出元素,请使用:
heapq.heappop(minheap) # pop from minheap
heapq._heappop_max(maxheap) # pop from maxheap
回答 2
解决方案是在将值存储在堆中时取反值,或像这样反转对象比较:
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)
最大堆的示例:
maxh = []
heapq.heappush(maxh, MaxHeapObj(x))
x = maxh[0].val # fetch max value
x = heapq.heappop(maxh).val # pop max value
但是您必须记住包装和解包值,这需要知道您要处理的是最小堆还是最大堆。
MinHeap,MaxHeap类
为MinHeap
和添加类MaxHeap
可以简化代码:
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
用法示例:
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 3
最简单理想的解决方案
将值乘以-1
妳去 现在,所有最高的数字都是最低的,反之亦然。
只需记住,当您弹出一个元素以使其与-1相乘时,才能再次获得原始值。
回答 4
我实现了heapq的最大堆版本并将其提交给PyPI。(对heapq模块的CPython代码稍作更改。)
https://pypi.python.org/pypi/heapq_max/
https://github.com/he-zhe/heapq_max
安装
pip install heapq_max
用法
tl; dr:与heapq模块相同,只不过在所有函数中添加了“ _max”。
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 5
如果您插入的是可比较的但不是int的键,则可能会覆盖它们上的比较运算符(即<=变为>,而>变为<=)。否则,您可以在heapq模块中覆盖heapq._siftup(最后只是Python代码)。
回答 6
允许您选择任意数量的最大或最小项目
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 7
扩展int类并覆盖__lt__是方法之一。
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 8
我创建了一个堆包装器,该堆包装器将这些值取反以创建一个最大堆,以及一个用于最小堆的包装器类,以使库更像OOP。这里是要点。一共有三节课;堆(抽象类),HeapMin和HeapMax。
方法:
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 9
如果您想使用最大堆来获取最大的K元素,可以执行以下技巧:
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]
回答 10
遵循艾萨克·特纳(Isaac Turner)的出色回答,我想举一个基于使用最大堆的K个最接近原点的示例。
from math import sqrt
import heapq
class MaxHeapObj(object):
def __init__(self, val):
self.val = val.distance
self.coordinates = val.coordinates
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)
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 peek(self):
return heapq.nsmallest(1, self.h)[0].val
def __getitem__(self, i):
return self.h[i].val
class Point():
def __init__(self, x, y):
self.distance = round(sqrt(x**2 + y**2), 3)
self.coordinates = (x, y)
def find_k_closest(points, k):
res = [Point(x, y) for (x, y) in points]
maxh = MaxHeap()
for i in range(k):
maxh.heappush(res[i])
for p in res[k:]:
if p.distance < maxh.peek():
maxh.heappop()
maxh.heappush(p)
res = [str(x.coordinates) for x in maxh.h]
print(f"{k} closest points from origin : {', '.join(res)}")
points = [(10, 8), (-2, 4), (0, -2), (-1, 0), (3, 5), (-2, 3), (3, 2), (0, 1)]
find_k_closest(points, 3)
回答 11
为了详细说明https://stackoverflow.com/a/59311063/1328979,这里是针对一般情况的完整记录,带注释和经过测试的Python 3实现。
from __future__ import annotations # To allow "MinHeap.push -> MinHeap:"
from typing import Generic, List, Optional, TypeVar
from heapq import heapify, heappop, heappush, heapreplace
T = TypeVar('T')
class MinHeap(Generic[T]):
'''
MinHeap provides a nicer API around heapq's functionality.
As it is a minimum heap, the first element of the heap is always the
smallest.
>>> h = MinHeap([3, 1, 4, 2])
>>> h[0]
1
>>> h.peek()
1
>>> h.push(5) # N.B.: the array isn't always fully sorted.
