fliscopt 0.3.0

Flight scheduling optimization using Genetic Algorithm variants and other algorithms.

Fliscopt

FLIght SCheduling OPTimization 🛫 or fliscopt is a simple optimization library for flight scheduling and related problems in the discrete domain. The library supports plotting, asynchronous multiprocessing, and unimodal optimization benchmarks. The following repository contains code for the paper "XYZ". The experiments were performed in PyPy3.7 and CPython 3.8.10.

Following algorithms have been implemented and test as of date:

Algorithms:

• Hill Climbing
• Random Search
• Simulated Annealing
• Genetic Algorithm
• Genetic Algorithm in Reverse Mode
• Genetic Algorithm with Reversals
• Genetic Algorithm with Random Search as a Reversal/Reverse Process
• Iterated Chaining

Getting Started

Install the library using pip:

pip install fliscopt

Or for unreleased versions:

pip install 'git+https://github.com/Agrover112/fliscopt/fliscopt@branchname

Or for development:

git clone https://github.com/Agrover112/fliscopt.git
cd fliscopt
pip install .

Download the flights.txt file from the following link and add it to a data/ directory within your parent directory.

Checkout out the examples in the examples directory or run in Google Collab

For PyPy users

The instructions for setup are mentioned in the setup directory. Alternatively, you can set up using this bash script. A requirements file is provided just in case. The script creates and activates a PyPy Conda environment with all libraries and dependencies.

cd ./setup.sh
source setup.sh

Then install using:

pypy -mpip install fliscopt

Testing

After adding any new algorithm, you can run the tests to check if the code is working properly.

./run_tests.sh

Results

Experimental Results

Results were compared by using the same seeds. The following table shows the results of the experiments. (Will be shortly added)

Accessing results

After running the experiments, the results are stored in the results directory. The results are stored in the following format in subdirectories:

.
├── multi_proc
│   ├── ackley_N2
│   │   ├── genetic_algorithm_results.csv
│   │   ├── genetic_algorithm_reversed_results.csv
│   │   ├── genetic_algorithm_with_reversals_results.csv
│   │   ├── hill_climb_results.csv
│   │   ├── random_search_results.csv
│   │   └── simulated_annealing_results.csv
│   ├── booth/...
|   |
|   |
│   └── zakharov
│       ├── genetic_algorithm_results.csv
│       ├── genetic_algorithm_reversed_results                  
│       ├── genetic_algorithm_with_reversals_results.csv
│       ├── random_search_results.csv
│       └── simulated_annealing_results.csv
├── plots
│   ├── ackley_N2
│   ├── fitness_function
│   │   ├── hill_climb.png
│   │   └── random_search.png
│   ├── flight_scheduling
│   │   ├── simulated_annealing.png
│   │   ├── sol_chaining.png
│   │   └── sol_chaining_a1.png
│   └── griewank

References

Read the following for detailed understanding of our project.

[1] Alicea B., Grover A., Lim A. ,Parent J, Unified Theory of Switching. Flash Talk to be presented at: 4th Neuromatch Conference; December 1 - 2, 2021

Contributing Guidelines

Refer Contributing.md and Project Board for mode details. This repository follows conventional commits!