Script to generate connected graphs

A tool to generate graphs, based on the Barabási–Albert (BA) model. You can use this to generate test data that resembles relations on a social graph.

Usage

The script accepts the following argumens

• n Number of nodes in the graph (default 10000)
• m How many edges to create for each node (default 10)
• s The format string (default {a} {b})
• d If present, will at the end print the generated distribution (the distribution of node degrees).

Note that the script always generate edges in pairs, a->b, b->a. So a -m value of 10, will actually create a network whose average # of edges per node is 20.

Use -s to specify your output format, most likely some sort of CSV to import into your DB; format is the one used for python's string formatting.

Example

Runnning time is acceptable for medium-sized networks. On a modest notebook, you can expect

time python ba_graph.py -m 50 -n 200000 -s "{a};{b}" -d > data.csv
real    20m41.900s
user    20m30.190s
sys     0m3.980s

that is 20' to generate 200000 nodes and 20000000 (twenty million) edges.

For the above example, the resulting node degree distribution is:

(note the log10 scale)

Have fun!