HEC Montréal, Canada, May 6 - 8, 2013
2013 Optimization Days
HEC Montréal, Canada, 6 — 8 May 2013
MA9 Applications de l'optimisation robuste / Applications of Robust Optimization
May 6, 2013 10:30 AM – 12:10 PM
Location: Dutailier International
Chaired by Erick Delage
10:30 AM - 10:55 AM
Robust Minimum-Cost Flow Problem with Nonlinear Cost Functions
We propose a tractable formulation to address the robust minimum-cost multicommodity flow problem with nonlinear cost functions known only to be bounded above and below by two convex functions. The conservatism of our approach is limited by a budget that prevents too many functions to take on the most pessimistic value. A minimal average transfer time traffic routing problem will be discussed.
10:55 AM - 11:20 AM
Price of Robustness in Inventory Problems
This research addresses multi-period inventory problem under budgeted demand uncertainty. We propose polynomial time approximation that has valuable theoretical properties. An empirical study also suggests that it performs better than currently available approximation methods for this problem.
11:20 AM - 11:45 AM
Robust Optimization of Radiation Treatment in the Presence of Spatiotemporal Uncertainties
The treatment of cancerous tumors with external radiation is planned based on initial data, resulting in strategies that do not vary over the course of the treatment. However, various properties of the tissue change over time. This is the case for spatial information, such as location and size, as well as tissue's response to ionizing radiation over time, particularly when chemotherapeutic agents are combined. Based on clinical cases, we demonstrate that robust plans account for temporal changes and are intrinsically insensitive to deviations from the assumed evolution path.
11:45 AM - 12:10 PM
Accounting for Risk Measure Ambiguity when Optimizing Financial Positions
Since the financial crisis of 2007-2009, there has been a renewed interest towards quantifying more appropriately the risks involved in financial positions. In this work, we show that one can account precisely for (neither more nor less than) what we know of the risk preferences of an investor/policy maker when comparing and optimizing financial positions.