def GenTasks(tname,step,args):
"""
generate the tasks
args, arguments as a string
nnodes, number of nodes
tsize, number of nodes per task
dis, degree distribution
"""
d = simplejson.loads(args[0])
nnodes = d["nnodes"]
tsize = d["tsize"]
dis = d["dis"]
ns = 0
tc = 0
while ns < nnodes:
tname = "G" + str(tc)
tc += 1
ne = ns + tsize
if ne > nnodes:
ne = nnodes
tinfo = "%d\t%d\t%s" % (ns, ne-1, dis)
print tname, tinfo
ns = ne
comm.mroute("T0", tname, tinfo)
def GenTasks(tname, step, args):
"""
generate the tasks
args, arguments as a string
nnodes, number of nodes
tsize, number of nodes per task
dis, degree distribution
"""
d = simplejson.loads(args[0])
nnodes = d["nnodes"]
tsize = d["tsize"]
dis = d["dis"]
ns = 0
tc = 0
while ns < nnodes:
tname = "G" + str(tc)
tc += 1
ne = ns + tsize
if ne > nnodes:
ne = nnodes
tinfo = "%d\t%d\t%s" % (ns, ne - 1, dis)
print tname, tinfo
ns = ne
comm.mroute("T0", tname, tinfo)
def GenGraph(tname,step,args):
"""
generate the graph edges
args, arguments as a string
"""
# extract the stubs from the args
# iterate through the input queue and add new items to the stub list
stubs = []
for item in args:
l = simplejson.loads(item)
stubs.extend(l)
#print stubs
print tname,stubs
# randomize the items
random.shuffle(stubs)
#print tname + "-r",stubs
# get the pairs
pairs = zip(stubs[::2], stubs[1::2])
print tname,pairs
# distribute the stubs randomly to the tasks
tsize = comm.mgetconfig("tsize")
# get edges for a specific task
edges = {}
for pair in pairs:
esrc = pair[0]
edst = pair[1]
# add the edge twice for both directions
tdst = TaskId(esrc, tsize)
if not edges.has_key(tdst):
edges[tdst] = []
l = [esrc, edst]
edges[tdst].append(l)
tdst = TaskId(edst, tsize)
if not edges.has_key(tdst):
edges[tdst] = []
l = [edst, esrc]
edges[tdst].append(l)
print tname,edges
for key, value in edges.iteritems():
tdst = "E%s" % (str(key))
targs = simplejson.dumps(value)
print tdst,targs
comm.mroute(tname,tdst,targs)
def GenGraph(tname, step, args):
"""
generate the graph edges
args, arguments as a string
"""
# extract the stubs from the args
# iterate through the input queue and add new items to the stub list
stubs = []
for item in args:
l = simplejson.loads(item)
stubs.extend(l)
#print stubs
print tname, stubs
# randomize the items
random.shuffle(stubs)
#print tname + "-r",stubs
# get the pairs
pairs = zip(stubs[::2], stubs[1::2])
print tname, pairs
# distribute the stubs randomly to the tasks
tsize = comm.mgetconfig("tsize")
# get edges for a specific task
edges = {}
for pair in pairs:
esrc = pair[0]
edst = pair[1]
# add the edge twice for both directions
tdst = TaskId(esrc, tsize)
if not edges.has_key(tdst):
edges[tdst] = []
l = [esrc, edst]
edges[tdst].append(l)
tdst = TaskId(edst, tsize)
if not edges.has_key(tdst):
edges[tdst] = []
l = [edst, esrc]
edges[tdst].append(l)
print tname, edges
for key, value in edges.iteritems():
tdst = "E%s" % (str(key))
targs = simplejson.dumps(value)
print tdst, targs
comm.mroute(tname, tdst, targs)
def GenStubs(tname, step, args):
"""
determine degrees for all the nodes, generate the stubs and distribute them
args, arguments as a string
"""
argwords = args[0]
largs = argwords.split("\t")
#print largs
ns = int(largs[0])
ne = int(largs[1])
dis = largs[2]
print "*task* %s %d %d %d %s" % (tname, step, ns, ne, dis)
# determine node degrees
i = ns
ddeg = {}
while i <= ne:
deg = StdDist(distmean, distvar)
#deg = 3
ddeg[i] = deg
print "*task* %s %d, node %s, degree %s" % (tname, step, str(i),
str(deg))
i += 1
print ddeg
# distribute the stubs randomly to the tasks
ntasks = comm.mgetconfig("tasks")
print "__tasks__ %s\t%s" % (tname, str(ntasks))
# one item per task, each task has a list of stubs
dstubs = {}
for key, value in ddeg.iteritems():
for i in range(0, value):
t = int(random.random() * ntasks)
if not dstubs.has_key(t):
dstubs[t] = []
dstubs[t].append(key)
for key, value in dstubs.iteritems():
tdst = "S%s" % (str(key))
targs = simplejson.dumps(value)
print tdst, targs
comm.mroute(tname, tdst, targs)
def GenStubs(tname,step,args):
"""
determine degrees for all the nodes, generate the stubs and distribute them
args, arguments as a string
"""
argwords = args[0]
largs = argwords.split("\t")
#print largs
