02229 - Systems Optimization: Group Project

Authors

Milán Balázs

Nicolai Herrmann

[redacted]

Antoine Sébert

Prerequisites

Python

Get the interpreter on the official website.

We will be working with the version 3.7.x.

You can check the interpreter's version with:

$ python --version

Package Installer

We will use pip as package installer.

You can install it by following this guide.

Then, check if the package installer has been installed for Python 3.7 with:

$ pip --version

The packages index can be accessed here.

Python Dependency Manager

To manage the deps, we will use poetry.

You can install it by following this guide.

Then, check if the package installer has been installed with:

$ poetry --version

Get started

Get the sources

Clone the branch with:

$ git clone --single-branch --branch constraint-programming https://gitlab.com/Zeltex/02229---systems-optimization

Create workflow

Go to the repository that has just been cloned and install the dependencies by running the following:

$ poetry install

You can update the dependencies by running the following:

$ poetry update

Launch

Change your working directory to the project's directory and run it with:

$ poetry run python src/main.py

Usage

You can show the CLI usage with:

$ poetry run python src/main.py --help
usage: SOLVER [-h] [-f FORMAT] [--verbose] [--version]
              (--case FOLDER | --collection FOLDER)

Solve task scheduling problems using graph coloration.

optional arguments:
  -h, --help            show this help message and exit
  -f FORMAT, --format FORMAT
                        Either one of xml, json, raw, svg
  --verbose             Toggle program verbosity.
  --version             show program's version number and exit
  --case FOLDER         Import problem description from FOLDER (only the first
                        *.tsk and *.cfg files found are taken, all potential
                        others are ignored).
  --collection FOLDER   Recursively import problem descriptions from FOLDER
                        and/or subfolders (only the first *.tsk and *.cfg
                        files found of each folder are taken, all potential
                        others are ignored).

Example

Solve the test case [data/Case 1](data/Case 1) with:

$ poetry run python src/main.py --case "data/Case 1"

Solve all the test cases data/100pct with:

$ poetry run python src/main.py --collection "data/100pct"

Redirects the console output including the solution to a file:

$ poetry run python src/main.py --case "data/Case 1" 2> file.txt

Note: If the file does not exist, it will be created.

Tests suite and style checks

Run the tests with:

$ poetry run python -m unittest

Note: this feature is not supported yet. See the roadmap section.

Perform a style check on the whole source with:

$ poetry run flake8

Simplified operation

graph LR
Start(Start) -->|XML| Builder
Builder -->|IR| Solver
Solver -->|Solution| Formatter
Formatter -->|Fmt Solution| End(End)

File Hierarchy

--root/
  +--build/					// future support for Cython builds
  +--data/					// test cases
  |  +--100pct/					// complex test case
  |  +--200pct/					// complex test case
  |  +--300pct/					// complex test case
  |  +--400pct/					// complex test case
  |  +--500pct/					// complex test case
  |  +--Case 1/					// simple test case
  |  +--Case 2/					// simple test case
  |  +--Case 3/					// simple test case
  |  +--Case 4/					// simple test case
  |  +--Case 5/					// simple test case
  |  +--Case 6/					// simple test case
  |  +--readme.txt				// test case data model definition
  +--src/					// sources
  |  +--builder.py				// problem builder
  |  +--datatypes.py			// structured data
  |  +--draw.py					// problem & solution painter
  |  +--format.py				// solution formatter
  |  +--log.py					// logging handler
  |  +--main.py					// entry point
  |  +--rate_monotonic.py		// rate monotonic -related utilities
  |  +--solver.py				// problem solver / solution generator
  |  +--timed.py				// timed function wrapper
  +--tests/					// tests
  |  +--main.py					// tests for src/main.py
  +--.flake8				// project-specific styles
  +--.gitattributes			// project attributes
  +--.gitignore				// unstaged files and folders
  +--.gitlab-ci.yml			// Gitlab pipeline configuration
  +--pyproject.toml			// poetry configuration
  +--README.md				// this file
  +--setup.py				// python package definition

Roadmap

This month - 22/11/2019

catch all exceptions : https://docs.python.org/3/library/contextlib.html
preemption
benchmark : https://github.com/python/pyperformance
optimize : https://docs.python.org/3/library/collections.html, https://docs.python.org/3/library/stdtypes.html, https://stackoverflow.com/questions/45123238/python-class-vs-tuple-huge-memory-overhead/45123561
setup.py : https://setuptools.readthedocs.io/en/latest/pkg_resources.html, https://setuptools.readthedocs.io/en/latest/setuptools.html
cython : http://docs.cython.org/en/latest/src/quickstart/build.html, http://docs.cython.org/en/latest/
write tests : https://docs.python.org/3/library/unittest.html
random problem generator

This week - 12/11/2019

add references support to avoid data copy
workload update (static attrs in core and cpu, invalidated on modification)
generate visual representation (SVG formatter)

06/11/2019

replace named tuples by mutable equivalent : https://bitbucket.org/intellimath/recordclass/src/default/
OR-tools : get feasible solution + support Period, Deadline, Offset, MaxJitter
linux cluster for benchmarks

AntoineSebert / scheduling-solver