The definitive version of the programme is available here. The list of speaker is available here. The booklet of the workshop is available here.

All talks will be given in Room "Alberto Bertoni" (Aula Magna) located on the ground floor of the venue (click here for more information on how to reach the venue).
The coffee breaks will be served in the nearby Aula Epsilon located on the same floor.
The registration desk will be in the Aula Epsilon on February 15th morning. Attendees and speakers who arrive after the scheduled registration time can ask for their badge and conference bag to the organizers during the conference.

Wednesday 15th	+

09:00-10:00	Registration
10:00-10:30	Opening
10:30-11:30	Bin Packing
D. Croci	A Support Planes Beam Search Algorithm for the Pallet Loading Problem
R. Giusti	Logistics Capacity Planning with Carrier Selection - A Stochastic Bin Packing Methodology
C. Turbian	Models and Methods for a 2D Bin Packing Problem in the Sheet Metal Industry
11:30-12:10	Disaster Management
C. Truden	Allocating temporary blackout supply points within walkable distances
S. Al-Natoor	Temporary logistic hubs prepositioning for preparedness and response disaster operations
12:10-14:00	Lunch
14:00-15:00	Interactions of Combinatorial Optimization and Machine Learning – 1
S. Milanesi	The BeMi Stardust: a Structured Ensemble of Binarized Neural Networks
A. Consolo	Randomized regression trees: a model variant and a decomposition training algorithm
F. Saccomanno	A Reinforcement Learning approach to solve the bin packing problem
15:00-15:40	Travelling Salesman problems
F. Michelotto	An Exact and Heuristic Approach for the Traveling Salesman Problem with Drone and Variable Drone Speed Selection
E. Vercesi	On the generation of Metric TSP instances with a large integrality gap by branch-and-cut
15:40-16:10	Coffee Break
16:10-16:50	Decomposition Methods
D. Palasgo	Algorithms for the Pickup and Delivery Problem with Time Windows and Last-in-first-out Loading
R. Praxedes	A unified exact approach for a set of vehicle routing problems with simultaneous pickup and delivery
16:50-17:50	Manufacturing and automated warehouses
K. Gdowska	Automated Analysis of Shewart Charts in a High Throughput Manufacturing Using Machine Learning Models
D. Zehetner	Towards Large Scale Collaborative Production Planning in Additive Manufacturing
V. Mosca	An adaptive large neighborhood search for the order picking process: the case of a retail distribution company in Italy

Thursday 16th	+

08:40-10:00	Applications to Finance and Revenue Management
E. Concas	Dynamic Bundles Offer Management in an Airline Context
M. Luzzi	Application of auctions mechanisms in restaurant business
D. Merolla	A non-parametric model for constrained retail assortment optimization
C. Gambella	FICO Decision Optimizer – Generating causal predictive models
10:00-10:30	Coffee Break
10:30-11:30	Plenary Talk
Paolo Boldi	University of Milan Centralities in graphs. The strange case of directedness & monotonicity
11:30-12:30	Scheduling
E. Lancini	Poly/Mono-chromatic Edge Coloring for Job Scheduling
A. Masone	Minimizing the sum of completion times in the AGV Scheduling Problem with battery constraints
A. Mancuso	An online optimization-based tool for surgery scheduling and re-scheduling
12:30-14:00	Lunch
14:00-15:00	Sponsor Session
Optit	District Energy Production Management Optimisation: beyond the unit committment problem
Multiprotexion	Optimization problems arising in surveillance companies
15:00-15:40	Nonlinear Programming
B. Manca	The Ellipsoidal Separation Machine
C. Piermarini	Computing Negative Curvature Directions for Large Scale Optimization: exploiting SYMMBK
15:40-16:10	Coffee Break
16:10-17:30	Energy
A. Bernardelli	A linear approximation for a stochastic optimal power flow problem based on wind energy sources
S. Cordieri	Transactive energy trading using a Solar Organic Rankine Cycle
N. Frieß	Simulation and Optimization of Renewable Energy Communities
G. Micheli	Operational Equilibrium of Electricity and Natural Gas Systems with Bi-Directional Energy Flows
17:30-17:45	AY Meeting
18:30-20:15	Leonardo3 Exhibition, Piazza Della Scala, at the entrance of Galleria Vittorio Emanuele
21:00	Social Dinner @ "Casa Lodi", 3, Via Cappellari

