Mercurial > public > think_complexity
view SmallWorldGraph.py @ 28:15ff31ecec7a
Use a linear scale on the y-axis.
author | Brian Neal <bgneal@gmail.com> |
---|---|
date | Sun, 06 Jan 2013 14:57:01 -0600 |
parents | f6073c187926 |
children | 305cc03c2750 |
line wrap: on
line source
"""This module contains the SmallWorldGraph class for 4.4, exercise 4. """ import heapq import itertools import random from Graph import Edge from RandomGraph import RandomGraph INFINITY = float('Inf') class SmallWorldGraph(RandomGraph): """A small world graph for 4.4, exercise 4 in Think Complexity.""" def __init__(self, vs, k, p): """Create a small world graph. Parameters: vs - a list of vertices k - regular graph parameter k: the number of edges between vertices p - probability for rewiring. First a regular graph is created from the list of vertices and parameter k, which specifies how many edges each vertex should have. Then the rewire() method is called with parameter p to rewire the graph according to the algorithm by Watts and Strogatz. """ super(SmallWorldGraph, self).__init__(vs=vs) self.add_regular_edges(k) self.rewire(p) # give each edge a default length of 1; this is used by the # shortest_path_tree() method for e in self.edges(): e.length = 1 def rewire(self, p): """Assuming the graph is initially a regular graph, rewires the graph using the Watts and Strogatz algorithm. """ # We iterate over the edges in a random order: edges = self.edges() random.shuffle(edges) vertices = self.vertices() for e in edges: if random.random() < p: v, w = e # remember the 2 vertices this edge connected self.remove_edge(e) # remove from graph # pick another vertex to connect to v; duplicate edges are # forbidden while True: x = random.choice(vertices) if v is not x and not self.get_edge(v, x): self.add_edge(Edge(v, x)) break def clustering_coefficient(self): """Compute the clustering coefficient for this graph as defined by Watts and Strogatz. """ cv = {} for v in self: # consider a node and its neighbors nodes = self.out_vertices(v) nodes.append(v) # compute the maximum number of possible edges between these nodes # if they were all connected to each other: n = len(nodes) if n == 1: # edge case of only 1 node; handle this case to avoid division # by zero in the general case cv[v] = 1.0 continue possible = n * (n - 1) / 2.0 # now compute how many edges actually exist between the nodes actual = 0 for x, y in itertools.combinations(nodes, 2): if self.get_edge(x, y): actual += 1 # the fraction of actual / possible is this nodes C sub v value cv[v] = actual / possible # The clustering coefficient is the average of all C sub v values if len(cv): return sum(cv.values()) / float(len(cv)) return 0.0 def shortest_path_tree(self, source, hint=None): """Finds the length of the shortest path from the source vertex to all other vertices in the graph. This length is stored on the vertices as an attribute named 'dist'. The algorithm used is Dijkstra's. hint: if provided, must be a dictionary mapping tuples to already known shortest path distances. This can be used to speed up the algorithm. """ if not hint: hint = {} for v in self.vertices(): v.dist = hint.get((source, v), INFINITY) source.dist = 0 queue = [v for v in self.vertices() if v.dist < INFINITY] sort_flag = True while len(queue): if sort_flag: queue.sort(key=lambda v: v.dist) sort_flag = False v = queue.pop(0) # for each neighbor of v, see if we found a new shortest path for w, e in self[v].iteritems(): d = v.dist + e.length if d < w.dist: w.dist = d queue.append(w) sort_flag = True def shortest_path_tree2(self, source): """Finds the length of the shortest path from the source vertex to all other vertices in the graph. This length is stored on the vertices as an attribute named 'dist'. The algorithm used is Dijkstra's with a Heap used to sort/store pending nodes to be examined. """ for v in self.vertices(): v.dist = INFINITY source.dist = 0 queue = [] heapq.heappush(queue, (0, source)) while len(queue): _, v = heapq.heappop(queue) # for each neighbor of v, see if we found a new shortest path for w, e in self[v].iteritems(): d = v.dist + e.length if d < w.dist: w.dist = d heapq.heappush(queue, (d, w)) def all_pairs_floyd_warshall(self): """Finds the shortest paths between all pairs of vertices using the Floyd-Warshall algorithm. http://en.wikipedia.org/wiki/Floyd-Warshall_algorithm """ vertices = self.vertices() dist = {} for i in vertices: for j in vertices: if i is j: dist[i, j] = 0.0 else: e = self.get_edge(i, j) dist[i, j] = e.length if e else INFINITY for k in vertices: for i in vertices: for j in vertices: d_ik = dist[i, k] d_kj = dist[k, j] new_cost = d_ik + d_kj if new_cost < dist[i, j]: dist[i, j] = new_cost return dist def big_l(self): """Computes the "big-L" value for the graph as per Watts & Strogatz. L is defined as the number of edges in the shortest path between two vertices, averaged over all vertices. Uses the shortest_path_tree() method, called once for every node. """ d = {} for v in self.vertices(): self.shortest_path_tree(v, d) t = [((w, v), w.dist) for w in self.vertices() if v is not w] d.update(t) if len(d): return sum(d.values()) / float(len(d)) return 0.0 def big_l2(self): """Computes the "big-L" value for the graph as per Watts & Strogatz. L is defined as the number of edges in the shortest path between two vertices, averaged over all vertices. Uses the all_pairs_floyd_warshall() method. """ dist = self.all_pairs_floyd_warshall() vertices = self.vertices() result = [dist[i, j] for i in vertices for j in vertices if i is not j] if len(result): return sum(result) / float(len(result)) return 0.0 def big_l3(self): """Computes the "big-L" value for the graph as per Watts & Strogatz. L is defined as the number of edges in the shortest path between two vertices, averaged over all vertices. Uses the shortest_path_tree2() method, called once for every node. """ d = {} for v in self.vertices(): self.shortest_path_tree2(v) t = [((v, w), w.dist) for w in self.vertices() if v is not w] d.update(t) if len(d): return sum(d.values()) / float(len(d)) return 0.0