“Daniela di Serafino”
International Doctoral Award - 2025 (first edition)
Award Ceremony for the Winner
Doctoral Dissertation Award Winner and Honorable Mentions
Jury: Stefania Bellavia, Daniela Calvetti, Renato De Leone, Emanuele Galligani, Luca Gemignani
The award ceremony will be held on 12 January 2026 at the Rectorate of the University of Campania Luigi Vanvitelli in Viale Ellittico 31, Caserta, at 10:30 a.m. The winner will hold a seminar on the topic of his PhD dissertation.
Winner
Igor Simunec (Institute of Mathematics, Ècole Polytechnique Fédérale de Lausanne)
Title: Advances in polynomial and rational Krylov methods for matrix functions with applications
PhD School: Scuola Normale Superiore
This thesis addresses two fundamental problems of numerical linear algebra, namely, the evaluation of the action of a matrix function on a given vector and the trace estimation problem for functions of matrices focusing on large-scale problems for which using eigendecomposition or Schur form of the underlying matrix is unfeasible. The committee found that because of the wealth of original theoretical results, the several new algorithms and the attention to implementational details, this thesis best honored the legacy of Daniela di Serafino scientific work.
Honorable Mentions
Miryam Gnazzo (Dipartimento di Matematica, Università degli Studi di Pisa)
Title: On the approximation of the closest singular matrix-valued functions
PhD School: Gran Sasso Science Institute
This thesis contains innovative and computationally appealing results on the numerical approximation of the distance from singularity for matrix-valued functions and nearness problems in numerical linear algebra, based on a deep combination of advanced tools in numerical linear algebra and optimization.
Pierluigi Mansueto (Dipartimento di Ingegneria dell’Informazione, Università degli Studi di Firenze)
Title: Pareto Front Reconstruction of Multi-Objective Optimization Problems
PhD School: Università degli Studi di Firenze
This thesis represents a significant contribution and an innovative approach to the solution of multiobjective optimization problems, providing numerical optimization procedures for approximating the entire Pareto front. It shows a depth of theoretical and practical research that holds promise not only within the scientific community but also for impactful real-world applications.
Mohammadhossein Mohammadisiahroudi (Department of Mathematics and Statistics, University of Maryland, Baltimore County)
Title: Quantum Computing and Optimization Methods
PhD School: Lehigh University
This thesis explores cutting edge intersections of numerical linear algebra methods and interior point methods for mathematical optimization in the framework of innovative and rapidly growing field of quantum computing with a combination of theoretical rigor and innovative computational tools and environments.
Konstantin Riedl (Mathematical Institute, University of Oxford)
Title: Mathematical Foundations of InteractingMulti-Particle Systems for Optimization
PhD School: Technical University of Munich
This thesis presents theoretically well-founded iterative algorithms based on multi-particle models for solving relevant non-smooth and non-convex optimization problems. It provides a deep understanding of the internal mechanisms of Consensus-based Optimization methods that underlie their success. Furthermore, the accompanying open-source software enhances the practical applicability of the proposed methods, making them accessible to practitioners
Ilaria Trombini (Dipartimento di Matematica e Informatica, Università degli Studi di Ferrara)
Title: On the hyperparameters setting for first-order stochastic optimization methods in machine learning
PhD School: Università degli Studi di Parma
This thesis, concerned with the study and the development of the mathematical foundation of stochastic optimization methods for large scale machine learning problems, rigorously examines the theoretical convergence of the methods in various settings. A wide numerical experimentation on a very large data set against the state-of-the art algorithms not only highlight the robustness of the presented methods but also shows excellent implementation skills and understanding of practical challenges.
