4 Presentations

• 
  03:30 PM - 03:55 PM

  Planning data collection of thunderstorm supercells

  • Olivier Quirion-Blais, presenter, DRDC - RDDC
  • Camille Archambault, DRDC - RDDC
  • Abder Benaskeur, DRDC - RDDC
  • Normand Pageau, DRDC - RDDC

  Thunderstorm supercells are complex phenomena subject to many researches. Recent techniques of analysis propose the use of unmanned aerial systems (UASs) to meet the thunderstorms near their trajectory and drop sensors in their core. To obtain good information during an event, many droppings can be done by one or more UASs. To support collection strategies, we propose to develop tools and concepts to optimize the routing of the UASs. The complexity of the problem is that we not only have to assign the meeting points to the UASs and decide in which order to visit them, but the meeting points must also be determined according to the trajectories of the thunderstorms, which are highly uncertain. We also look at other factors adding to the complexity of the problem such as the time windows during which collection is allowed, and the balance between the safety of the UASs and the quality of the measurements. We also face situations where the number of UASs limits the number of measurements. We present a strategy to model the problem and as well as a mathematical formulation.
• 
  03:55 PM - 04:20 PM

  Post-COVID Changes to the Relative Predictability of Women’s Sports Teams and Men’s Sports Teams in World Championships

  • Raymond Stefani, presenter, California State University Long Beach (USA)

  A pre-COVID 2007-2019 data base was gathered to study all team sports that had international recognition, had an official rating system by the governing federation and had a world championship (WC). The data base included 40 WCs for the 13 men’s team sports, 35 WCs for the 12 women’s team sports and a total of 3936 games in which the percentage of games won by each higher-rated team was tabulated. The women’s teams won only 0.25% more than the men’s teams. Post-COVID, nine WCs were contested during 2021and 2022, along with one in 2023, using the same rating system as pre-COVID: four for men and five for women. In all four of the men’s WCs, the percentage of games won by the higher-rated team was lower post-COVID: curling (-6.1%), rugby 7s (-12.9%) and T20 cricket (-10.9% and -6.9%). However, the women’s higher-rated teams had higher percentages of games won in all five WCs post-COVID: curling (+7.3%), rugby 7s (+6.2%), T20 cricket (+7.0%), basketball (+5.5%) and rugby union (+0.4%). During the COVID-era restrictions, women’s physical conditioning and insightful learning of tactical skills increased team cohesiveness, whereas men emerged with less team cohesiveness.

  sports predictions, gender differences, team cohesiveness, post-COVID changes
• 
  04:20 PM - 04:45 PM

  Optimizing database index selection using constraint programming

  • Philippe Olivier, presenter,

  As the size of a database grows, so does the time required to execute query workloads. Database indexes are special data structures that improve the speed of query workloads, at the cost of slower writes and storage space. The Index Selection Problem aims to find the set of indexes that maximizes the speed of the query workloads, under given budget constraints.

  We show that the quality of a set of indexes can be more richly evaluated based on competing criteria, such as the global speed of a query workload vs. the worst-case speed of a query from that workload. We propose a multi-objective constraint programming model that selects good sets of indexes based on these criteria and on the requirements of a database administrator.

  We compare our model with an integer programming formulation for the basic Index Selection Problem, and discuss the importance of solution convergence for certain types of workloads. We present the steps of the solving process when optimizing for multiple objectives.
• 
  04:45 PM - 05:10 PM

  Using a Blackbox Optimization Framework to Toggle Between Deterministic and Stochastic Formulations of the Short-Term Hydropower Scheduling Problem

  • Sara Séguin, presenter, Université du Québec à Chicoutimi
  • Maissa Daadaa, Université du Québec à Chicoutimi
  • Miguel F. Anjos, GERAD, Polytechnique Montréal
  • Kenjy Demeester,

  Solving the short-term hydropower optimization model is complex since the problem is nonlinear, nonconvex and has integer variables related to the choice of the turbines in operation. In addition, the inflows in the reservoirs are uncertain, leading to a stochastic nonlinear integer problem. A formulation of the problem that maximizes the efficiency of the turbines is proposed and the deterministic equivalent of the problem, in which inflows are represented by scenario trees is solved. The problem is solved in a rolling-horizon fashion for a 10 day planning horizon, but solving the problem with uncertain inflows when there is low variability in the inflows is time consuming and does not improve the solution. Therefore, in this talk, we propose a blackbox optimization framework that determines when a deterministic or a stochastic formulation of the problem should be used in order to maximize the energy production. The blackbox optimization variables are the periods when the formulation should be changed, as well as which type of formulation is required (deterministic vs stochastic). For the production planners, choosing when uncertain inflows should be considered is difficult, but the use of an optimization framework to take this decision allows to produce more energy and to make better decisions in an operational context. Numerical results based on real data from a hydropower system located in the province of Québec, Canada are presented.