How to create a graph from an adjacency matrix in C?

Usage
,
1

Hello!

Is there a way how to create a igraph_t graph directly from a binary adjacency matrix in C igraph? I know it is possible in Python, but I need to use the develop version of C igraph because of the latest change in label propagation function. I tried to covert the adjacency matrix to .lgl in like this:

import sys
import igraph as ig

G = ig.Graph.Read_Adjacency(sys.argv[1], sep=',',mode='undirected')
path = sys.argv[1][0:-3]+"lgl"
file = open(path,"w")

G.write_lgl(path, names=None, weights=None)
file.close()

Then, I used the function igraph_read_graph_lgl(&g, input, 0, IGRAPH_ADD_WEIGHTS_NO, 0); with to create the graph in C but I have a problem that the order of nodes is changed. I need the order of nodes to be the same as in the adjacency matrix (node 0 = first row of the matrix). The graph is unweighted and undirected. Do you know how to do it?

Thank you,
Kateřina

2

Yes, you can use igraph_adjacency().

I’ll get back to you about this later.

3

Thank you! But I can’t pass .csv file to igraph_adjacency(), right? I’m sorry, I forgot to mention this in the original question but I have data in .csv. I want to process more matrices, so I don’t want to hard code it to the C code.

4

If Python is more convenient for you then perhaps it would be best if you compiled the latest 0.9.x version of python-igraph. This will be relatively easy on Linux or macOS. Instruction are here:

You will need to clone the git repository and use the master branch. This already includes the fix to label propagation that you contributed.

5

The easiest way to install the version from the master branch is pip install git+https://github.com/igraph/python-igraph. This automatically fetches the repository, builds the package and installs it.

We should perhaps also include this in the documentation you reference @szhorvat.

1 Like
6

Well, I didn’t know that, so please do include it :slight_smile:

1 Like
7

Thank you very much, pip install git+https://github.com/igraph/python-igraph works for me.

1 Like
8

Great, good to hear!

I’ve updated the instructions to also include this (CC @szhorvat).