Unsupported operand type(s) for +: 'float' and 'tuple'

F

Francesc Segura

Hello all, I'm new to this and I'm having problems on summing two
values at python.

I get the following error:

Traceback (most recent call last):
File "C:\edge-bc (2).py", line 168, in <module>
if (costGG <= cost + T0):
TypeError: unsupported operand type(s) for +: 'float' and 'tuple'

I'm working with networkx and my program is this one:


import networkx as NX
import pylab as P
from math import exp, log
import random
try:
import matplotlib.pyplot as plt
except:
raise

##N=27
N=30
ITERATIONS = 60
T0 = 0.5
RO = 0.99
NK = 20

def costG(G):

bc = NX.edge_betweenness_centrality(G,normalized=True)
edges = NX.edges(G)

for i in range(len(edges)):
total = 0
cost = 0
factor = 1

liedges = list(edges)
linode1 = list(liedges[0])
linode2 = list(liedges[1])

distance = (((linode2[0]-linode1[0])%N)^2)+(((linode2[1]-
linode1[1])%N)^2)

edgecentrality = bc[edges]

factor = (distance-19790)*(-0.000055586)

cost = distance*edgecentrality*factor

total = total + cost

return(total)

def avedistance(G):




return (AvgDist)

def costGeasy(G):
bc = NX.edge_betweenness_centrality(G,normalized=True)
total = 0
for i in range(len(bc)):
total=total+bc.values()

return (total)


G = NX.grid_2d_graph(N,N,True)

for i in range(N):
for j in range(N):
G.add_edge((i,j),((i+1) % N ,(j+1) % N))
G.add_edge((i,j),((i-1) % N ,(j+1) % N))


NODES=NX.number_of_nodes(G)
nod=NX.nodes(G)
EDGES=NX.number_of_edges(G)
edg=NX.edges(G)

pos={}
for i in range(NODES):
pos[nod]=(nod[0]/(N*1.0),nod[1]/(N*1.0))

NX.draw(G,pos,node_color='r',node_size=20,with_labels=False,width=1)
plt.title("Inicial graph, Toroidal 27x27, degree 8")
plt.savefig("initial_grid_malla.png")
plt.show()

pos=NX.spring_layout(G,iterations=100)
NX.draw(G,pos,node_color='r',node_size=20,with_labels=False,width=1)
plt.title("Inicial graph, Toroidal 27x27, degree 8")
plt.savefig("initial_grid_tor.png")
plt.show()

initGr = G
best_graph = G
best_cost = costG(G)
average_distance = avedistance(G)
initCost = best_cost
initGHist = NX.degree_histogram(G)

##print NX.info(initGr)
##print 'Diameter = %f ' % (NX.diameter(G))
##print 'Avg Clust Coeff. NetworkX = %-6.4f ' %
(NX.average_clustering(G))
##print 'Avg Dist. NetworkX = %-6.4f ' % (average_distance)
##print 'Distribucio de Graus'
##print initGHist
##print 'Cost inicial'
##print initCost

T0 = initCost*0,1

for y in range(NK):
for x in range(ITERATIONS):
cost = costG(G)
if (cost < (best_cost)):
best_graph = G
best_cost = cost
GG = G
u = random.randint(0,NODES-1)
while GG.degree(nod) <= 1:
u = random.randint(0,NODES-1)
v = random.randint(0,GG.degree(nod)-1)
GG.remove_edge(nod,GG[nod].keys()[v])
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
adj=G.adjacency_list()
while ((nod in adj[a]) or (b == a)):
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
GG.add_edge(nod[a],nod)
while (NX.is_connected(GG) == 0):
GG = G
u = random.randint(0,NODES-1)
while GG.degree(nod) <= 1:
u = random.randint(0,NODES-1)
v = random.randint(0,GG.degree(nod)-1)
GG.remove_edge(nod,GG[nod].keys()[v])
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
adj=GG.adjacency_list()
while ((nod in adj[a]) or (b == a)):
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
GG.add_edge(nod[a],nod)

costGG = costG(GG)
if (costGG <= cost):
G = GG
else:
if (costGG <= cost + T0):
G = GG

T0 = T0 * RO
print 'IT %d' % y
print 'BEST %f ' % best_cost


best_graph = G
print NX.info(best_graph)
print 'Diameter = %f ' % (NX.diameter(best_graph))
print 'Avg Clust Coeff. NetworkX = %-6.4f ' %
(NX.average_clustering(best_graph))
average_distance = avedistance(best_graph)
print 'Avg Dist. NetworkX = %-6.4f ' % (average_distance)
print 'Distribucio de Graus'
print NX.degree_histogram(best_graph)
print 'Millor Cost'
print best_cost

NX.write_edgelist(best_graph,'optimal-graph.dat')

pos={}
for i in range(NODES):
pos[nod]=(nod[0]/(N*1.0),nod[1]/(N*1.0))


NX.draw(best_graph,pos,node_color='r',node_size=20,with_labels=False,width=1)
plt.title("Final graph, Toroidal 27x27, degree 8")
plt.savefig("final_grid_malla.png")
plt.show()

pos=NX.spring_layout(G,iterations=100)
NX.draw(best_graph,pos,node_color='r',node_size=20,with_labels=False,width=1)
plt.title("Final graph, Toroidal 27x27, degree 8")
plt.savefig("final_grid_tor.png")
plt.show()
 
T

Tim Chase

Hello all, I'm new to this and I'm having problems on summing two
values at python.

