158 lines
5.4 KiB
Python
158 lines
5.4 KiB
Python
# -*- coding: utf-8 -*-
|
|
#
|
|
# Copyright (C) 2014
|
|
# ysitu <ysitu@users.noreply.github.com>
|
|
# All rights reserved.
|
|
# BSD license.
|
|
"""
|
|
Stoer-Wagner minimum cut algorithm.
|
|
"""
|
|
from itertools import islice
|
|
|
|
import networkx as nx
|
|
from ...utils import BinaryHeap
|
|
from ...utils import not_implemented_for
|
|
from ...utils import arbitrary_element
|
|
|
|
__author__ = 'ysitu <ysitu@users.noreply.github.com>'
|
|
|
|
__all__ = ['stoer_wagner']
|
|
|
|
|
|
@not_implemented_for('directed')
|
|
@not_implemented_for('multigraph')
|
|
def stoer_wagner(G, weight='weight', heap=BinaryHeap):
|
|
r"""Returns the weighted minimum edge cut using the Stoer-Wagner algorithm.
|
|
|
|
Determine the minimum edge cut of a connected graph using the
|
|
Stoer-Wagner algorithm. In weighted cases, all weights must be
|
|
nonnegative.
|
|
|
|
The running time of the algorithm depends on the type of heaps used:
|
|
|
|
============== =============================================
|
|
Type of heap Running time
|
|
============== =============================================
|
|
Binary heap $O(n (m + n) \log n)$
|
|
Fibonacci heap $O(nm + n^2 \log n)$
|
|
Pairing heap $O(2^{2 \sqrt{\log \log n}} nm + n^2 \log n)$
|
|
============== =============================================
|
|
|
|
Parameters
|
|
----------
|
|
G : NetworkX graph
|
|
Edges of the graph are expected to have an attribute named by the
|
|
weight parameter below. If this attribute is not present, the edge is
|
|
considered to have unit weight.
|
|
|
|
weight : string
|
|
Name of the weight attribute of the edges. If the attribute is not
|
|
present, unit weight is assumed. Default value: 'weight'.
|
|
|
|
heap : class
|
|
Type of heap to be used in the algorithm. It should be a subclass of
|
|
:class:`MinHeap` or implement a compatible interface.
|
|
|
|
If a stock heap implementation is to be used, :class:`BinaryHeap` is
|
|
recommended over :class:`PairingHeap` for Python implementations without
|
|
optimized attribute accesses (e.g., CPython) despite a slower
|
|
asymptotic running time. For Python implementations with optimized
|
|
attribute accesses (e.g., PyPy), :class:`PairingHeap` provides better
|
|
performance. Default value: :class:`BinaryHeap`.
|
|
|
|
Returns
|
|
-------
|
|
cut_value : integer or float
|
|
The sum of weights of edges in a minimum cut.
|
|
|
|
partition : pair of node lists
|
|
A partitioning of the nodes that defines a minimum cut.
|
|
|
|
Raises
|
|
------
|
|
NetworkXNotImplemented
|
|
If the graph is directed or a multigraph.
|
|
|
|
NetworkXError
|
|
If the graph has less than two nodes, is not connected or has a
|
|
negative-weighted edge.
|
|
|
|
Examples
|
|
--------
|
|
>>> G = nx.Graph()
|
|
>>> G.add_edge('x', 'a', weight=3)
|
|
>>> G.add_edge('x', 'b', weight=1)
|
|
>>> G.add_edge('a', 'c', weight=3)
|
|
>>> G.add_edge('b', 'c', weight=5)
|
|
>>> G.add_edge('b', 'd', weight=4)
|
|
>>> G.add_edge('d', 'e', weight=2)
|
|
>>> G.add_edge('c', 'y', weight=2)
|
|
>>> G.add_edge('e', 'y', weight=3)
|
|
>>> cut_value, partition = nx.stoer_wagner(G)
|
|
>>> cut_value
|
|
4
|
|
"""
|
|
n = len(G)
|
|
if n < 2:
|
|
raise nx.NetworkXError('graph has less than two nodes.')
|
|
if not nx.is_connected(G):
|
|
raise nx.NetworkXError('graph is not connected.')
|
|
|
|
# Make a copy of the graph for internal use.
|
|
G = nx.Graph((u, v, {'weight': e.get(weight, 1)})
|
|
for u, v, e in G.edges(data=True) if u != v)
|
|
|
|
for u, v, e, in G.edges(data=True):
|
|
if e['weight'] < 0:
|
|
raise nx.NetworkXError('graph has a negative-weighted edge.')
|
|
|
|
cut_value = float('inf')
|
|
nodes = set(G)
|
|
contractions = [] # contracted node pairs
|
|
|
|
# Repeatedly pick a pair of nodes to contract until only one node is left.
|
|
for i in range(n - 1):
|
|
# Pick an arbitrary node u and create a set A = {u}.
|
|
u = arbitrary_element(G)
|
|
A = set([u])
|
|
# Repeatedly pick the node "most tightly connected" to A and add it to
|
|
# A. The tightness of connectivity of a node not in A is defined by the
|
|
# of edges connecting it to nodes in A.
|
|
h = heap() # min-heap emulating a max-heap
|
|
for v, e in G[u].items():
|
|
h.insert(v, -e['weight'])
|
|
# Repeat until all but one node has been added to A.
|
|
for j in range(n - i - 2):
|
|
u = h.pop()[0]
|
|
A.add(u)
|
|
for v, e, in G[u].items():
|
|
if v not in A:
|
|
h.insert(v, h.get(v, 0) - e['weight'])
|
|
# A and the remaining node v define a "cut of the phase". There is a
|
|
# minimum cut of the original graph that is also a cut of the phase.
|
|
# Due to contractions in earlier phases, v may in fact represent
|
|
# multiple nodes in the original graph.
|
|
v, w = h.min()
|
|
w = -w
|
|
if w < cut_value:
|
|
cut_value = w
|
|
best_phase = i
|
|
# Contract v and the last node added to A.
|
|
contractions.append((u, v))
|
|
for w, e in G[v].items():
|
|
if w != u:
|
|
if w not in G[u]:
|
|
G.add_edge(u, w, weight=e['weight'])
|
|
else:
|
|
G[u][w]['weight'] += e['weight']
|
|
G.remove_node(v)
|
|
|
|
# Recover the optimal partitioning from the contractions.
|
|
G = nx.Graph(islice(contractions, best_phase))
|
|
v = contractions[best_phase][1]
|
|
G.add_node(v)
|
|
reachable = set(nx.single_source_shortest_path_length(G, v))
|
|
partition = (list(reachable), list(nodes - reachable))
|
|
|
|
return cut_value, partition
|