Usage

Graphical mode

To start the application in graphical mode, just invoke complexnetspp whithout arguments from command line.

$ ./complexnetspp

Command line mode

ComplexNets++ can also run using the command line.

Loading a graph from an input file

The syntax to load a graph from an input file is the following:

$ ./complexnetspp -i <filename>

If the file represents a weighted graph, the command must include the --weighted argument:

$ ./complexnetspp -i <filename> --weighted

If the file represents a digraph, the command must include the --digraph argument:

$ ./complexnetspp -i <filename> --digraph

NOTE: Currently, both --weighted AND --digraph arguments CANNOT be used at the same time, as weighted digraphs aren't still supported by complexnets.

Creating a graph using a model

The graphs can also be created using a model. To achieve this, the command must have the following syntax:

$ ./complexnetspp <model_parameters>

Where <model_parameters> can take one of the following values depending on the used model:

• Erdos-Renyi:

--erdos -n <number_of_nodes> -p <probability>

• Barabási-Albert:

--barabasi -n <number_of_nodes> -c <number_of_initial_nodes> -m <number_of_attachment_nodes>

• Extended Hot:

--hot -n <number_of_nodes> -m <new_vertex_edges> -x <neighbors_select> -q <added_edges> -r <edges_select> -t <root_regeneration_rate>

• Molloy-Reed:

--molloy -k <ks_file>

• Hyperbolic:

--hyperbolic --n=<number of nodes> -a=<radial density> --deg=<degree coefficient>

Computing a property on a graph

To compute a property given a file, the command must have the following syntax:

$ ./complexnetspp <graph> <property>

where <graph> are the arguments to load a graph or create it using a model, and <property> can take the following values depending on the calculated property:

• Betweenness:

--betweenness <vertex_index>

• Degree distribution:

--ddist <degree>

NOTE: when working with a digraph, this command will show the number of vertices having <degree> as in-degree and out-degree at the same time.

• Clustering coefficient:

--clustering <vertex_index>

• Nearest neighbors degree:

--knn <vertex_index>

• Shell index:

--shell <vertex_index>

• Maximal clique:

 --maxCliqueExact <max_time>
 --maxCliqueAprox

NOTE: The method maxCliqueExact takes to long. That's why it has a time-out (in seconds). If looking for a quick answer use the approximation method.
NOTE2: If you want to obtain a result the most accurate possible but with a time limit use:

 $ ./complexnetspp <graph> --maxCliqueExact <max_time> || ./complexnetspp <graph> --maxCliqueAprox

Fitting data against a power law

To test a graph (either generated or loaded from a file) for a power law distribution, use the --print-deg option and redirect the output to the appropiate script. For instance:

$ ./complexnetspp <graph> --print-deg=1 | python powerlawpy/script.py

The script is independent to the project and requires several extra dependencies, mainly: python, python-dev, numpy, scigy. Please refer to the appropiate README for installation details.

Saving a graph in a file

To save a loaded or model-generated graph in a file, the command must have the following syntax:

$ ./complexnetspp <graph> -o <file>

where <graph> are the arguments to load a graph or create it using a model.

Generating a Gnuplot readable output

To generate a file that can be read and plotted by Gnuplot, the command must have the following syntax:

$ ./complexnetspp <graph> -o <output_file> <function>

where <graph> are the arguments to load a graph or create it using a model and <function> can take the following values depending on the calculated function:

• Betweenness vs. Degree:

--betweenness-output

• Degree distribution:

--ddist-output

If the output is expected to be grouped using log-bins, the command must include the log-bin option:

--ddist-output --log-bin=<number_of_bins>

NOTE: when working with digraphs, this command will output the in-degree probabilities into a file with _in_degree suffix and the out-degree probabilities into a file with _out_degree suffix.

• Clustering coefficient vs. Degree:

--clustering-output

• Nearest neighbors degree vs. Degree:

--knn-output

• Shell index vs. Degree:

--shell-output

• Maximal clique distribution (for each node ¿witch is the biggest clique it is part of?):

--maxCliqueExact-output <max_time>
--maxCliqueAprox-output

Using a bash script to generate plots

A bash script can be used to generate the plot for each property of a graph. To achieve this, the following command must be invoked:

$ sh plotter.sh <input_file>

This command creates a directory in which the plots for betweenness, degree distribution, clustering coefficient, nearest neighbors degree and shell index are saved for the graph represented by <input_file>. The input file must have an extension.

If the command is invoked with one more parameter:

$ sh plotter.sh <input_file> <number_of_bins>

then the degree distribution is also plotted using the log-bin method, with the number of bins specified by <number_of_bins>.