[1, 2, 4, 3, 5]
>>> h.pop()
1
>>> h.pop()
2
>>> h.pop()
3
>>> h.push(3).push(2)
[2, 3, 4, 5]
>>> h.replace(1)
2
>>> h
[1, 3, 4, 5]
'''
def __init__(self, array: Optional[List[T]] = None):
if array is None:
array = []
heapify(array)
self.h = array
def push(self, x: T) -> MinHeap:
heappush(self.h, x)
return self # To allow chaining operations.
def peek(self) -> T:
return self.h[0]
def pop(self) -> T:
return heappop(self.h)
def replace(self, x: T) -> T:
return heapreplace(self.h, x)
def __getitem__(self, i) -> T:
return self.h[i]
def __len__(self) -> int:
return len(self.h)
def __str__(self) -> str:
return str(self.h)
def __repr__(self) -> str:
return str(self.h)
class Reverse(Generic[T]):
'''
Wrap around the provided object, reversing the comparison operators.
>>> 1 < 2
True
>>> Reverse(1) < Reverse(2)
False
>>> Reverse(2) < Reverse(1)
True
>>> Reverse(1) <= Reverse(2)
False
>>> Reverse(2) <= Reverse(1)
True
>>> Reverse(2) <= Reverse(2)
True
>>> Reverse(1) == Reverse(1)
True
>>> Reverse(2) > Reverse(1)
False
>>> Reverse(1) > Reverse(2)
True
>>> Reverse(2) >= Reverse(1)
False
>>> Reverse(1) >= Reverse(2)
True
>>> Reverse(1)
1
'''
def __init__(self, x: T) -> None:
self.x = x
def __lt__(self, other: Reverse) -> bool:
return other.x.__lt__(self.x)
def __le__(self, other: Reverse) -> bool:
return other.x.__le__(self.x)
def __eq__(self, other) -> bool:
return self.x == other.x
def __ne__(self, other: Reverse) -> bool:
return other.x.__ne__(self.x)
def __ge__(self, other: Reverse) -> bool:
return other.x.__ge__(self.x)
def __gt__(self, other: Reverse) -> bool:
return other.x.__gt__(self.x)
def __str__(self):
return str(self.x)
def __repr__(self):
return str(self.x)
class MaxHeap(MinHeap):
'''
MaxHeap provides an implement of a maximum-heap, as heapq does not provide
it. As it is a maximum heap, the first element of the heap is always the
largest. It achieves this by wrapping around elements with Reverse,
which reverses the comparison operations used by heapq.
>>> h = MaxHeap([3, 1, 4, 2])
>>> h[0]
4
>>> h.peek()
4
>>> h.push(5) # N.B.: the array isn't always fully sorted.
[5, 4, 3, 1, 2]
>>> h.pop()
5
>>> h.pop()
4
>>> h.pop()
3
>>> h.pop()
2
>>> h.push(3).push(2).push(4)
[4, 3, 2, 1]
>>> h.replace(1)
4
>>> h
[3, 1, 2, 1]
'''
def __init__(self, array: Optional[List[T]] = None):
if array is not None:
array = [Reverse(x) for x in array] # Wrap with Reverse.
super().__init__(array)
def push(self, x: T) -> MaxHeap:
super().push(Reverse(x))
return self
def peek(self) -> T:
return super().peek().x
def pop(self) -> T:
return super().pop().x
def replace(self, x: T) -> T:
return super().replace(Reverse(x)).x
if __name__ == '__main__':
import doctest
doctest.testmod()
https://gist.github.com/marccarre/577a55850998da02af3d4b7b98152cf4
回答 12
这是MaxHeap
基于的简单实现heapq
。虽然它仅适用于数值。
import heapq
from typing import List
class MaxHeap:
def __init__(self):
self.data = []
def top(self):
return -self.data[0]
def push(self, val):
heapq.heappush(self.data, -val)
def pop(self):
return -heapq.heappop(self.data)
用法:
max_heap = MaxHeap()
max_heap.push(3)
max_heap.push(5)
max_heap.push(1)
print(max_heap.top()) # 5