Installation¶

There are two supported ways to install HermesPy on your system. Users may either pull pre-built wheels from the official python package index PyPi or clone the source code from GitHub and build the package from scratch.

All methods will be described in the following sections.

Install from PyPi¶

Pulling HermesPy from PyPi is the recommended installation method for end-users who do not intend to extend the HermesPy source code.

HermesPy is registered as an official package within the python package index. Binaries for Windows and most Linux distributions, built from the newest release version of HermesPy, are distributed via the index, enabling easy installations for most operating systems. Since releases are scheduled bi-anually, the latest features and bug fixes may not be available in the distributed package.

Install the package by executing the following commands within a terminal:

conda create -n <envname> python=3.11
conda activate <envname>
conda install pip
pip install hermespy

python -m venv <envname>
<envname>\\Scripts\\activate
pip install hermespy

python -m venv env
source env/bin/activate
pip install hermespy

Executing these statements sequentially results in the following actions:

Creation of a new virtual environment titled <envname>
Activate the newly created environment
Install the HermesPy wheel from PyPi within the environment

Note that if you plan on utilizing HermesPy within an already existing Python environment, you may omit step one and replace <envname> by the title of the existing environment.

Hermes lists a set of optional features requiring the installation of additional dependencies. In order for those features to function properly, the respective identifiers have to be appended to the installation command

pip install "hermespy[feature,feature,feature]"

The following feature identifiers are available:

Optional Features¶
Feature	Description
quadriga	Support for the Quadriga channel model
uhd	Support for the USRP Hardware Driver
audio	Support for audio hardware
sionna	Support for sionna raytracing via Tensorflow
documentation	Dependencies to build the documentation from source
develop	Dependencies to build binaries from source

Install from Source¶

Cloning the HermesPy source code and manually building / installing its package is the recommended way for developers who plan on extending the HermesPy source code. Additionally, this mathod can also be applied by users as a fall-back who, for any reason, are unable to install HermesPy from the index.

Before cloning, make sure to have the LFS extension to Git installed. Using the Git command line interface, the HermesPy source code can be copied to any system by executing

git clone --recursive <this-repo>
cd hermespy/

within a terminal. Please note that the –recursive flag is required to clone the repository including its submodules.

Some parts of HermesPy are provided as C++ implementations with Python bindings for improved performance. Therefore, building the package from source requires your system to have a build chain detectable by CMake installed and configured. For Windows users, we recommend downloading and installing the Visual Studio Build Tools.

Build and install the package contained within the repository by executing the following commands within a terminal:

conda create -n <envname> python=3.11
conda activate <envname>
conda install pip
pip install .

python -m venv <envname>
<envname>\\Scripts\\activate
pip install .

python -m venv env
source env/bin/activate
pip install .

Executing these statements sequentially results in the following actions:

Creation of a new virtual environment titled <envname>
Activate the newly created environment
Install the HermesPy wheel from source within the environment

Note that if you plan on utilizing HermesPy within an already existing Python environment, you may omit step one and replace <envname> by the title of the existing environment.

If you plan to alter the source code in any way, we recommend installing the package as editable. As a result, all combined binaries and source files will remain within the repository directory during installation.

conda create -n <envname> python=3.11
conda activate <envname>
conda install pip
pip install -e ".[develop]"

python -m venv <envname>
<envname>\Scripts\activate
pip install -e ".[develop]"

python -m venv env
source env/bin/activate
pip install -e ".[develop]"

Install Quadriga¶

In addition to its native channel models, HermesPy supports the Quadriga channel model as an external dependency. For it to be used, some preliminary steps need to be taken. Quadriga is based on Matlab and can be executed by either the Matlab interpreter or its open-source equivalent Octave.

In order to execute the Matlab interpreter the matlab.engine package provided by Matlab needs to be installed manually.

In order to execute the Octave interpreter the additional oct2py package needs to be installed (pip install oct2py). Under Windows, an extension of the PATH variable may be required for oct2py to be able to locate the octave command line interface:

setx PATH "%PATH%;<path-to-octave-cli>"

When installing HermesPy from the distributed package, the Quadriga source code needs to be installed manually. Download the latest version of Quadriga and extract the zip archive in a location of your choice. Afterwards, set the environment variable HERMES_QUADRIGA to point to the quadriga_src directory. This will point Hermes to search for the Quadriga files within the specified location during simulation runtime.