SISSA MathLab

📘 A new book chapter by Dario Coscia and Gianluigi Rozza, “Bayesian Perspectives in Scientific Machine Learning: Methods, Challenges, and Opportunities”, is featured in the eBook Numerical Analysis and Scientific Computing: an Anthology, published by Springer Nature in the Lecture Notes in Computational Science and Engineering series.

📊 The chapter explores Bayesian approaches to Scientific Machine Learning (SciML) and their potential to make data-driven scientific models more reliable. It reviews their role in uncertainty quantification, continual learning, interpretability, and robustness, while highlighting current challenges and promising research directions.

🔗 Preview the chapter:
https://link.springer.com/chapter/10.1007/978-3-032-33428-2_11

Bayesian Perspectives in Scientific Machine Learning: Methods, Challenges, and Opportunities Scientific Machine Learning (SciML) is an interdisciplinary field combining data-driven machine learning with physics-based modeling, for accurate predictions across scientific and engineering applications. Despite the great successes of SciML, the widespread...

🆕 A new article by Armin Sheidani, Michele Girfoglio, Annalisa Quaini and Gianluigi Rozza has been published in Results in Engineering (Elsevier): “Enhancing the accuracy of under-resolved numerical simulations of atmospheric flows with super resolution”.

🌦️ The study explores how deep learning-based super-resolution techniques can improve coarse-grid simulations of mesoscale atmospheric flows while containing computational costs. Among the approaches tested, a multi-scale convolutional neural network achieves the best balance of accuracy, robustness and computational efficiency, particularly for more complex flows.

🔗 Read the full article:
https://www.sciencedirect.com/science/article/pii/S2590123026036984



𝘍𝘪𝘨𝘶𝘳𝘦 𝘢𝘥𝘢𝘱𝘵𝘦𝘥 𝘧𝘳𝘰𝘮 𝘵𝘩𝘦 𝘢𝘶𝘵𝘩𝘰𝘳𝘴' 𝘱𝘳𝘦𝘱𝘳𝘪𝘯𝘵 𝘢𝘷𝘢𝘪𝘭𝘢𝘣𝘭𝘦 𝘰𝘯 𝘢𝘳𝘟𝘪𝘷

🧮 Our Gianluigi Rozza will be one of the speakers at the round table “Mathematics, Innovation and Knowledge Transfer”, bringing our perspective on the role of applied mathematics in connecting research, technology, and innovation.

🧮 The hidden mathematics behind innovation

❓What connects fundamental research to real-world solutions?
At the round table "Mathematics, Innovation and Knowledge Transfer", leading voices from academia and industry will discuss how mathematical research fuels technological, economic and social development.

📅 8 September 2026, 15:30
📍 Budinich Lecture Hall, ICTP – Abdus Salam International Centre for Theoretical Physics
🎟️ Open event

The event is organised within the 1st Italy–China Mathematical Bilateral Meeting by ICTP: International Centre for Theoretical Physics, SISSA and INdAM – Istituto Nazionale di Alta Matematica "Francesco Severi", together with the Beijing International Center for Mathematical Research at Peking University, Great Bay University and Dongguan University of Technology.

➡ https://www.sissa.it/news/hidden-mathematics-behind-innovation-italy-and-china-meet-trieste

🆕 A new article by Lander Besabe, Michele Girfoglio, Simona Perotto, Annalisa Quaini, and Gianluigi Rozza has been published in Advances in Computational Science and Engineering (): An isotropic recovery-based error estimator algorithm for mesh adaptation in a finite volume environment with application to atmospheric flows.

💨 The paper introduces an adaptive mesh refinement method for atmospheric simulations, dynamically increasing resolution where needed. The approach improves accuracy and stability while reducing computational costs compared with uniformly fine meshes.

🔗 Read the full article:
https://www.aimsciences.org/article/doi/10.3934/acse.2026012

📢 Looking for an internship in AI and engineering?

💼 NVIDIA is offering a 6-month internship in Zurich, Switzerland for Master's students in Mathematics, Data Science and AI, Computer Science, and Engineering.

⚙️ The internship will be co-supervised by SISSA mathLab and will focus on AI surrogate models and agentic approaches for the preprocessing of computer-aided engineering (CAE) applications.

📅 The starting date is flexible and can be agreed upon until December 1, 2026.

