Session Detail Information
Add this session to your itinerary

Cluster :  Optimization/Computational Optimization and Software

Session Information  : Sunday Oct 06, 16:30 - 18:00

Title:  Optimization Software and Applications
Chair: Hans Mittelmann,Arizona State University, Box 871804, Tempe AZ, United States of America, MITTELMANN@asu.edu

Abstract Details

Title: Automatic Algorithm Selection for Nonlinear Optimization Problems
 Presenting Author: Zsolt Csizmadia,FICO, FICO House, Starley Way, Birmingham B37 7GN, United Kingdom, ZsoltCsizmadia@fico.com
 
Abstract: Performance of nonlinear optimization problems depend on the solution method used, a nontrivial selection requiring expert knowledge. The best modeling practice has traditionally been to keep the solution methods available in mind, limiting the freedom of the modeler. The talk will discuss problem characteristics that may guide the automatic selection of the best fitting method including identification of special problem classes, presolve and deciding between first and second order methods.
  
Title: Tour de MOSEK 7: The Short Version
 Presenting Author: Erling Andersen,CEO, MOSEK ApS, Fruebjergvej 3, Box 16, DK-2100, Copenhagen, Denmark, e.d.andersen@mosek.com
 
Abstract: We present the new features in MOSEK version 7. These features includes support for semi-definite optimization, a new API for building conic optimization models easily and a new mixed-integer optimizer. Finally, we will discuss the expected performance improvements from upgrading from version 6 if time permits.
  
Title: Exact and Heuristic Approaches for Directional Sensor Control
 Presenting Author: Hans Mittelmann,Arizona State University, Box 871804, Tempe AZ, United States of America, MITTELMANN@asu.edu
 Co-Author: Domenico Salvagnin,DEI, University of Padova, Via Gradenigo 6/B, 35131 Padova, Italy, dominiqs@gmail.com
 
Abstract: The Directional Sensor Control problem (DSCP) consists in assigning a direction of view to each sensor with the goal of maximizing information gain on the location of a given set of target objects. In this paper we study and computationally evaluate exact and approximate approaches for the DSCP. In particular, we propose an exact mixed integer convex programming (MICP) formulation and several meta-heuristic approaches.
  
Title: Mixed Integer Programming: Analyzing 12 Years of Progress
 Presenting Author: Roland Wunderling,IBM, Annenstrasse 9, Graz 8020, Austria, roland.wunderling@at.ibm.com
 Co-Author: Tobias Achterberg,IBM, Takustra├če 7, Berlin, Germany, achterberg@zib.de
 
Abstract: In this talk we outline an unbiased way to analyze benchmark results and apply this scheme to assess the contribution of the main components in CPLEX 12.5 to the ability to solve MIPs. We highlight some of the more recent features, in particular the deterministic parallel optimizer.