I get the following error:

Traceback (most recent call last):
File "C:\edge-bc (2).py", line 168, in<module>
if (costGG<= cost + T0):
TypeError: unsupported operand type(s) for +: 'float' and 'tuple'

I'm working with networkx and my program is this one: ....
T0 = initCost*0,1

Here, you're setting T0 to the tuple "(initCost*0, 1)" == "(0,
1)". I think you mean to use a period instead of a comma.

You then try to add that to a float (cost), and Python doesn't
like that. I wouldn't either :)

-tkc
 
F

Francesc Segura

Here, you're setting T0 to the tuple "(initCost*0, 1)" == "(0,
1)".  I think you mean to use a period instead of a comma.

You then try to add that to a float (cost), and Python doesn't
like that.  I wouldn't either :)

-tkc

Thanks a lot, I am a noob retard!
 
G

Gabriel Genellina

Hello all, I'm new to this and I'm having problems on summing two
values at python.

I get the following error:

Traceback (most recent call last):
File "C:\edge-bc (2).py", line 168, in <module>
if (costGG <= cost + T0):
TypeError: unsupported operand type(s) for +: 'float' and 'tuple'

I see Tim Chase already told you about this error. Let me make a few
comments about the rest.
try:
import matplotlib.pyplot as plt
except:
raise

I guess the above try/except was left from some earlier debugging attempt
- such an except clause is useless, just omit it.
T0 = 0.5
RO = 0.99

Perhaps those names make sense in your problem at hand, but usually I try
to use more meaningful ones. 0 and O look very similar in some fonts.
for i in range(len(edges)):
total = 0
cost = 0
factor = 1
liedges = list(edges)
linode1 = list(liedges[0])
linode2 = list(liedges[1])


list(something) creates a new list out of the elements from `something`.
You're just iterating here, so there is no need to duplicate those lists.
In addition, Python is not C: the `for` statement iterates over a
collection, you don't have to iterate over the indices and dereference
each item:

for liedges in edges:
linode1 = liedges[0]
linode2 = liedges[1]

distance = (((linode2[0]-linode1[0])%N)^2)+(((linode2[1]-
linode1[1])%N)^2)

That doesn't evaluate what you think it does. ^ is the "bitwise xor"
operator, and I bet you want **, the "power" operator.

total = total + cost
return(total)

return is not a function but a statement; those () are unnecesary and
confusing.
And I think you want to initialize total=0 *before* entering the loop;
also, initializing cost and factor is unnecesary.
def costGeasy(G):
bc = NX.edge_betweenness_centrality(G,normalized=True)
total = 0
for i in range(len(bc)):
total=total+bc.values()

return (total)


bc = NX.edge_betweenness_centrality(G,normalized=True)
values = bc.values()
total = sum(values)
return total

==>

return sum(bc.values())

pos={}
for i in range(NODES):
pos[nod]=(nod[0]/(N*1.0),nod[1]/(N*1.0))


In Python version 2.x, 1/3 evals to 0, but that's a mistake; it is fixed
in the 3.x version. If you put this line at the top of your script:
from __future__ import division
then 1/3 returns 0.3333...
When you actually want integer division, use //, like 1//3

So we can rewrite the above as:

from __future__ import division
....
for node in nod:
pos[node] = (node[0] / N, node[1] / N)

Another way, not relying on true division:

divisor = float(N)
for node in nod:
pos[node] = (node[0] / divisor, node[1] / divisor)

or even:

pos = dict((node, (node[0] / divisor, node[1] / divisor)) for node in nod)

for y in range(NK):
for x in range(ITERATIONS):
cost = costG(G)
if (cost < (best_cost)):
best_graph = G
best_cost = cost
GG = G

Again, I think this doesn't do what you think it does. GG = G means "let's
use the name GG for the object currently known as G". GG is not a "copy"
of G, just a different name for the very same object. Later operations
like GG.remove_edge(...) modify the object - and you'll see the changes in
G, and in best_graph, because those names all refer to the same object.
I think you'll benefit from reading this:
http://effbot.org/zone/python-objects.htm
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
adj=G.adjacency_list()
while ((nod in adj[a]) or (b == a)):
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
GG.add_edge(nod[a],nod)


As above, I'd avoid using indexes, take two random nodes using
random.sample instead, and avoid adjacency_list():

while True:
a, b = random.sample(nod, 2)
if b not in G[a]:
break
GG.add_edge(a, b)

(mmm, I'm unsure of the adjacency test, I've used networkx some time ago
but I don't have it available right now)
 
G

Gabriel Genellina

Hello all, I'm new to this and I'm having problems on summing two
values at python.

