Networkit - A Growing Open-Source Toolkit For Large-Scale Network Analysis#399
NetworKit is an open-source tool suite for high-performance network analysis. Its aim is to provide tools for the analysis of large networks in the size range from thousands to billions of edges. For this purpose, it implements efficient graph algorithms, many of them parallel to utilize multicore architectures. These are meant to compute standard measures of network analysis. NetworKit is focused on scalability and comprehensiveness. NetworKit is also a testbed for algorithm engineering and contains novel algorithms from recently published research (see list of publications below).
NetworKit is a Python module. High-performance algorithms are written in C++ and exposed to Python via the Cython toolchain. Python in turn gives us the ability to work interactively and a rich environment of tools for data analysis and scientific computing. Furthermore, NetworKit's core can be built and used as a native library if needed.
You will need the following software to install NetworKit as a python package:
- A modern C++ compiler, e.g.: g++ (>= 6.1), clang++ (>= 3.9) or MSVC (>= 14.13)
- OpenMP for parallelism (usually ships with the compiler)
- Python3 (3.6 or higher is supported)
- Development libraries for Python3. The package name depends on your distribution. Examples:
apt-get install python3-dev
dnf install python3-devel
- Windows: Use the official release installer from www.python.org
- CMake version 3.6 or higher (Advised to use system packages if available. Alternative:
pip3 install cmake)
- Build system: Make or Ninja
- Cython version 0.29 or higher (e.g.,
pip3 install cython)
In order to use NetworKit, you can either install it via package managers or build the Python module from source.
Install via package manager
While the most recent version is in general available for all package managers, the number of older downloadable versions differ.
pip3 install [--user] networkit
conda (channel conda-forge)
conda config --add channels conda-forge conda install networkit [-c conda-forge]
brew install networkit
spack install py-networkit
Building the Python module from source
git clone https://github.com/networkit/networkit networkit cd networkit python3 setup.py build_ext [-jX] pip3 install -e .
The script will call
makeas fallback) to compile NetworKit as a library, build the extensions and copy it to the top folder. By default, NetworKit will be built with the amount of available cores in optimized mode. It is possible the add the option
-jNthe number of threads used for compilation.
To get an overview and learn about NetworKit's different functions/classes, have a look at our interactive notebooks-section, especially the Networkit UserGuide. Note: To view and edit the computed output from the notebooks, it is recommended to use Jupyter Notebook. This requires the prior installation of NetworKit. You should really check that out before start working on your network analysis.
We also provide a Binder-instance of our notebooks. To access this service, you can either click on the badge at the top or follow this link. Disclaimer: Due to rebuilds of the underlying image, it can takes some time until your Binder instance is ready for usage.
If you only want to see in short how NetworKit is used - the following example provides a climpse at that. Here we generate a random hyperbolic graph with 100k nodes and compute its communities with the PLM method:
>>> import networkit as nk >>> g = nk.generators.HyperbolicGenerator(1e5).generate() >>> communities = nk.community.detectCommunities(g, inspect=True) PLM(balanced,pc,turbo) detected communities in 0.14577102661132812 [s] solution properties: ------------------- ----------- # communities 4536 min community size 1 max community size 2790 avg. community size 22.0459 modularity 0.987243 ------------------- -----------
Install the C++ Core only
In case you only want to work with NetworKit's C++ core, you can either install it via package managers or build it from source.
Install C++ core via package manager
conda (channel conda-forge)
conda config --add channels conda-forge conda install libnetworkit [-c conda-forge]
brew install libnetworkit
spack install libnetworkit
Building the C++ core from source
We recommend CMake and your preferred build system for building the C++ part of NetworKit.
The following description shows how to use CMake in order to build the C++ Core only:
First you have to create and change to a build directory: (in this case named
mkdir build cd build
Then call CMake to generate files for the
makebuild system, specifying the directory of the root
..). After this
makeis called to start the build process:
cmake .. make -jX
To speed up the compilation with make a multi-core machine, you can append
-jXwhere X denotes the number of threads to compile with.
Use NetworKit as a library
This paragraph explains how to use the NetworKit core C++ library in case it has been built from source. For how to use it when installed via package managers, best refer to the official documentation (brew, conda, spack).
In order to use the previous compiled networkit library, you need to have it installed, and link it while compiling your project. Use these instructions to compile and install NetworKit in
cmake .. make -jX install
Once NetworKit has been installed, you can use include directives in your C++-application as follows:
You can compile your source as follows:
g++ my_file.cpp -lnetworkit
Building and running NetworKit unit tests is not mandatory. However, as a developer you might want to write and run unit tests for your code, or if you experience any issues with NetworKit, you might want to check if NetworKit runs properly. The unit tests can only be run from a clone or copy of the repository and not from a pip installation. In order to run the unit tests, you need to compile them first. This is done by setting the CMake
cmake -DNETWORKIT_BUILD_TESTS=ON ..
Unit tests are implemented using GTest macros such as
TEST_F(CentralityGTest, testBetweennessCentrality). Single tests can be executed with:
Additionally, one can specify the level of the logs outputs by adding
--loglevel <log_level>; supported log levels are:
Compiling with address/leak sanitizers
Sanitizers are great tools to debug your code. NetworKit provides additional Cmake flags to enable address, leak, and undefined behavior sanitizers. To compile your code with sanitizers, set the CMake
cmake -DNETWORKIT_WITH_SANITIZERS=leak ..
By setting this flag to
address, your code will be compiled with the
undefinedsanitizers. Setting it to
leakalso adds the
The most recent version of the documentation can be found online.
For questions regarding NetworKit, have a look at our issues-section and see if there is already an open discussion. If not feel free to open a new issue. To stay updated about this project, subscribe to our mailing list.
We encourage contributions to the NetworKit source code. See the development guide for instructions. For support please contact the mailing list.
List of contributors can be found on the NetworKit website credits page.
The program source includes:
- the TLX library
- the TTMath bignum library
The source code of this program is released under the MIT License. We ask you to cite us if you use this code in your project (c.f. the publications section below and especially the technical report). Feedback is also welcome.
The NetworKit publications page lists the publications on NetworKit as a toolkit, on algorithms available in NetworKit, and simply using NetworKit. We ask you to cite the appropriate ones if you found NetworKit useful for your own research.