def vertex_partitioning(sc, nodes, edges, num_partition=4):
"""
implementation of vertex partitioning Kruskal's algorithm
:param nodes: nodes for input.
:param edges: edges for input.
:param num_partition: number of partitions.
:return:
"""
# define function for calculating combinations of different vertex partitions
def combine(iterator):
for i in iterator:
if i[0] < i[1]:
yield i[0] + i[1]
vertices = sc.parallelize(shuffle(nodes), num_partition).glom()
vertices = vertices.cartesian(vertices) \
.mapPartitions(combine, preservesPartitioning=False)
# parallelize
vertices = sc.parallelize(vertices.collect(), num_partition)
# define function for calculating local MSTs
def local_kruskal(iterator):
for subset in iterator:
for edge in kruskal(nodes=set(subset), edges=edges):
yield edge
# calculate local MSTs
subtrees = vertices.mapPartitions(local_kruskal).distinct().collect()
# calculate the global MST
return kruskal(nodes=nodes, edges=subtrees)
def edge_partitioning(sc, nodes, edges, num_partition=4):
"""
implementation of edge partitioning Kruskal's algorithm
:param nodes: nodes for input.
:param edges: edges for input.
:param num_partition: number of partitions.
:return:
"""
# parallelize
edges = sc.parallelize(shuffle(edges), num_partition)
# define function for calculating local MSTs
def local_kruskal(iterator):
for edge in kruskal(nodes=nodes, edges=iterator):
yield edge
# calculate local MSTs
subtrees = edges.mapPartitions(local_kruskal).collect()
# calculate the global MST
return kruskal(nodes=nodes, edges=subtrees)
graphdata = f.readline().rstrip('\n').split(
' '
) #read first line to get graph data, strip newline, and put data into 2-element list
graphdata = map(int, graphdata)
Input = []
for line in f:
sublist = line.rstrip('\n').split(
' '
) # Remove newline character and split 2-column row into 2-element list
Input.append(map(int, sublist)) # Append 4-element rowlist to Input list
print graphdata
#T = prim(Input,1,graphdata)
uf = kruskal(Input, 1, graphdata, 4)
ufgroup = uf.groups()
print uf.groups()
fn = open('output.txt', 'wb')
pickle.dump(uf.groups(), fn)
print len(uf.groups())
minspace = float("inf")
for i in range(1):
group1 = [x for x in Input if x[0] in ufgroup[i]]
group2 = [x for x in group1 if x[1] not in ufgroup[i]]
print group2
minspacetemp = min(group2, key=itemgetter(2))
if minspacetemp < minspace:
minspace = minspacetemp
def local_kruskal(iterator):
for subset in iterator:
for edge in kruskal(nodes=set(subset), edges=edges):
yield edge
def local_kruskal(iterator):
for edge in kruskal(nodes=nodes, edges=iterator):
yield edge
import pickle
f = open('clustering1.txt','r')
graphdata = f.readline().rstrip('\n').split(' ') #read first line to get graph data, strip newline, and put data into 2-element list
graphdata = map(int,graphdata)
Input = []
for line in f:
sublist = line.rstrip('\n').split(' ') # Remove newline character and split 2-column row into 2-element list
Input.append(map(int,sublist)) # Append 4-element rowlist to Input list
print graphdata
#T = prim(Input,1,graphdata)
uf = kruskal(Input,1,graphdata,4)
ufgroup = uf.groups()
print uf.groups()
fn = open('output.txt', 'wb')
pickle.dump(uf.groups(), fn)
print len(uf.groups())
minspace = float("inf")
for i in range(1):
group1 = [x for x in Input if x[0] in ufgroup[i] ]
group2 = [x for x in group1 if x[1] not in ufgroup[i]]
print group2
minspacetemp = min(group2, key = itemgetter(2))
if minspacetemp < minspace:
minspace = minspacetemp