Fabian Torres

This page reports only the academic work registered in the databases of the Transport and Mobility Laboratory, and is not necessarily a comprehensive list of the work by Fabian Torres.

Personal webpage

Technical reports [1]

2023

Haering T., Legault R., Torres F., Ljubic I., Bierlaire M. (2023). Exact Algorithms for Continuous Pricing with Advanced Discrete Choice Demand Models. Report TRANSP-OR 231211


@TechReport{HaerLegaTorrLjubBier2023,
  author = {Tom Haering and Robin Legault and {Fabian Alejandro} Torres and Ivana Ljubic and Michel Bierlaire},
  title = {Exact Algorithms for Continuous Pricing with Advanced Discrete Choice Demand Models},
  institution = {Transport and Mobility Laboratory, Ecole Polytechnique F\'ed\'erale de Lausanne},
  year = {2023},
  type = {Technical Report},
  number = {TRANSP-OR 231211},
  address = {Lausanne, Switzerland}}}

Research projects [1]

2022

The Elevator Dispatching Problem with Destination Control

The Elevator Dispatching Problem with Destination Control We propose developing a Branch-and-Price (BP) exact method that can provide optimal solutions to instances. The size of the instances that BP can solve has to be determined with computational experiments once the method is implemented and cannot be known beforehand. BP is known to provide good lower bounds that can also be used to evaluate heuristic solutions. The aim of this study is to provide optimal solutions to the deterministic-static EDPDC where we have perfect information and compare with Schindler�s heuristic solution.
- Sep 22, 2022 - Apr 01, 2023
- Sponsor: Schindler
- Principal investigator: Michel Bierlaire
- Project manager: Fabian Torres

EPFL courses [4]

Spring 2024

Decision-aid methodologies in transportation.
- Level: Master
- Section: Civil Engineering
- Lecturers: Fabian Torres, Evangelos Paschalidis
- Teaching assistant: Tom Haering

Fall 2023

Introduction to optimization and operations research .
- Level: Bachelor
- Sections: Mechanical Engineering, Civil Engineering, Environmental Sciences and Engineering, Electrical and Electronics Engineering
- Lecturer: Michel Bierlaire
- Teaching assistants: Fabian Torres, L�a Ricard, Marija Kukic, Tom Haering, Negar Rezvany

Spring 2023

Decision-aid methodologies in transportation.
- Level: Master
- Section: Civil Engineering
- Lecturers: Fabian Torres, Evangelos Paschalidis
- Teaching assistants: Selin Atac, Tom Haering

Fall 2022

Introduction to optimization and operations research .
- Level: Bachelor
- Sections: Mechanical Engineering, Civil Engineering, Environmental Sciences and Engineering, Electrical and Electronics Engineering
- Lecturer: Michel Bierlaire
- Teaching assistants: Fabian Torres, Selin Atac, Marija Kukic, Negar Rezvany

Lectures [1]

2024

Introduction, Logit Estimation and Testing.
- Discrete Choice Analysis: Predicting Individual Behavior and Market Demand, EPFL, Lausanne, Switzerland
- Feb 11, 2024
- Lecturer: Tom Haering
- Assistants: Negar Rezvany, Nicola Ortelli, Fabian Torres

MSc theses [2]

2024

Optimizing Gas Storage Strategies: Addressing Price Fluctuations and Leakage through a Mercuria Case Study, Antoine Goupil de Bouille (Mathematics)
- Supervision: Michel Bierlaire, Cloe Cortes Balcells, Fabian Torres
- External expert:

2022-2023

Orders Allocation Management with Upstream Consolidation, Olivier Laferrere (None)
- Supervision: Fabian Torres, Michel Bierlaire
- External expert:

Semester projects [4]

2024

A computational comparison of open-source branch-and-price solvers, Jules Schneuwly

Supervision: Léa Ricard, Fabian Torres, Michel Bierlaire
Column generation algorithms are relevant for addressing problems featuring a large number of decision variables (i.e., columns), commonly encountered in transportation and logistics problems such as vehicle routing and scheduling of staff or resources. When these optimization problems involve integrality constraints, column generation algorithms can be integrated into a branch-and-bound tree. This solution method is also referred to as branch-and-price. Historically, such algorithms have been predominantly developed within closed-source solvers, which raises issues regarding the comparability and reproducibility of computational experiments. Recently, the academic community has introduced open-source branch-and-price solvers to address these concerns. This project aims to compare two such open-source solvers in terms of computational time and flexibility across various families of mixed-integer problems (MIPs). The selected open-source solvers are:

Solve the problem of the EPFL campus in MOBi,, Anne-Valérie Preto (Civil Engineering)

Supervision: Negar Rezvany, Fabian Torres, Michel Bierlaire

Construction of an Itinerary Optimization Algorithm for a Trip Planning App, Gabriel André Veigas Marques

Supervision: Fabian Torres, Tom Haering, Michel Bierlaire
The proliferation of mobile and web applications for travel planning necessitates advanced optimization algorithms to efficiently organize itineraries. This semester project aims to develop an innovative itinerary optimization algorithm for a trip planning application, addressing the Generalized Price-Collecting Traveling Salesman Problem with Resource Constraints. The algorithm's primary objective is to automate the organization of travel elements (hotels, restaurants, and activities) into a coherent daily schedule, optimizing for travel time, cost, and user preferences. The project begins by leveraging a recommendation algorithm that selects a set of events based on user input, consisting of activities, dining options, and accommodations. These events are organized into an extended set, allowing for greater flexibility and optimization potential. The core of the project involves the construction of an optimization algorithm that integrates travel time computations between events, using Google Maps API, and optimizes the sequence of events to minimize overall travel time, while considering the rank and cost of each event. Initial objectives include the implementation of travel time calculations and the development of a basic itinerary optimization model. Subsequent phases will expand the event set and refine the optimization model to balance travel time, event desirability, and cost, ultimately producing an itinerary that offers the best possible experience for the user.

2023-2024

Modeling Individual Activity Schedules and Behavior Changes in COVID-19 Using Metaheuristic Optimization, Pierre Alexis Hellich

Supervision: Cloe Cortes Balcells, Fabian Torres, Michel Bierlaire
The primary objective of this project is to develop an innovative approach that addresses these limitations by using a metaheuristic optimization technique already developed by TRANSP-OR lab, to generate personalized activity schedules for individuals. We will use synthetic population data including information about individuals� characteristics and a comprehensive activity network with coordinates for each facility. The proposed methodology aims to create schedules that 1 optimize a given utility function for every individual. This utility function will consider individual preferences, activity durations, and the impact of COVID-19 restrictions on activity choices. The expected outcomes of this project include a novel framework that provides insights into how individuals modify their behavior and activity plans during periods of COVID-19 restrictions. By integrating a comprehensive network and allowing individuals to select activities from outside their desired schedules, the proposed model will contribute to more accurate and comprehensive epidemiological models. The output of this model will be used as an input for the Interdisciplinary Behavioral Model (IBM) developed by the TRANSP-OR lab. This model integrates individual behavior into both epidemiological and mobility models by combining interdisciplinary research. It integrates epidemiology, mobility, and behavioral modeling, to create a comprehensive model that considers individual-specific behavior in the context of testing, infection levels, and daily activities. Policymakers can leverage these models to make informed decisions regarding the implementation of restrictions and the development of effective risk mitigation strategies.

Michel Bierlaire, EPFL. Sun Jun 30 18:50:32 2024