遍历图中所有可用路径
Posted
技术标签:
【中文标题】遍历图中所有可用路径【英文标题】:Traversing through all the available paths in a diagraph 【发布时间】:2018-06-05 15:26:34 【问题描述】:有一个带有如下数字的图形结构。
在 python 的图形对象中加载这个结构。我把它做成了一个多行字符串,如下所示。
myString='''1
2 3
4 5 6
7 8 9 10
11 12 13 14 15'''
将其表示为列表列表。
>>> listofLists=[ list(map(int,elements.split())) for elements in myString.strip().split("\n")]
>>> print(listofLists)
[[1], [2, 3], [4, 5, 6], [7, 8, 9, 10], [11, 12, 13, 14, 15]]
在 python 中使用下面的节点和边类创建图形结构
节点类,它需要位置作为一个元组和一个值 示例:元素及其位置、值
1 --- position (0,0) and value is 1
2 --- position (1,0) and value is 2
3 --- position (1,1) and value is 3
节点类
class node(object):
def __init__(self,position,value):
'''
position : gives the position of the node wrt to the test string as a tuple
value : gives the value of the node
'''
self.value=value
self.position=position
def getPosition(self):
return self.position
def getvalue(self):
return self.value
def __str__(self):
return 'P:'+str(self.position)+' V:'+str(self.value)
edge 类在两个节点之间创建一条边。
class edge(object):
def __init__(self,src,dest):
'''src and dest are nodes'''
self.src = src
self.dest = dest
def getSource(self):
return self.src
def getDestination(self):
return self.dest
#return the destination nodes value as the weight
def getWeight(self):
return self.dest.getvalue()
def __str__(self):
return (self.src.getPosition(),)+'->'+(self.dest.getPosition(),)
有向图类如下。图结构构建为字典邻接列表。 node1:[node2,node3],node2:[node3,node4]......
class Diagraph(object):
'''the edges is a dict mapping node to a list of its destination'''
def __init__(self):
self.edges =
'''Adds the given node as a key to the dict named edges '''
def addNode(self,node):
if node in self.edges:
raise ValueError('Duplicate node')
else:
self.edges[node]=[]
'''addEdge accepts and edge class object checks if source and destination node are present in the graph '''
def addEdge(self,edge):
src = edge.getSource()
dest = edge.getDestination()
if not (src in self.edges and dest in self.edges):
raise ValueError('Node not in graph')
self.edges[src].append(dest)
'''getChildrenof returns all the children of the node'''
def getChildrenof(self,node):
return self.edges[node]
'''to check whether a node is present in the graph or not'''
def hasNode(self,node):
return node in self.edges
'''rootNode returns the root node i.e node at position (0,0)'''
def rootNode(self):
for keys in self.edges:
return keys if keys.getPosition()==(0,0) else 'No Root node for this graph'
一个创建和返回图形对象的函数。
def createmygraph(testString):
'''input is a multi-line string'''
#create a list of lists from the string
listofLists=[ list(map(int,elements.split())) for elements in testString.strip().split("\n")]
y = Diagraph()
nodeList = []
# create all the nodes and store it in a list nodeList
for i in range(len(listofLists)):
for j in range(len(listofLists)):
if i<=j:
mynode=node((j,i),listofLists[j][i])
nodeList.append(mynode)
y.addNode(mynode)
# create all the edges
for srcNode in nodeList:
# iterate through all the nodes again and form a logic add the edges
for destNode in nodeList:
#to add the immediate down node eg : add 7 (1,0) to 3 (0,0) , add 2 (2,0) to 7 (1,0)
if srcNode.getPosition()[0]==destNode.getPosition()[0]-1 and srcNode.getPosition()[1]==destNode.getPosition()[1]-1:
y.addEdge(edge(srcNode,destNode))
#to add the bottom right node eg :add 4 (1,1) to 3 (0,0)
if srcNode.getPosition()[0]==destNode.getPosition()[0]-1 and srcNode.getPosition()[1]==destNode.getPosition()[1]:
y.addEdge(edge(srcNode,destNode))
return y
如何列出两个节点之间所有可用的路径。特别是 1---->11 , 1---->12 , 1---->13 , 1---- >14 , 1---->15 对于这种情况,我尝试了左优先深度优先方法。但它无法获得路径。
def leftFirstDepthFirst(graph,start,end,path,valueSum):
#add input start node to the path
path=path+[start]
#add the value to the valueSum variable
valueSum+=start.getvalue()
print('Current Path ',printPath(path))
print('The sum is ',valueSum)
# return if start and end node matches.
if start==end:
print('returning as start and end have matched')
return path
#if there are no further destination nodes, move up a node in the path and remove the current element from the path list.