ns = int(largs[0])
ne = int(largs[1])
dis = largs[2]
print "*task* %s %d %d %d %s" % (tname, step, ns, ne, dis)
# determine node degrees
i = ns
ddeg = {}
while i <= ne:
deg = StdDist(distmean,distvar)
#deg = 3
ddeg[i] = deg
print "*task* %s %d, node %s, degree %s" % (tname, step, str(i), str(deg))
i += 1
print ddeg
# distribute the stubs randomly to the tasks
ntasks = comm.mgetconfig("tasks")
print "__tasks__ %s\t%s" % (tname, str(ntasks))
# one item per task, each task has a list of stubs
dstubs = {}
for key,value in ddeg.iteritems():
for i in range(0,value):
t = int(random.random() * ntasks)
if not dstubs.has_key(t):
dstubs[t] = []
dstubs[t].append(key)
for key, value in dstubs.iteritems():
tdst = "S%s" % (str(key))
targs = simplejson.dumps(value)
print tdst,targs
comm.mroute(tname,tdst,targs)
def GetDist(tname,step,args):
"""
find the node distance
args, arguments as a string
"""
print "GetDist", tname
task = tname
tsize = comm.mgetconfig("tsize")
# get the initial arguments: starting node and its neighbors
dinit = simplejson.loads(args[0])
node = dinit["node"]
nbrlist = dinit["nbrs"]
# no visited nodes yet, first iteration
visited = {}
distance = 0
print "*distance*", tname, node, distance
visited[node] = distance
while True:
# process the new neighbors
distance += 1
newnodes = []
# process all the input elements
for arg in args:
dinit = simplejson.loads(arg)
srcnode = dinit["node"]
nbrlist = dinit["nbrs"]
for item in nbrlist:
# skip nodes already visited
if item in visited:
continue
# add new nodes to the visited nodes and the new nodes
print "*distance*", tname, item, distance
visited[item] = distance
newnodes.append(item)
# done, if there are no more new nodes
if len(newnodes) <= 0:
break
# send new visited nodes to the graph nodes to find their neighbors
# collect nodes for the same task
dtasks = {}
for ndst in newnodes:
tn = TaskId(ndst,tsize)
if not dtasks.has_key(tn):
dtasks[tn] = []
dtasks[tn].append(ndst)
print "dtasks", dtasks
# send the messages
for tn,args in dtasks.iteritems():
dmsg = {}
dmsg["task"] = task
dmsg["nodes"] = args
tdst = "E%s" % (str(tn))
targs = simplejson.dumps(dmsg)
print tdst,targs
comm.mroute(tname,tdst,targs)
# wait for another iteration
yield
# get the input queue
args = comm.mgetargs(tname)
# end of while, repeat the loop
#print "*finish*", tname, visited
dist = {}
for key,value in visited.iteritems():
if not dist.has_key(value):
dist[value] = []
dist[value].append(key)
for key,value in dist.iteritems():
value.sort()
#print "*finish1*", tname, dist
distance = 0
compsize = 0
while dist.has_key(distance):
print "*dist*", distance, len(dist[distance])
#print "*dist*", distance, len(dist[distance]), dist[distance]
compsize += len(dist[distance])
distance += 1
print "*distall*", distance, compsize
return
def GenNbr(tname,step,args):
"""
generate the graph neighbors
args, arguments as a string
"""
task = tname
# extract neighbors from the args
# iterate through the input queue and add new items to the neighbor list
edges = []
for item in args:
l = simplejson.loads(item)
edges.extend(l)
print edges
print tname,edges
# collect neighbors for each node
nbrs = {}
for item in edges:
src = item[0]
dst = item[1]
if not nbrs.has_key(src):
nbrs[src] = set()
nbrs[src].add(dst)
for node, edges in nbrs.iteritems():
print tname, node, edges
# select a random node for stats
#nsel = random.choice(list(nbrs.keys()))
nsel = list(nbrs.keys())[0]
print tname, "sel", nsel, list(nbrs.keys())
# send the node and its neighbors to the distance task
tdst = "D%s" % (str(nsel))
dmsg = {}
dmsg["node"] = nsel
dmsg["nbrs"] = list(nbrs[nsel])
targs = simplejson.dumps(dmsg)
print tdst,targs
comm.mroute(tname,tdst,targs)
while True:
yield
args = comm.mgetargs(tname)
if args == None:
continue
# iterate through the input queue, get the nodes, report neighbors
for item in args:
d = simplejson.loads(item)
tdst = d["task"]
nodes = d["nodes"]
for node in nodes:
dmsg = {}
dmsg["node"] = node
dmsg["nbrs"] = list(nbrs[node])
targs = simplejson.dumps(dmsg)
#tdst = "D%s" % (str(node))
print tdst,targs
comm.mroute(tname,tdst,targs)
d = {}
d["type"] = "iter"
d["def" ] = GetDist
dispatch["D"] = d
comm.msetdispatch(dispatch)
d = {}
d["nnodes"] = nnodes
d["tsize"] = tsize
