antoineallard / a-portable-graph-library Goto Github PK

PGL provides a set of C++ objects and Python scripts to analyze, manipulate and simulate the structure of graphs (or complex networks).

Except for a few exceptions, every objects provided by the PGL are written in standard C++. They are provided via self-contained header files making them easily integrable into any C++ project (can be included similarly to any header-only libraries such as Boost or the STL).

The development of this library was motivated by the need of a fast, lightweight and customizable code, which was easy to install and straightforward to deploy on supercomputing facilities whose softwares are not always up-to-date.

At the beginning of this project, none of the existing high-quality libraries such as graph-tool, NetworkX or igraph fulfilled the aforementionned requirements.

I therefore decided to start writing my own code, adding new features following the needs of my research. This is a work in progress; comments, suggestions, contributions and bug reports are welcome.

• A C++11 (or newer) compliant compiler
• Python 3.x

Clone this reporsitory using the --recurse-submodules option (see Git documentation), for example:

git clone --recurse-submodules https://github.com/antoineallard/a-portable-graph-library.git

and then

• copy the required header file (<sub-module>/src/<class object>.hpp) file in your project directory.
• add #include <class object>.hpp to your C++ code

• copy the required script file (<sub-module>/src/<script file>.hpp) file in your project directory.
• add import <script file> to your Python script

There is no extensive documentation. However we provide several tutorials in the form of C++ codes and Jupyter notebooks to illustrate how to use the PGL. Codes validating several functionalities of the PGL provide additional examples and are located in most sub-module's validation/ sub-directory.

• directed_graph: Provides functions to compute and extract various structural properties of simple unweighted directed graphs without self-loops.

• geometric_Sd_model: Generates random geometric graphs according to the S^D model.

• l-cloning Generates an undirected edgelist using the L-cloning algorithm.

• percolation_on_graph: Provides functions to efficiently simulate bond percolation on simple unweighted undirected graphs without self-loops.

Most sub-modules of the PGL import graphs using an edgelist format (see each sub-module for details). A compatible edgelist file can be generated from NetworkX via

networkx.write_edgelist(your_graph_object, filename_of_edgelist, delimiter=' ', data=False)