GeNet provides tools to represent and work with a multi-modal transport network with public transport (PT)
services. It is based on MATSim's representation of such networks. The underlying
network available to PT services (roads, railways, but also ferry/flight connections) uses a networkx.MultiDiGraph
with additional methods for 'links' which are unique in genet.Network (networkx.MultiDiGraph accepts multiple
edges between the same from and to node pair; referring to an edge in networkx.MultiDiGraph and genet.Network
has the same effects, i.e. the result is a dictionary indexed by the multi edge index). The PT services are
represented through genet.Schedule class which relies on other genet
classes: the Schedule relies on a list of genet.Service's, which in turns consists of a list of genet.Route's.
Each Route class object has an attribute stops which consists of genet.Stops objects. The Stops carry spatial
information for the PT stop.
The goal of GeNet is to:
- Provide a formalised in-memory data structure for representing a multi-modal network with a PT service
- Enable using the data structure for tasks such as generating auxiliary MATSim files e.g. Road Pricing
- Simplify the process of modifying a network and provide simple change log to track the differences between the input and output networks.
- Provide validation methods to check for simple errors such as: whether a
Routehas more than oneStopor that the underlying graph doesn't have any dead-ends or sources (a place which you can leave but cannot get back to).
To run pre-baked scripts that use GeNet in a number of different scenarios you can use docker, which will save you the work of installing GeNet locally:
Docker is the recommended way to use GeNet if you do not plan to make any code changes.
docker build -t "genet" .
docker run genet reproject_network.py -h
usage: reproject_network.py [-h] -n NETWORK [-s SCHEDULE] [-v VEHICLES] -cp
CURRENT_PROJECTION -np NEW_PROJECTION
[-p PROCESSES] -od OUTPUT_DIR
Reproject a MATSim network
optional arguments:
-h, --help show this help message and exit
-n NETWORK, --network NETWORK
Location of the network.xml file
-s SCHEDULE, --schedule SCHEDULE
Location of the schedule.xml file
-v VEHICLES, --vehicles VEHCILES
Location of the vehicles.xml file
-cp CURRENT_PROJECTION, --current_projection CURRENT_PROJECTION
The projection network is currently in, eg.
"epsg:27700"
-np NEW_PROJECTION, --new_projection NEW_PROJECTION
The projection desired, eg. "epsg:27700"
-p PROCESSES, --processes PROCESSES
The number of processes to split computation across
-od OUTPUT_DIR, --output_dir OUTPUT_DIR
Output directory for the reprojected network
Otherwise, you can install genet as a python package, in your base installation
of python or a virtual environment.
Run the pre-baked scripts, write your own scripts or use IPython shell or Jupyter Notebook to load up a network,
inspect or change it and save it out to file. Check out the
wiki pages and
example jupyter notebooks
for usage examples.
Note: if you plan only to use GeNet rather than make code changes to it, you can avoid having to perform any local installation by using GeNet's Docker image. If you are going to make code changes, or you don't want to use Docker for some reason...
GeNet uses some Python libraries that rely on underlying native libraries for things like geospatial calculations and linear programming solvers. Before you install GeNet's Python dependencies, you must first install these native libraries.
Note: The default CBC solver is pre-installed inside GeNet's Docker image, which can save you some installation effort
To use methods which snap public transit to the graph, GeNet uses a mathematical solver. If you won't be using such functionality, you do not need to install this solver. Methods default to CBC, an open source solver. Another good open source choice is GLPK. The solver you use needs to support MILP - mixed integer linear programming.
The commands for installing the necessary native libraries vary according to the operating system you are using, for example:
| OS | Commands |
|---|---|
| Mac OS | brew install spatialindex brew install gdal --HEAD brew install gdal brew tap coin-or-tools/coinor brew install coin-or-tools/coinor/cbc |
| Ubuntu | sudo apt install libspatialindex-dev sudo apt install libgdal-dev sudo apt install coinor-cbc |
(Use equivalent linux or Windows package management as appropriate for your environment)
brew install python3.7
brew install virtualenv
Create and activate a Python virtual environment
virtualenv -p python3.7 venv
source venv/bin/activate
Finally install GeNet's Python dependencies
pip install -e .
Please follow kepler's installation instructions to be able to use the visualisation methods. To see the maps in a jupyter notebook, make sure you enable widgets.
jupyter nbextension enable --py widgetsnbextension
We welcome community contributions to GeNet; please see our guide to contributing and our community code of conduct. If you are making changes to the codebase, you should use the tools described below to verify that the code still works. All of the following commands assume you are in the project's root directory.
python -m pytest -vv tests
To generate an HTML coverage report at reports/coverage/index.html:
./bash_scripts/code-coverage.sh
./bash_scripts/lint-check.sh
./bash_scripts/notebooks-smoke-test.sh