Introduction

This is the master repository for transportation modelling projects of MMO lab

It is based on

• https://github.com/MeruzaKub/TransportNet
• https://github.com/tamamolis/TransportNet
• https://github.com/Lareton/transport_network_optimization

and also uses code written in other related projects of MMO lab.

Content

Repo contains implementations of basic algorithms for the equilibrium traffic assignment problem: $$\sum_e \sigma_e(f_e) \to \min_{f \in F_d},$$

and the combined travel-demand (twostage) problem: $$\gamma \sum_e\sigma_e(f_e) + \sum_{ij}d_{ij}\ln d_{ij} \to \min_{\substack{f\in F_d \\ \sum_j d_{ij}=l_i\ \sum_i d_{ij}=w_j}}.$$

Installation

1. Grab bstabler's TransportationNetworks sumbodule: use git clone --recurse submodules or do git submodule update --init after clone
2. Install conda if not yet
3. Create and activate conda environment

conda env create -f environment.yml
conda activate tm

4. Add this conda environment to your jupyter notebook

ipython kernel install --user --name=tm

After that you can select tm kernel from notebook's kernel menu. Alternatively, you can install jupyter into the environment and run it from there (but it gave me an error while launching the notebook app)

conda install jupyter -n tm

More details about jupyter with conda env here

Docker image might be created on demand to simplify the installation process. We also have remote linux servers for internal use

Experiments :

1. Пример запуска экспериментов : python3 compare_methods.py.
2. Модуль для запуска экспериментов src/test.py.
3. Сохранение результатов экспериментов в директорию experiments_result происходит, если выставить флаг save=True в методе test.plot().
4. В TransportationNetworks лежат датасеты городов. (загруженные из репозитория bstabler)
5. Алгоритмы расположены в my_algs.py и algs.py.
6. NFW вмержен в основной репозиторий mmo_tm.