d["dis"] = "std"
targs = simplejson.dumps(d)
#random.seed(0)
comm.mroute("00","T0",targs)
# generate the tasks and assign nodes to them
#GenTasks(nnodes, tsize, "std")
comm.mexec()
# generate node degrees and distribute stubs to tasks
#comm.mexec(GenStubs)
comm.mexec()
# generate the random graph
#comm.mexec(GenGraph)
comm.mexec()
# generate the graph statistics
#comm.mexec(GenNbr)
def GetDist(tname, step, args):
"""
find the node distance
args, arguments as a string
"""
print "GetDist", tname
task = tname
tsize = comm.mgetconfig("tsize")
# get the initial arguments: starting node and its neighbors
dinit = simplejson.loads(args[0])
node = dinit["node"]
nbrlist = dinit["nbrs"]
# no visited nodes yet, first iteration
visited = {}
distance = 0
print "*distance*", tname, node, distance
visited[node] = distance
while True:
# process the new neighbors
distance += 1
newnodes = []
# process all the input elements
for arg in args:
dinit = simplejson.loads(arg)
srcnode = dinit["node"]
nbrlist = dinit["nbrs"]
for item in nbrlist:
# skip nodes already visited
if item in visited:
continue
# add new nodes to the visited nodes and the new nodes
print "*distance*", tname, item, distance
visited[item] = distance
newnodes.append(item)
# done, if there are no more new nodes
if len(newnodes) <= 0:
break
# send new visited nodes to the graph nodes to find their neighbors
# collect nodes for the same task
dtasks = {}
for ndst in newnodes:
tn = TaskId(ndst, tsize)
if not dtasks.has_key(tn):
dtasks[tn] = []
dtasks[tn].append(ndst)
print "dtasks", dtasks
# send the messages
for tn, args in dtasks.iteritems():
dmsg = {}
dmsg["task"] = task
dmsg["nodes"] = args
tdst = "E%s" % (str(tn))
targs = simplejson.dumps(dmsg)
print tdst, targs
comm.mroute(tname, tdst, targs)
# wait for another iteration
yield
# get the input queue
args = comm.mgetargs(tname)
# end of while, repeat the loop
#print "*finish*", tname, visited
dist = {}
for key, value in visited.iteritems():
if not dist.has_key(value):
dist[value] = []
dist[value].append(key)
for key, value in dist.iteritems():
value.sort()
#print "*finish1*", tname, dist
distance = 0
compsize = 0
while dist.has_key(distance):
print "*dist*", distance, len(dist[distance])
#print "*dist*", distance, len(dist[distance]), dist[distance]
compsize += len(dist[distance])
distance += 1
print "*distall*", distance, compsize
return
def GenNbr(tname, step, args):
"""
generate the graph neighbors
args, arguments as a string
"""
task = tname
# extract neighbors from the args
# iterate through the input queue and add new items to the neighbor list
edges = []
for item in args:
l = simplejson.loads(item)
edges.extend(l)
print edges
print tname, edges
# collect neighbors for each node
nbrs = {}
for item in edges:
src = item[0]
dst = item[1]
if not nbrs.has_key(src):
nbrs[src] = set()
nbrs[src].add(dst)
for node, edges in nbrs.iteritems():
print tname, node, edges
# select a random node for stats
#nsel = random.choice(list(nbrs.keys()))
nsel = list(nbrs.keys())[0]
print tname, "sel", nsel, list(nbrs.keys())
# send the node and its neighbors to the distance task
tdst = "D%s" % (str(nsel))
dmsg = {}
dmsg["node"] = nsel
dmsg["nbrs"] = list(nbrs[nsel])
targs = simplejson.dumps(dmsg)
print tdst, targs
comm.mroute(tname, tdst, targs)
while True:
yield
args = comm.mgetargs(tname)
if args == None:
continue
# iterate through the input queue, get the nodes, report neighbors
for item in args:
d = simplejson.loads(item)
tdst = d["task"]
nodes = d["nodes"]
for node in nodes:
dmsg = {}
dmsg["node"] = node
dmsg["nbrs"] = list(nbrs[node])
targs = simplejson.dumps(dmsg)
#tdst = "D%s" % (str(node))
print tdst, targs
comm.mroute(tname, tdst, targs)
d = {}
d["type"] = "iter"
d["def"] = GetDist
dispatch["D"] = d
comm.msetdispatch(dispatch)
d = {}
d["nnodes"] = nnodes
d["tsize"] = tsize
d["dis"] = "std"
targs = simplejson.dumps(d)
#random.seed(0)
comm.mroute("00", "T0", targs)
# generate the tasks and assign nodes to them
#GenTasks(nnodes, tsize, "std")
comm.mexec()
# generate node degrees and distribute stubs to tasks
#comm.mexec(GenStubs)
comm.mexec()
# generate the random graph
#comm.mexec(GenGraph)
comm.mexec()
# generate the graph statistics
#comm.mexec(GenNbr)