508. A combinatorial Benders decomposition approach for a scheduling problem arising in a sports institution
Invited abstract in session MA-12: Mathematical Programming in Project Scheduling , stream Scheduling and Project Management.
Monday, 8:30-10:00Room: Clarendon SR 1.02
Authors (first author is the speaker)
| 1. | Javier Marenco
|
| Business School, Universidad Torcuato Di Tella | |
| 2. | Ivo Koch
|
| BASF | |
| 3. | Luciana Skakovsky
|
| Departamento de Computación, Universidad de Buenos Aires |
Abstract
In this work we consider the following scheduling problem. We are given a set of tasks, a set of employees, and a set of days. Each task has a priority, a length in hours, and a number of needed employees who must be scheduled to perform the task simultaneously. We also have a compatibility matrix between tasks and employees, specifying which employees can perform each task. The problem consists in determining which tasks are to be scheduled and determining a day, a starting time, and a set of employees for each scheduled task in such a way that (a) no employee performs two tasks at the same time, (b) task constraints are obeyed, and (c) the total priority of the scheduled tasks is optimized. We present a computational approach for this problem based on the combinatorial Benders decomposition. We show that this approach is effective, outperforming a commercial integer programming solver over a natural formulation for real instances coming from a sports institution. As part of this decomposition, we consider the problem of explaining the compatibility task-employee matrix in terms of skills. We seek to assign a minimum-sized set of skills to tasks and employees, in such a way that an employee is compatible with a task if and only if the skills set of the employee is included in the skills set of the task. The resulting explanation of the compatibility matrix is used as a means to strengthening the primary subproblem in the combinatorial Benders decomposition described above.
Keywords
- Scheduling
- Programming, Mixed-Integer
- Combinatorial Optimization
Status: accepted
Back to the list of papers