Using accumarray to sum all values associated with a given node

Given a graph with the following edges

edges = [1 2; 2 3; 3 1; 2 4; 4 5; 4 1];

I have some data associated with each edge

data=[0.276

0.679

0.655

0.162

0.118

0.333];

For a given node, I want to sum the data values corresponding to all edges it's connected to. How do I do this? accumarray looks promising, but this gives six nonzero values, when there should just be 5 (and I don't know how to distinguish which one corresponds to which node).

sum_data = accumarray(edges,data);

Answer 1

[Matt J]

One way:

edges = [1 2; 2 3; 3 1; 2 4; 4 5; 4 1];

data=[0.276
0.679
0.655
0.162
0.118
0.333];

G=graph(edges(:,1), edges(:,2), data);

out=sum(adjacency(G,'weighted'),2)

out = 
   (1,1)       1.2640
   (2,1)       1.1170
   (3,1)       1.3340
   (4,1)       0.6130
   (5,1)       0.1180

[L'O.G.]

Thanks, is there a way to do it without computing the graph explicitly?

[Matt J]

What do you mean by "compute the graph explicitly"? There are no computations involved in the creation of the graph object G. It just stores the table of edges and weights that you already have.

[L'O.G.]

Ah ok, I was just wondering if I could've used accumarray after all. Thank you!

[Matt J]

You could (see my other answer), but I'm pretty sure that's what the adjacency() command is already doing internally.

Answer 2

[Matt J]

edges = [1 2; 2 3; 3 1; 2 4; 4 5; 4 1];
data=[0.276
0.679
0.655
0.162
0.118
0.333];

m=max(edges(:));

A=accumarray(edges,data,[m,m],[],0,1);

out=sum(A+A.',2) %assume it's a non-directed graph

out = 
   (1,1)       1.2640
   (2,1)       1.1170
   (3,1)       1.3340
   (4,1)       0.6130
   (5,1)       0.1180