Date: 3 March 2026
Time: 15:00 CET
Abstract
This webinar gives an overview of the new features and performance improvements that are included in the recently released 2026.0 version of GROMACS. Performance improvements include a complete HIP backend for AMD GPUs and non-bonded free-energy kernels for CUDA GPUs. Particular attention is devoted to new features for neural-network potentials and the inclusion and validation of the newest AMBER force fields.
Presenters
Berk Hess
Berk Hess is a professor of Theoretical Biophysics at KTH. He has designed algorithms for the GROMACS simulation package for over two decades. His current research focuses on advanced sampling methods, aggregation of molecules and studying wetting of surfaces at the molecular scale.
Lukas Müllender
Lukas completed his Bachelors and Masters degree in physics with a specialisation in condensed matter theory at RWTH Aachen University. For his Master’s thesis he worked on developing machine-learning collective variables for enhanced sampling from massively parallel path sampling simulations under the supervision of Prof. Paolo Carloni at Forschungszentrum Jülich. Since 2023, he is a PhD student within the AQTIVATE MSCA doctoral programme with Prof. Erik Lindahl at KTH Royal Institute of Technology, working on the implementation and application of neural network potentials in GROMACS.
Vedran Miletic
Vedran Miletic is an HPC application expert at MPCDF specializing in GROMACS and other biomolecular simulation software. With more than 15 years of experience in scientific software development and supercomputing, he has contributed code to numerous high-profile open-source projects across both application and compiler/toolchain domains. He has also co-authored several publications in in the areas of computational biochemistry and computational biophysics.
