The subsequently described installation steps are demonstrated as shell commands, where the path before the
% sign denotes the
directory in which the commands following the
% should be entered.
Building the model with the scripts in this repository runs on a normal computer. The entire energy model is relatively heavy and it easily requires > 40Gb of available memory in the disk (HHD/SSD); the exact space requirements depend on the specific models under interest.
The complete list of software needed before installing PyPSA Africa is listed below.
Python 3 (mandatory): Python is used as our main programming language, thus its knowledge is mandatory. To refresh the knowledge, there are plenty of online courses free-of-charge, e.g. CSDojo playlist. Useful content to watch refer to numpy, pandas
conda (mandatory): in order to use packages in python, it is highly recommended to use a conda package manager, such as Anaconda or Miniconda (recommended). There are many things you can do wrong with conda. This article provides you a crystal clear explanation of conda (excellent read).
Git (mandatory): Git is a free open source system aimed at tracking changes in the code development and enable to coordinate the parallel software development between many developers. It is mandatory to learn the git basics.
Java (recommendation): A Java distribution is needed for using powerplantmatching package. To have a better user experience, please install the redistribution from the website according to your operating system.
IDE Python (recommendation): in order to write python code, you need an Integrated Development Environment (IDE) that is a software used to write code. Any program can be used, however, we recommend Visual Studio Code, which is freely available online. Other alternatives are also viable if you are familiar with them, such as PyCharm, however we recommend Visual Studio Code also given its easy to use interface with Git. Note: if you decide to use Visual Studio Code, check out the tutorial about how to use Git and Github in Visual Studio Code
Solver (mandatory): an optimization solver is needed to solve the mathematical problem that is build with the automated workflow. With the goal of supporting completely open source initiative, we focus on relying on Open-Source solvers, such as CBC , GLPK, WinGLPK or HiGHS; to further improve performances, commercial solvers like Gurobi or CPLEX (both commercial licenses with free academic options) can also be used. A recommended instruction to install the HiGHS solver is given here.
Be aware that the list of software listed above is only the prerequisite elements needed to successfully install the PyPSA Africa model. The complete list of recommended software and prerequisite needed to enjoy the full PyPSA Africa experience is listed in the Tutorial section. Most of the dependencies needed will be automatically installed using the conda environments listed below
Clone the Repository
First of all, clone the PyPSA meets Africa repository using the version control system
The path to the directory into which the
git repository is cloned, must not have any spaces!
If you do not have
git installed, follow installation instructions here.
/some/other/path % cd /some/path/without/spaces /some/path/without/spaces % git clone https://github.com/pypsa-meets-africa/pypsa-africa.git
Install Python Dependencies
PyPSA meets Africa relies on a set of other Python packages to function.
We recommend using the package manager and environment management system
conda to install them.
Make sure that
conda is already installed on your system or install one of the following two distributions:
For instructions for your operating system follow the
conda installation guide.
The python package requirements are curated in the envs/environment.yaml file. The environment can be installed and activated using
.../pypsa-africa % conda env create -f envs/environment.yaml .../pypsa-africa % conda activate pypsa-africa
Sometimes the conda pypsa-africa installation can take more than one our
(indicates some conflicts in the packages). In case a quick installation is necessary,
you might also want to try out
mamba. It was observed to take around 5-15min to
install all pypsa-africa dependencies with it.
To use jupyter lab (new jupyter notebooks) continue with the ipython kernel installation and test if your jupyter lab works:
.../pypsa-africa % ipython kernel install --user --name=pypsa-africa .../pypsa-africa % jupyter lab
Please, make sure to have properly installed java, from the official website or equivalent.
In linux only, that is possible through the following command.
To verify the successful installation, you can verify that by using the following code. .. code: bash
…/pypsa-africa % java -version
The expected output should resemble the following text: .. code: bash
java version “1.8.0_341” Java(TM) SE Runtime Environment (build 1.8.0_341-b10) Java HotSpot(TM) 64-Bit Server VM (build 25.341-b10, mixed mode)
Snakemake, which is one of the major dependencies, will be automatically installed in the environment pypsa-africa,
thereby there is no need to install it manually.
The snakemake included in the conda environment pypsa-africa installed with the above-mentioned procedure can be executed with the following procedure:
The entire distribution, including the data for the whole Africa, is very heavy (>40Gb) and it involves a large number of files.
To simplify the installation of the github folder, the main source code is available in the Github folder, whereas the data are stored in cloud.
retrieve_databundle_light has been specifically developed to set up the raw data, and the procedure below guides in setting up the needed data.
Duplicate the file
config.default.yamland rename the copy as
config.yamlusing any text editor
Make sure that the option
Execute the following code on the shell to download initial files. Please, note that around 20Gb zipped files will be downloaded, so make sure you have a stable connection, time and around 50 Gb available in your system. If no errors show up, then you can proceed.
.../pypsa-africa % conda activate pypsa-africa .../pypsa-africa % snakemake -j1 retrieve_databundle_light --force
In the file
config.yamlset the option
falseand save the file:
Once these tasks have been completed, the package is ready to use.