Friday 17th	+

9:00-10:00	Public and Multimodal Transportation
C. Tomasi	Evaluating Public Transport by Multimodal Schedule-based Routing
M. Truvolo	A MILP model for multimodal logistics cooperation
M. Tessitore	On the Fragility of a Train Timetable
10:00-10:30	Coffee Break
10:30-11:30	Plenary Talk
Claudia D'Ambrosio	Ècole Polytechnique, Paris Mathematical optimization to guarantee safety in Urban Air Mobility
11:30-12:30	Multiobjective optimization
V. Bonomi	Fairness in Home Healthcare: a lexicographic approach to investigate the impact of conflicting stakeholder’s goals
M. Cerulli	A bilevel pricing and routing problem
M. Cosmi	Mathematical programming for managing the profitability-sustainability trade-off in complex chemical value chains
12:30-14:00	Lunch
14:00-15:00	Interactions of Combinatorial Optimization and Machine Learning – 2
L. Baty	Winning Approach for the EURO-NeurIPS Dynamic Vehicle Routing Competition
N. Gòomez-Vargas	Explainability in predict-and-optimize
L. Bonasera	Optimal Shapelets Tree for Time Series Interpretable Classification
15:00-16:00	Sponsor Session
Reply Target	Ship Certifications: Operational search with Genetic Algorithms
Reply LAIFE	From data to smart optimization science: a real case study in the healthcare sector
16:00-16:40	Software
S. Basso	PathWise: an open-source library for the Resource Constrained Shortest Path
F. Da Ros	JuLeS: A Julia Framework for White-box Metaheuristic Design
16:40-17:20	Electric Vehicles
F. Taverna	Waste collection with EVs: a MILP formulation
E. Scalzo	A GRASP for a Green Location Problem
17:20-17:30	Closing

---

Invited Speakers

Paolo Boldi

Paolo Boldi is full Professor at the Università degli Studi di Milano since 2015, where he is currently the co-ordinator of the Computer Science Degree.

Thursday 16th, 10:30 - 11:30: "Centralities in graphs. The strange case of directedness & monotonicity"	+

Is it always beneficial to create a new relationship (have a new follower/friend) in a social network? This question can be formally stated as a property of the centrality measure that defines the importance of the actors of the network. In this lecture, we will explain the role of centrality, introduce the most important centrality scores, and discuss some notions of monotonicity. In particular, we will present some recent results about the effect of directedness.

Claudia D'Ambrosio

Claudia D'Ambrosio is CNRS research director since 01/10/19 and adjunct professor at Ècole Polytechnique (Paris, France) since 01/09/19.

Friday 17th, 10:30 - 11:30: "Mathematical optimization to guarantee safety in Urban Air Mobility"	+

Urban air mobility (UAM) represents a new era for public transportation. For example, passengers’ transportation via eVTOLs, i.e., electric flying vehicles, will allow exploiting the sky to help smooth ground traffic in densely populated areas. Clearly, one of the biggest challenges in UAM is to ensure safety. We present how mathematical optimization can be applied in this context to solve the tactical deconfliction (TD) problem. TD is an online problem aimed at identifying and fixing potential conflict between each pair of aircraft passing through the same air space. While TD is a well-known problem in air traffic management, its specialization to the urban air mobility case has some peculiarities. Our mathematical model is one of the first attempts to fully formalize such a problem and to propose a solution to it. We test our method on three different scenarios, where the following sources of conflict are generated: i. standard conflicts due to delay w.r.t. the original planning; ii. a priority flight that has to be integrated into the original planning; iii. an intruder flight. In the computational study, we, thus, compare our approach against a variant considering only pairwise conflicts, on three sets of realistic instances.

---

Industrial Talk

Optit - Caterina Tamburini

Thursday 16th, 14:00 - 14:30: "District Energy Production Management Optimisation: beyond the unit committment problem"	+

As decarbonization becomes a global priority, there is a need to increase the efficiency of energy production. Combined Heat and Power (CHP) is an energy efficient technology that generates electricity and captures the heat that would be wasted otherwise, in order to provide thermal energy, often used to feed district heating networks. Unit Commitment (UC) is a key problem in this context. The goal in UC is to determine a schedule for the machines that maximize the operative margin, satisfying a forecasted heat demand coming from a district heating network as well as functional and regulatory constraints deriving from system composition and placement. This gives rise to UC problems. In this work, we present some key features of the UC problem and our approach to perform the optimization of real world CHP systems, in both short and long term cases. Along with more classic and Operations Research based perspectives, we show how machine learning and data driven models can support and integrate the optimization process. We formulate and solve a Mixed Integer Linear Problem for the short-term optimization problem. Since the ability to predict the heat demand of the network is a relevant factor for the accuracy of the energy production plan, we developed a forecasting module that, given a series of historical data, automatically builds accurate prediction models. Further, we present a metaheuristic algorithm for the long-term case, founded on a time-based decomposition of the problem, that leverage on a clustering module and on the above mentioned MILP for the short-term optimization. Moreover we show how the MILP for the short-term optimization can play a central role in the automatization of some complex processes, such as trading in power market sessions and managing plants where machines are operated in series. In the first case, we have a portfolio of plants and we use the optimization model to assess cost-opportunity of alternative scenarios in order to maximize the revenues obtained trading electric power across several market sessions. In the latter case, operating temperature and water flows become important decision variables, introducing some nonlinear and nonconvex relations that make the optimization problem very hard to be solved.

