We welcome the new year 2026 with a fresh lineup of speakers for the Atomistic Modeling Seminar. Get ready for exciting insights into materials science and atomistic modeling! 

Dr. Ivor Lončarić is a Senior Research Associate at the Condensed Matter and Statistical Physics Lab at the Division of Theoretical Physics at the Ruđer Bošković Institute in Zagreb, Croatia. He works on first-principles materials modelling using density functional theory (DFT) and machine-learning approaches that extend quantum simulations to larger length and time scales. Currently, Lončarić leads a project developing machine-learned interatomic potentials for efficient modelling of molecular crystals and other complex materials. His research also involves studying dynamical and structural properties of materials through advanced first-principles methods in collaboration with the broader computational materials community.

Machine‑learned interatomic potentials make it possible to compute mechanical and thermal properties of molecular crystals with first‑principles accuracy but far greater efficiency, enabling both interpretation of experiments and high‑throughput materials screening. Because such models depend on large, high‑quality datasets, the talk “Modeling Molecular Crystals with Machine Learning Interatomic Potentials” will also address the current state of available databases and strategies for using them effectively.

Date: Tuesday, February 3, 2026, 3:30 pm

Location: MIBE Lecture Hall

Abstract:

Molecular crystals are a common and important class of crystalline materials. However, modelling molecular crystals based on first principles (eg. with density functional theory) is often difficult due to the size of a typical unit cell. Therefore, high-throughput calculations for the discovery of useful properties are rare. In this presentation, I will show how machine-learned interatomic potentials can enable accurate and fast calculations of mechanical and thermal properties of molecular crystals, enabling an understanding of experimental observations as well as a high-throughput search for materials with the desired properties [1,2,3,4]. Since machine learning models rely on a sufficiently large database, calculated with the desired accuracy, I will also discuss the status of databases and how to best exploit existing databases. 

[1] Materials & Design, 254, 114047 (2025) https://doi.org/10.1016/j.matdes.2025.114047
[2] Crystal Growth & Design 25 (19), 8196-8202 (2025) https://doi.org/10.1021/acs.cgd.5c01043
[3] Journal of Chemical Physics 160 (15) (2024) https://doi.org/10.1063/5.0196232
[4] Crystal Growth & Design 24 (20), 8167–8173 (2024) https://doi.org/10.1021/acs.cgd.4c00905