Research

 Ongoing and past curiosity-driven research projects

Computational Optimal Transport

Background

Our group works on Computational Optimal Transport using Network Flow models and algorithms. We have developed a custom Parallel Network Simplex algorithm, which is competitive with state-of-the-art entropic regularized solvers.

Talks

Combinatorial Optimization

Background

The research group has strong expertise in developing branch-and-price and branch-and-cut algorithms for the solution of NP-hard combinatorial problems. 

Currently, we are actively working on three research topis:

Recent Papers

Talks

Optimization for Healthcare Management

Background

The management of health services is characterized by decision problems with high computational complexity and aspects of uncertainty and dynamicity. We design new mathematical models and algorithms for different problems arising in healthcare management with the aim of providing decision support to decision-makers. 

Hybrid methods involving optimization, simulation, and data science methodologies are developed to consider the inherent stochasticity of these problems and to find the better trade-off between patient-centered and facility-centered objectives. Two main lines of research can be identified:

Papers

Talks

Interpretable Machine Learning

Background

The research group is working on the development of Machine Learning models and algorithms based on interpretable mathematical models that can provide explainable outcomes. Currently, we are developing exact Mixed Integer Programming models for classifying time series.

Papers

Talks

Multimodal Routing

Background

One of the main EU policy priorities under the European Green Deal is to achieve climate neutrality by 2050. Transport is a key player in that task, as it is a major energy consumer and contributes significantly to greenhouse gas emissions. Rail and buses, in particular, can represent a more sustainable mode of transport. To monitor the performance of public transport in the EU and to be able to inform the relevant policy decisions on the topic, we use comprehensive data to compute performance and accessibility-to-opportunities measures associated with different types of public transport. 

Underneath these measures are hidden different Operations Research challenges. One example is the computation of accessibility measures across Europe, which is related to a schedule-based, time-dependent, all-pairs scheduling problem on very large multimodal networks.

Status

Talks

Robust Optimal Power Flow

Background

Development of models and algorithms regarding the Optimal Power Flow problem in a stochastic setting that arises from Renewable Energy Sources.

Status

Talks

Past European Research Projects

Spatial-KWD, Eurostat 2020

Background

The goal of this project was to design and implement an open-source library for the computation of exact and approximate Kantorovich-Wasserstein distances between pairs of very large-scale spatial maps. The exact method is intended for the validation on small to medium size instances of the results obtained via an approximate method (with bound guarantees) that should scale to larger-size instances. 

The core of the algorithm is based on a recent efficient implementation of the Network Simplex on sparse graphs. The core algorithm includes additional features, such as the possibility of dealing with non-convex input maps, the computation of distances between unbalanced maps, and a key incremental sub-procedure that can be used to estimate the spatial density within a focused sub-area.

Deliverables

Talks

FastPath, EU-JRC 2017

Background

The goal of this project was to design and implement a time-dependent shortest-path solver that

The project's main outcome was a detailed technical specification of the proposed time-dependent shortest path solver, along with corresponding source code developed in ANSI standard C/C++17 and an MS-Window solver, callable from the command line.

Deliverables