Local search optimisation with OscaR, explained to my neighbour

Speaker(s) : Renaud De Landtsheer Christophe Ponsard

  • Language : Français
  • Nature : Conference
  • Date : Wednesday 8 July 2015
  • Schedule : 15h20
  • Duration : 40 minutes
  • Place : 205

Video : https://rmll.ubicast.tv/permalink/v1253b4381693qqgat95

OscaR is an open source framework for solving optimization problems. OscaR includes several optimization engines, namely: constraint-based local search, constraint programming, and linear programming. OscaR is implemented in Scala.

After an overview of the different engines and the shared infrastructure, the presentation will focus on the local search engine based on constraints. It will first present its main principles and engine architecture before focusing on the declarative layer enabling the easy definition of the problem constraints and objective function, and also the compact expression of powerful search procedures.

We will show how those open and extensible mechanisms not only provide a productive framework for developing effective solutions for large optimization problems, but also allow the emergence of business-specific modules (routing, scheduling) making the adoption even more easier. The presentation will be illustrated by several examples including feedback from an adopting company.

Renaud De Landtsheer , Christophe Ponsard ,
After a PhD in software engineering, formal methods and security, Renaud decided to head to technology transfer, and industry-driven research and development at CETIC in Belgium. Since 2011, he is the actively developing the CBLS engine of the OscaR open source optimisation framework within a community also gathering the beCool team from University of Louvain and the N-side company.

Slides (Powerpoint - 564.1 kb)
Slides (PDF - 569.5 kb)