I get the following error:

Traceback (most recent call last):
File "C:\edge-bc (2).py", line 168, in <module>
if (costGG <= cost + T0):
TypeError: unsupported operand type(s) for +: 'float' and 'tuple'

I see Tim Chase already told you about this error. Let me make a few
comments about the rest.
try:
import matplotlib.pyplot as plt
except:
raise

I guess the above try/except was left from some earlier debugging attempt
- such an except clause is useless, just omit it.
T0 = 0.5
RO = 0.99

Perhaps those names make sense in your problem at hand, but usually I try
to use more meaningful ones. 0 and O look very similar in some fonts.
for i in range(len(edges)):
total = 0
cost = 0
factor = 1
liedges = list(edges)
linode1 = list(liedges[0])
linode2 = list(liedges[1])


list(something) creates a new list out of the elements from `something`.
You're just iterating here, so there is no need to duplicate those lists.
In addition, Python is not C: the `for` statement iterates over a
collection, you don't have to iterate over the indices and dereference
each item:

for liedges in edges:
linode1 = liedges[0]
linode2 = liedges[1]

distance = (((linode2[0]-linode1[0])%N)^2)+(((linode2[1]-
linode1[1])%N)^2)

That doesn't evaluate what you think it does. ^ is the "bitwise xor"
operator, and I bet you want **, the "power" operator.

total = total + cost
return(total)

return is not a function but a statement; those () are unnecesary and
confusing.
And I think you want to initialize total=0 *before* entering the loop;
also, initializing cost and factor is unnecesary.
def costGeasy(G):
bc = NX.edge_betweenness_centrality(G,normalized=True)
total = 0
for i in range(len(bc)):
total=total+bc.values()

return (total)


bc = NX.edge_betweenness_centrality(G,normalized=True)
values = bc.values()
total = sum(values)
return total

==>

return sum(bc.values())

pos={}
for i in range(NODES):
pos[nod]=(nod[0]/(N*1.0),nod[1]/(N*1.0))


In Python version 2.x, 1/3 evals to 0, but that's a mistake; it is fixed
in the 3.x version. If you put this line at the top of your script:
from __future__ import division
then 1/3 returns 0.3333...
When you actually want integer division, use //, like 1//3

So we can rewrite the above as:

from __future__ import division
....
for node in nod:
pos[node] = (node[0] / N, node[1] / N)

Another way, not relying on true division:

divisor = float(N)
for node in nod:
pos[node] = (node[0] / divisor, node[1] / divisor)

or even:

pos = dict((node, (node[0] / divisor, node[1] / divisor)) for node in nod)

for y in range(NK):
for x in range(ITERATIONS):
cost = costG(G)
if (cost < (best_cost)):
best_graph = G
best_cost = cost
GG = G

Again, I think this doesn't do what you think it does. GG = G means "let's
use the name GG for the object currently known as G". GG is not a "copy"
of G, just a different name for the very same object. Later operations
like GG.remove_edge(...) modify the object - and you'll see the changes in
G, and in best_graph, because those names all refer to the same object.
I think you'll benefit from reading this:
http://effbot.org/zone/python-objects.htm
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
adj=G.adjacency_list()
while ((nod in adj[a]) or (b == a)):
a = random.randint(0,NODES-1)
b = random.randint(0,NODES-1)
GG.add_edge(nod[a],nod)


As above, I'd avoid using indexes, take two random nodes using
random.sample instead, and avoid adjacency_list():

while True:
a, b = random.sample(nod, 2)
if b not in G[a]:
break
GG.add_edge(a, b)

(mmm, I'm unsure of the adjacency test, I've used networkx some time ago
but I don't have it available right now)
 
T

Terry Reedy

Hello all, I'm new to this and I'm having problems on summing two
values at python.

I get the following error:

Traceback (most recent call last):
File "C:\edge-bc (2).py", line 168, in<module>
if (costGG<= cost + T0):
TypeError: unsupported operand type(s) for +: 'float' and 'tuple'

I'm working with networkx and my program is this one:

[snip about 100 lines]

Before posting code that does not work, it is a good idea to reduce to
to some minimum needed to exhibit the problem. If you had done that, you
may well have found the answer.

In this specific case, you should have searched for all lines making
assignments to the names causing a problem. This is easier to do by eye
with a minimal example. Or, use a search function for 'T0 =' (or 'T0='
if needed, but it is not) with Find Next and the second hit is the
offending line ("T0 = initCost*0,1").
 
T

Thomas Rachel

Am 11.06.2011 03:02 schrieb Gabriel Genellina:
Perhaps those names make sense in your problem at hand, but usually I try
to use more meaningful ones.

Until here we agree.
0 and O look very similar in some fonts.

That is right - but who would use such fonts for programming?


Thomas
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
474,045
Messages
2,570,389
Members
47,052
Latest member
ketan

Latest Threads

Top