examination-scheduling

This project consist in finding a solution to the examination scheduling at TUM (Munich). Heuristics, Optimization problems and algorithms will be developped.

Main programmation language

Python

Project structure

The following folders are important for understanding the project. Inside the folders there is a Readme.md with instructions and details.

• inputData: Generating real data
• evaluation: Running solvers to get results
• heuristics: Implementations for different heuristics
• model: Constraints checking and data format
• GurobiModel: ILP models
• results: this one is for result files generated by the solvers
• tests: unit tests to ensure correctness

Using the Real Data

For installing the real data please refer to the file inputData/README.md !

Conventions

Problems Two way have to be considered when creating a problem:

• create a class inheriting of class BaseProblem and complete method build_constants, build_variables, build_constraints and build_dimensions
• create a method that returns a model guroby.Model. Variable names are in this case: "x_%s_%s_%s" % (i, j, k) or "x_%s_%s" % (i, j). x may be replaced by y. Only x and y can be used.

Variables Two way to create new variables:

• if the variable is contained in a guroby model, its name has to be: "x_%s_%s_%s" % (i, j, k) or "x_%s_%s" % (i, j). x may be replaced by y. Only x and y can be used
• if not x and y have to be dictionaries: x = {(i, k): 0.0 for ...} or x = {(i, k, l): 0.0 for ..}. x may be replaced by y. Only x and y can be used

Data The main data are:

• c for room capacities
• s for students capacities (for each exam)
• T for open time of rooms
• Q for conflicts
• conflicts for sparse version of Q
• h for beginning time of each timeslot
• n the number of exams
• p the number of timeslots
• r the number of rooms
  Upper cases are matrix [[]] and lower cases are lists []
  Other data can be introduced, however they must be linked to the main data.

** Test **
In the script test_constraints.py, in the method SetUp(self), create the problem, solve it and add it to the lists depending which constraints you want to solve.
If you want a best descriptif of eventual failures, do the same in test_constraint_advanced.py

Visualization

To visualize the produced solutions of a problem or heuristics, first use function update_variables for a problem or transform_variables for variables to get the right form of x, y variables. These functions can be found in tools/utils.py. Then run the function generate_file (in script tools/graphicer.py). It will generate a file in the folder model/Data/visualization/name.csv where name is the name you gave if you gave one else visual-date-time where date and time are respectively the current date and time the current time.

Installation for github (Windows)

Git for windows:
follow this link: https://github.com/git-for-windows/git/releases/tag/v2.8.1.windows.1
Then download and install it being careful with your windows version

Copy the project to your account:
Create first a github account
Then search for this project writing its name in the search bar
Above right click on Fork

clone the repository into your computer:

$ cd folder_path
$ git clone https://github.com/username/examination-scheduling.git
$ git init

Use github

Following command lines are essential to work on github

$ git commit -am "change descriptions" // To save localy all the change you did about the project
$ git push // to update on github the changes you saved with `git commit`
$ git remote add examination https://github.com/mickael-grima-TUM/examination-scheduling.git // Need to be done just one time. See `git pull` under
$ git pull examination master // update your code with the master banch. Conflicts may appear, solve them and commit then push