if not graph.getChildrenof(start):
path.pop()
valueSum=valueSum-start.getvalue()
return leftFirstDepthFirst(graph,graph.getChildrenof(path[-1])[1],end,path,valueSum)
else:
for aNode in graph.getChildrenof(start):
return leftFirstDepthFirst(graph,aNode,end,path,valueSum)
print('no further path to explore')
测试代码。
#creating a graph object with given string
y=createmygraph(myString)
函数返回终端节点,如 11、12、13、14、15。
def fetchTerminalNode(graph,position):
terminalNode=[]
for keys in graph.edges:
if not graph.edges[keys]:
terminalNode.append(keys)
return terminalNode[position]
运行深度优先左前函数。
source=y.rootNode() # element at position (0,0)
destination=fetchTerminalNode(y,1) #ie. number 12
print('Path from ',start ,'to ',destination)
xyz=leftFirstDepthFirst(y,source,destination,[],0)
为元素 11 和 12 获取路径,但不是为 13、14 或 15 获取路径。即 destination=fetchTerminalNode(y,2)
不起作用。请任何人提出解决此问题的方法。
【问题讨论】:
【参考方案1】:给定一个tree
tree = \
[ [1]
, [2, 3]
, [4, 5, 6]
, [7, 8, 9, 10]
, [11, 12, 13, 14, 15]
]
还有一个traverse
函数
def traverse (tree):
def loop (path, t = None, *rest):
if not rest:
for x in t:
yield path + [x]
else:
for x in t:
yield from loop (path + [x], *rest)
return loop ([], *tree)
遍历所有路径...
for path in traverse (tree):
print (path)
# [ 1, 2, 4, 7, 11 ]
# [ 1, 2, 4, 7, 12 ]
# [ 1, 2, 4, 7, 13 ]
# [ 1, 2, 4, 7, 14 ]
# [ 1, 2, 4, 7, 15 ]
# [ 1, 2, 4, 8, 11 ]
# [ 1, 2, 4, 8, 12 ]
# ...
# [ 1, 3, 6, 9, 15 ]
# [ 1, 3, 6, 10, 11 ]
# [ 1, 3, 6, 10, 12 ]
# [ 1, 3, 6, 10, 13 ]
# [ 1, 3, 6, 10, 14 ]
# [ 1, 3, 6, 10, 15 ]
或者将所有路径收集到一个列表中
print (list (traverse (tree)))
# [ [ 1, 2, 4, 7, 11 ]
# , [ 1, 2, 4, 7, 12 ]
# , [ 1, 2, 4, 7, 13 ]
# , [ 1, 2, 4, 7, 14 ]
# , [ 1, 2, 4, 7, 15 ]
# , [ 1, 2, 4, 8, 11 ]
# , [ 1, 2, 4, 8, 12 ]
# , ...
# , [ 1, 3, 6, 9, 15 ]
# , [ 1, 3, 6, 10, 11 ]
# , [ 1, 3, 6, 10, 12 ]
# , [ 1, 3, 6, 10, 13 ]
# , [ 1, 3, 6, 10, 14 ]
# , [ 1, 3, 6, 10, 15 ]
# ]
在上面,我们使用了生成器,它是 Python 中的高级功能。也许您想了解如何使用更原始的功能实现解决方案...
我们在这里寻找的通用机制是 list monad,它捕捉了模糊计算的概念;一些可能返回多个值的过程。
Python 已经提供了列表以及使用[]
构造它们的方法。我们只需要提供绑定操作,下面命名为flat_map
def flat_map (f, xs):
return [ y for x in xs for y in f (x) ]
def traverse (tree):
def loop (path, t = None, *rest):
if not rest:
return map (lambda x: path + [x], t)
else:
return flat_map (lambda x: loop (path + [x], *rest), t)
return loop ([], *tree)
print (traverse (tree))
# [ [ 1, 2, 4, 7, 11 ]
# , [ 1, 2, 4, 7, 12 ]
# , [ 1, 2, 4, 7, 13 ]
# , ... same output as above ...
# ]
哦,Python 有一个内置的product
函数,它恰好可以按照我们的需要工作。唯一的区别是路径将输出为元组()
,而不是列表[]
from itertools import product
tree = \
[ [1]
, [2, 3]
, [4, 5, 6]
, [7, 8, 9, 10]
, [11, 12, 13, 14, 15]
]
for path in product (*tree):
print (path)
