Harnessing Rare-Event Methods and Machine Learning to Advance Molecular Modeling.

Palestrante: Prof. Dr. Enrico Riccardi – University of Stavanger

Abstract: Molecular modeling is a computational approach able to sample systems at atomic and subatomic resolution. Its accuracy and simulation is limited by the time and dimension scale accessible by the available computational power. In the study of chemical reactions, conformation changes, phase transitions, permeations, the computational requirements further increase as the initial and final stable states have to be sampled, as well as the transition between them. At molecular scale, such transitions are often extremely spurious. Rare event methods are intended to boost the simulation power of molecular models for these cases. Path sampling, in particular, boosts the sampling efficiency without altering the ‘natural’ system dynamics, allowing the study of un-probable events. The presentation will have three parts. In the first, I will introduce a path sampling rare event method and software (www.pyretis.org) that we have recently developed. In the second part, I will introduce a set of studies where a rare events method is applied (CO2 adsorption in confined space, sodium batteries electrolytes deterioration and proton transfer reaction in water). In the last part, I will introduce a procedure to learn how to effectively train machine learning approaches for molecular dynamics data (e.g. convolutional neural network) to effectively sample transitions from sparse data.

Enrico Riccardi is a Professor of Computational Engineering in the Department of Mathematics and Physics at the University of Stavanger. He joined UiS as an Associate Professor in 2023 and was promoted to Full Professor in 2025. His academic path includes a Postdoctoral Researcher position at the University of Oslo and nearly a decade at NTNU, where he worked both as a Post-Doc and later as a Researcher. Throughout these roles, he consolidated his expertise in computational chemistry, molecular modeling, and soft-matter systems, contributing to interdisciplinary research at the interface of physics, chemistry, and materials science.His international background is extensive, with research appointments across Europe and the United States. He held a Post-Doctoral Researcher position at Technische Universität Darmstadt, where he authored several publications on molecular dynamics, polymeric adsorbent media, and interfacial phenomena. He also served as a Visiting Researcher at institutions including the University of Amsterdam and the Danish Cancer Society, working on topics such as antibiotic–DNA interactions and biophysical modeling. His academic training includes a Ph.D. in Chemical Engineering from Missouri University of Science and Technology and a master’s degree from Politecnico di Torino, complemented by an early research period as a Visiting Scholar in France.Beyond his research positions, Riccardi has been actively engaged in academic leadership and policy. He served as a Board Member of SiN – the Norwegian association for PhD candidates and postdocs – and of DION at NTNU, advocating for early-career researchers. He also coordinated the EURODOC Work Group, strengthening ties within the European research community. These roles reflect his commitment not only to scientific advancement but also to improving research environments and governance in academia.Riccardi’s scientific profile spans computational chemistry, physical chemistry, machine learning applied to materials science, protein chemistry, polymers, and nanotechnology. His publication record covers molecular dynamics method development, soft-matter interphases, adsorption phenomena, and biomolecular transport in complex media.