Summer seminar

Please press Ctrl+F5 to load the most recent version of this page...

This seminar is part of the CUSO Doctoral Programme in Operations Research. For details, please refer to the invitation sent by e-mail.

Dates: June 26 - June 30, 2016
Venue: Hotel Europe, Zinal (VS)

Organization: Norbert Trautmann (University of Bern), Michel Bierlaire (EPFL), Nadine Zumsteg (University of Bern)

Keynote speakers:

Cinzia Cirillo (University of Maryland)

Lecture 1: Revenue Management I - basic concepts and methods

The lecture will provide an outline of the revenue management problem and will discuss its complexities. Research progress will be reported on the following four areas: forecasting, overbooking, seat inventory control, and pricing. The talk will also illustrate methodologies that incorporate customers’ behavior, individual preferences and product choice into the revenue management problem.

Lecture 2: Revenue Management II - choice based methods

Most case studies in revenue management are in the airline sector, because airlines have the longest history of development in revenue management. This lecture will illustrate the difficulties encountered when transferring methods developed for airlines to other sectors (in particular railways). Methods to account for heterogeneity in individuals’ preferences and for dynamic in customers’ choices will be also discussed.

Lecture 3: Optimization methods for econometrics models

In this lecture we discuss the leading classes of optimization methods that are of particular importance in modern econometrics. In particular, methods for choice analysis have progressed enormously in latest decades in order to overcome limitations of early applications. More complex model specifications require simulation, the maximization of ill-behaved functions, and are computationally intensive. Optimization based on Monte Carlo and quasi Monte-Carlo simulations, adaptive Trust Region methods and on the Genz algorithm will be illustrated and results obtained for specific choice models will be discussed.

Lecture 4: A multi-disciplinary approach to Big data analytics

Modern societies continuously produce data. This data revolution is providing researchers with an increasing amount of information from relatively low-cost sources and requires new methods for analysis and research collaboration among scientists from different disciplines. Recent work on travel time estimation from Vehicle Probe Project data will be presented. Two techniques for travel time prediction in real-time will be illustrated. The first identifies a relatively small number of hidden variables using methods from statistical machine learning and stream computing. The second develop Bayesian and approximate Bayesian methods, borrowing ideas from the existing literature on the rapidly growing interdisciplinary field of small area estimation, in order to produce instantaneous prediction of travel time.


Pierre L'Ecuyer (University of Montreal)

Title: Stochastic Simulation

We examine some key ideas and methods to simulate stochastic systems efficiently by computer and to estimate some of their distributional properties. Principles, techniques, and theory will be reviewed and illustrated by examples from various areas. We will review Monte Carlo methods for estimating an  integral (an expectation), a quantile, an entire distributions (densities), an optimum, etc. We will discuss uniform and non-uniform random number generators with multiple streams and substreams. We will examine various techniques to reduce the variance and more generally improve the efficiency of Monte Carlo simulations. We will see that rare event simulation has many more applications than one might think at first sight.

Lecture 1: Introduction to Monte Carlo Simulation

Stochastic models and simulation, Monte Carlo to estimate an integral, a function of several integrals, a quantile, a distribution, a density, and a conditional density. Confidence intervals. Multiple streams of random numbers. Comparing systems with common random numbers. Discrete-event simulation.

Lecture 2: Random Number Generation

Uniform random number generators, multiple streams and substreams, theoretical analysis of uniformity vs empirical testing, main classes of recommendable generators, combined generators, nonlinear generators, generators for parallel computers. Nonuniform generators, inversion, rejection method, using changes of variables.

Lecture 3: Variance Reduction Methods and Efficiency Improvement

Measures of efficiency, variance versus computing time, variance reduction, common random numbers, antithetic variates, quasi-Monte Carlo methods, stratification, conditional Monte Carlo, multilevel Monte Carlo, variance reduction for estimating a derivative.

Lecture 4: Variance Reduction in Rare Event Contexts

Rare event simulation, change of probability measure, importance sampling, splitting, robustness to rare events. Applications of rare event simulation.