# (1, 2, 4, 7, 11)
# (1, 2, 4, 7, 12)
# (1, 2, 4, 7, 13)
# (1, 2, 4, 7, 14)
# (1, 2, 4, 7, 15)
# ... same output as above ...
在您的程序中,您尝试通过各种类node
和edge
和diagraph
来抽象这种机制。最终,您可以随心所欲地构建您的程序,但要知道它不需要比我们在此处编写的更复杂。
更新
正如@user3386109 在 cmets 中指出的那样,我上面的程序会生成路径,就好像每个父节点都连接到 所有 子节点一样。然而,这是一个错误,因为您的图表显示父母只连接到 相邻 孩子。我们可以通过修改我们的程序来解决这个问题——以下更改为粗体
def traverse (tree):
def loop (path, i, t = None, *rest):
if not rest:
for (j,x) in enumerate (t):
if i == j or i + 1 == j:
yield path + [x]
else:
for (j,x) in enumerate (t):
if i == j or i + 1 == j:
yield from loop (path + [x], j, *rest)
return loop ([], 0, *tree)
在上面,我们使用索引i
和j
来确定哪些节点是“相邻的”,但它把我们的loop
弄得乱七八糟。另外,新代码看起来像是引入了一些重复。给这个意图一个名字adjacencies
让我们的函数更干净
def traverse (tree):
def adjacencies (t, i):
for (j, x) in enumerate (t):
if i == j or i + 1 == j:
yield (j, x)
def loop (path, i, t = None, *rest):
if not rest:
for (_, x) in adjacencies (t, i):
yield path + [x]
else:
for (j, x) in adjacencies (t, i):
yield from loop (path + [x], j, *rest)
return loop ([], 0, *tree)
使用是一样的,但是这次我们得到的是原题中指定的输出
for path in traverse (tree):
print (path)
# [1, 2, 4, 7, 11]
# [1, 2, 4, 7, 12]
# [1, 2, 4, 8, 12]
# [1, 2, 4, 8, 13]
# [1, 2, 5, 8, 12]
# [1, 2, 5, 8, 13]
# [1, 2, 5, 9, 13]
# [1, 2, 5, 9, 14]
# [1, 3, 5, 8, 12]
# [1, 3, 5, 8, 13]
# [1, 3, 5, 9, 13]
# [1, 3, 5, 9, 14]
# [1, 3, 6, 9, 13]
# [1, 3, 6, 9, 14]
# [1, 3, 6, 10, 14]
# [1, 3, 6, 10, 15]
这个简单的adjancies
函数之所以起作用,是因为您的输入数据是统一且有效的。您可以通过对图像中的路径进行颜色编码来清楚地显示索引i
和i + 1
。我们永远不必担心 index-out-of-bounds 错误,因为您可以看到 i + 1
永远不会在没有子节点的节点上计算(即最后一行)。如果您要指定无效数据,traverse
不保证有效结果。
【讨论】:
此答案中的每个输出都包含1, 2, 4, 7, 13
行。从问题顶部的图像来看,这是不可能的,因为 7 与 13 没有优势。
我好像忽略了这个细节!我的想法的核心是相同的,但是需要进行更改以仅绑定相邻的孩子而不是所有孩子。如果我有时间,我稍后会重新访问。
@user3386109 我进行了更新以解决您的评论:D
感谢您的详细解释,上面的代码提供了从开始节点1到结束节点11、12、13、14、15的所有可能路径。但是,我一直在寻找一个使用图论来回答问题的答案。我同意but know that it doesn't need to be more complex than we have written it here.
我正在学习具有图算法的 edx 课程 6.00.2x,尽管我可以在这里应用它。如果您能建议一种使用图表来解决这个问题的方法,那将会很有帮助。【参考方案2】:
我想出的 PFB 函数使用 breathfirstSearch 打印根节点和端节点之间的所有可用路径。
谷歌Colab/github链接
def breathfirstalgo(graph,tempPaths,finalPath):
## iterates over all the lists inside the tempPaths and checks if there are child nodes to its last node.
condList=[graph.getChildrenof(apartList[-1]) for apartList in tempPaths if graph.getChildrenof(apartList[-1])]
tempL=[]
if condList:
for partialList in tempPaths:
#get the children of the last element of partialList
allchild=graph.getChildrenof(partialList[-1])
if allchild:
noOfChild=len(allchild)
#create noOfChild copies of the partialList
newlist=[partialList[:] for _ in range(noOfChild)]
#append the a child element to the new list
for i in range(noOfChild):
newlist[i].append(allchild[i])
#append each list to the temp list tempL
for alist in newlist:
tempL.append(alist)
else:
pass
#after completion of the for loop i.e iterate through 1 level
return breathfirstalgo(graph,tempL,finalPath)
else:
#append all the lists from tempPaths to finalPath that will be returned
for completePath in tempPaths:
finalPath.append(completePath)
return finalPath
如下测试呼吸优先搜索解决方案。
mygraph=createmygraph(myString)
print('The graph object is ',mygraph)
print('The root node is ',mygraph.rootNode())
#print(mygraph)
all_list=breathfirstalgo(mygraph,tempPaths=[[mygraph.rootNode()]],finalPath=[])
print('alllist is ')
for idx,partlist in enumerate(all_list):
print(printPath(partlist))
将节点类的__str__
修改为str(self.value)
后输出如下
The graph object is <__main__.Diagraph object at 0x7f08e5a3d128>
The root node is 1
alllist is
-->1-->2-->4-->7-->11
-->1-->2-->4-->7-->12
-->1-->2-->4-->8-->12
-->1-->2-->4-->8-->13
-->1-->2-->5-->8-->12
-->1-->2-->5-->8-->13
-->1-->2-->5-->9-->13
-->1-->2-->5-->9-->14
-->1-->3-->5-->8-->12
-->1-->3-->5-->8-->13
-->1-->3-->5-->9-->13
-->1-->3-->5-->9-->14
-->1-->3-->6-->9-->13
-->1-->3-->6-->9-->14
-->1-->3-->6-->10-->14
-->1-->3-->6-->10-->15
【讨论】:
以上是关于遍历图中所有可用路径的主要内容,如果未能解决你的问题,请参考以下文章