Why do I get a wrong answer when I use FIND function in this case?

I have a graph structure "G_dmg.Nodes.Load" which has typically 40 "-inf" values and 10 other numbers ~= -inf.

In my script:

inActive = find(G_dmg.Nodes.Load==-inf);
for ii = 1:length(inActive)
    b = neighbors(G_dmg,inActive(ii)); 
   idx = find(G_dmg.Nodes.Load(b)~= -inf); 
   Load = G_dmg.Nodes.Load(idx);
end

So, I'm trying to find the indices of inActive nodes' neighbors that do not have -inf Load values. FIND function is not giving me back the correct indices.

say b = [10 11 2 3 4] and the 10th, 11th entries in G_dmg.Nodes.Load ~=-inf but 2nd, 3rd, 4th have -inf values. idx variable in my script is giving me back 1 2 which are basically are the first and second entries in b.

Answer 1

[Rik]

When you subindex a variable before using find, it will return the indices relative to that subindexed variable. This is because find doesn't know about the indexing.

This might fix the problem:

idx=b(idx);

[Waseem AL Aqqad]

WOW! It works.

Thank you so much for your explanation.

Answer 2

[KSSV]

Replace the line:

inActive = find(G_dmg.Nodes.Load==-inf);

with

idx = isinf(G_dmg.Nodes.Load) ; 
inActive = find(sign(G_dmg.Nodes.Load).*idx == -1)

Follow the same for other lines where you are trying to find -inf.

You can achieve the above without loop.

[Waseem AL Aqqad]

Thanks for telling me this technique.

Unfortunately, I have to use a loop here since the built in function "neighbors" accepts only scalar inputs, I can't pass the vector inActive to it.

I believe it has this ristriction as each node in a graph has different number of neighboring nodes. Of course, I can concatenate them like this:

b = [];
b = [b, neighbors(G_dmg,inActive(ii))];

But still, I cannot tell which nodes are neighbors to node X and which are neighbors to node Y.