Tentative Program

  1. Baumann Philipp, University of Bern
  2. Bierlaire Michel, EPFL
  3. Binder Stefan, EPFL
  4. Cirillo Cinzia, University of Maryland
  5. Daudet Laurent, EPFL
  6. De Lapparent Matthieu, EPFL
  7. Fernandez Antolin Anna, EPFL
  8. Forrer Salome, University of Bern
  9. Galby Esther, University of Fribourg
  10. Gallay Olivier, University of Lausanne
  11. Gnägi Mario, University of Bern
  12. L'Ecuyer Pierre, University of Montreal
  13. Maknoon Yousef, EPFL
  14. Nikolic Marija, EPFL
  15. Pacheco Paneque Meritxell, EPFL
  16. Ries Bernard, University of Fribourg
  17. Rihm Tom, University of Bern
  18. Scarinci Riccardo, EPFL
  19. Sharif Azadeh Shadi, EPFL
  20. Strub Oliver, University of Bern
  21. Trautmann Norbert, University of Bern
  22. Widmer Marino, University of Fribourg
  23. Zimmermann Adrian, University of Bern

Bastin, F., Cirillo, C., & Toint, P. L. "Application of an adaptive Monte-Carlo algorithm for mixed logit estimation", Transportation Research Part B, 40, 7 (2006), 577-593.

Bastin, F., Cirillo, C., & Xu, R. "A dynamic formulation for car ownership modeling", Transportation Science, 50, 1 (2016), 322–335.

Cirillo, C., & Hetrakul, P. "A latent class choice based model system for railway optimal pricing and seat allocation", Transportation Research Part E, 61, 1 (2014), 68-83.

Papers Pierre L'Ecuyer

Avramidis, A. N., & L'Ecuyer, P. "Ecient Monte Carlo and Quasi-Monte Carlo Option Pricing Under the Variance-Gamma Model", Management Science, 52, 12 (2006), 1930-1944.

Avramidis, A. N., Chan, W., Gendreau, M., L'Ecuyer, P., & Pisacane, O. "Optimizing Daily Agent Scheduling in a Multiskill Call Center'', Earlier version in CIRRELT Report 2007-44. European Journal of Operations Research, 200, 3 (2010), 822-832.

Botev, Z. I., L'Ecuyer, P., Simard, R., & Tun, B. "Static Network Reliability Estimation under the Marshall-Olkin Copula", ACM Transactions on Modeling and Computer Simulation, 26, 2 (2016), Article 14.

Cezik, T., & L'Ecuyer, P. "Staffing Multiskill Call Centers via Linear Programming and Simulation'', first draft in 2004, Management Science, 54, 2 (2008), 310-323. The data for the large example is available here.

L'Ecuyer, P. "Quasi-Monte Carlo Methods with Applications in Finance", Finance and Stochastics, 13, 3 (2009), 307-349.

L'Ecuyer, P. "Random Number Generation with Multiple Streams for Sequential and Parallel Computers", short review article, Proceedings of the 2015 Winter Simulation Conference, IEEE Press, 2015, 31-44.

L'Ecuyer, P. "SSJ: A Java Library for Stochastic Simulation", DIRO, Universite de Montreal,

L'Ecuyer, P., & Buist, E. "On the Interaction Between Stratification and Control Variates, with Illustrations in a Call Center Simulation'', Journal of Simulation, 2 (2008), 29-40.

L'Ecuyer, P., & Perron, G. "On the Convergence Rates of IPA and FDC Derivative Estimators for Finite-Horizon Stochastic Simulations'', Operations Research, 42, 4 (1994), 643--656.

L'Ecuyer, P., & Simard, R. "TestU01: A C Library for Empirical Testing of Random Number Generators", ACM Transactions on Mathematical Software, 33, 4 (2007), Article 22, 40 pages.

L'Ecuyer, P., Lecot, C., & Tun, B. "A Randomized Quasi-Monte Carlo Simulation Method for Markov Chains", Operations Research, 56, 4 (2008), 958-975.

L'Ecuyer, P., Blanchet, J., Tun, B., & Glynn, P. W. "Asymptotic Robustness of Estimators in Rare-Event Simulation", ACM Transactions on Modeling and Computer Simulation, 20, 1 (2010), Article 6, 41 pages.