Multiprotexion - Valentina Morandi

Thursday 16th, 14:30 - 15:00: "Optimization problems arising in surveillance companies"	+

Multiprotexion is a company that deals with the surveillance of international transport vehicles and facilities with a telematics escort option for vehicles/companies with a high monetary value of transported goods. The company provides multiple services including the 24/7 operational centre to manage the alarms received from the drivers and the assisted fleet route guidance for fuel savings. We will illustrate how Operations Research has been and will be effective in evaluating and decision-making in provided services, and which optimization techniques have been effectively used. Many of the problems arising in Multiprotexion, and possibly in many companies doing remote surveillance, can be solved through optimization and machine learning approaches. Many have been already identified and will be faced in the next years. Some promising research and business areas will be proposed and discussed during the talk.

TARGET REPLY - Andrea Mosconi

Friday 17th, 15:00 - 15:30: "Ship Certifications: Operational search with Genetic Algorithms"	+

The client is a world leader in ship certification, classification, verification of conformity and testing. Ships need continuous checkings, different controls require different qualifications for the surveyors that perform them. Every day, hundreds of ships approach some of the hundreds of ports where the client has an office. How to choose which is the best suitable surveyor for each of the ships? In this session we will show how to: Use Open Data: Vesseltracker (for ship fleet scheduling) and OpenStreet for georeferentiation of ports and offices Manage a scalable big-data ready architecture based on MongoDB Define a “best match” function to rank surveyors choose the best surveyor for each ship by mean of Genetic Algorithm based operational search

LAIFE REPLY - Nicholas Draghetti

Friday 17th, 15:30 - 16:00: "From data to smart optimization science: a real case study in the healthcare sector"	+

The problem of scarcity of resources to cope with an increasing number of demands is a problem that is assuming great prominence in every sector, not least the health sector. This issue came to the forefront during the Covid 19 pandemic where, faced with an increasing number of requests, an increasingly optimized approach to the use of available resources was required. During the talk, the approach put in place to identify and subsequently solve a specific "business critical" problem in healthcare management will be presented. Specifically, it will be shown how the joint use of data and optimization models is the key element in decision support within a real-world context. In particular, by exploding the pipeline set up, the data-driven approach (based on the proprietary hegos solution) adopted in order to identify the “business-critical” issue under study will be presented in the first instance. Next, the mathematical modeling approach of the identified problem (set covering problem) will be shown through the comparison of 2 heuristic algorithms (greedy algorithm vs hunt-colony optimization) showing strengths and weaknesses of the heuristics considered with respect to the specific case study

---

Social Events

Social Event - 16/02 18:30-20:15 - Leonardo3 Exhibition

On February 16th, we will take a guided tour of Leonardo3 Exhibition. The tour will begin from 18:30 and will last for 1h 45 minutes.
The tour is free of charges for all partecipants (while seats last)!
The appointment is at 18:15 at Piazza Della Scala, at the entrance of Galleria Vittorio Emanuele. Be on time!

From the venue of the workshop, the fastest way to reach the museum is by metro: from the venue reach station Piola of metro line L2 (green line). Take the metro towards Assago/Abbiategrasso and get off at Loreto. At Loreto transfer to metro line L1 (red line) towards Rho/Biseglie, and get off at Duomo. From Piazza Duomo you can reach Piazza Della Scala heading north through Galleria Vittorio Emanuele.

Social Dinner - 16/02 21:00 @ "Casa Lodi"

On February 16th, after the social event, for those who expressed desider to join and executed the payment successfully, a dinner has beeen organized at "Casa Lodi", in via Cappellari 3, near Piazza Duomo.
The dinner will take place starting at 21:00. Special arrangements will be offered for people with special dietary requirements.
After the guided tour at the museum we will directly move to the restaurant.
For those willing to reach the restaurant on their own, The appointment is at 20:50 in front of the restaurant: from the venue of the workshop reach Piazza Duomo following the instruction above. The restaurant is at 1 minute walk from Piazza Duomo, heading south towards Piazza Diaz.