📩 For more information and applications, please contact Davide Fransos and Gianluigi Rozza:
https://www.linkedin.com/in/davidefransos/
https://www.linkedin.com/in/gianluigi-rozza-8447903/

🌍 This week, a large part of our group will be in Munich for WCCM-ECCOMAS 2026, the 17th World Congress on Computational Mechanics and 10th European Congress on Computational Methods in Applied Sciences and Engineering, supported by IACM (International Association for Computational Mechanics), European Community on Computational Methods in Applied Sciences, and GACM (German Association for Computational Mechanics).

🎤 Throughout the congress, our Gianluigi Rozza, Pasquale C. Africa, Dario Coscia, Lorenzo Fabris, Isabella C. Gonnella, Rahul Halder, Anna Ivagnes, Hammad Khaliq, Gaspare Li Causi, Federico Pichi, and Lorenzo Tomada will present contributions spanning reduced order modeling, scientific machine learning and uncertainty quantification—with applications including naval engineering, sustainable mobility, and cardiac electrophysiology.

📍 Gianluigi Rozza and Federico Pichi are also chairing minisymposium series during the congress. In addition, Gianluigi will chair the plenary lecture "Agentic Scientific Machine Learning" by George Karniadakis.

👉 Discover the full scientific program: https://wccm-eccomas2026.org/event/programme

UPDATE: Applications are closed.

📢 A new opportunity for early-career mathematicians at SISSA - Scuola Internazionale Superiore di Studi Avanzati!

🆕 The SISSA Mathematics Area is offering a new PhD fellowship in mathematical analysis, modeling and applications, funded by Fincantieri SpA.

🚢 The research project deals with the development of surrogate methods for the parametric optimization of the structural analysis of passenger ships.

⏰ The application deadline is August 27, 2026.

🔗Apply here: https://www.sissa.it/bandi/selection-conferment-phd-fellowship-funded-fincantieri-spa

www.sissa.it

🆕 A new review article by Shenhui Ruan, Andreas G. Class, and Gianluigi Rozza has been published in “Archives of Computational Methods in Engineering” (Springer Nature): “A Structured Review of Reduced Order Modeling for Domain Decomposition Problems: State of the Art and Perspectives”.

📚 The review provides a comprehensive overview of reduced order modeling techniques combined with domain decomposition, an approach that accelerates large-scale engineering simulations by dividing complex problems into smaller subdomains and constructing local reduced models.

👉 Read the full article:
https://link.springer.com/article/10.1007/s11831-026-10690-9



𝘐𝘮𝘢𝘨𝘦 𝘢𝘥𝘢𝘱𝘵𝘦𝘥 𝘶𝘯𝘥𝘦𝘳 𝘵𝘩𝘦 𝘊𝘊 𝘉𝘠 4.0 𝘭𝘪𝘤𝘦𝘯𝘴𝘦: http://creativecommons.org/licenses/by/4.0/

🆕 A new article by Lorenzo Tomada, Federico Pichi, and Gianluigi Rozza has been published in the Journal of Computational Physics (Elsevier): “Latent Dynamics Graph Convolutional Networks for model order reduction of parameterized time-dependent PDEs”.

🧠 The paper introduces a data-driven reduced-order modeling framework that combines graph neural networks with a low-dimensional latent representation of dynamical systems. The proposed approach enables the efficient reconstruction of solutions to parameterized time-dependent PDEs, including problems defined on complex geometries, while reducing computational costs.

👉 Read the full article:
https://www.sciencedirect.com/science/article/pii/S0021999126005024

🆕 A new article by Harsh*th Gowrachari, Mattia Giuseppe Barra, Giovanni Stabile, Gianluca Bazzaro and Gianluigi Rozza is now available online as a pre-proof in Results in Engineering (Elsevier): “Data-driven reduced order model for residence time distribution analysis of an industrial-scale continuous casting tundish”.

⚙️ The paper presents a data-driven reduced order model for predicting the residence time distribution in an industrial continuous-casting tundish, a key component of the steel production process. By accurately reproducing the results of full-order simulations at a fraction of the computational cost, the proposed approach enables efficient real-time analysis and supports process monitoring and optimization in industrial environments.

👉 Read the article:
https://www.sciencedirect.com/science/article/pii/S2590123026025867