For best experience please turn on javascript and use a modern browser!
You are using a browser that is no longer supported by Microsoft. Please upgrade your browser. The site may not present itself correctly if you continue browsing.
The Spotlight introduces a different Data Science Centre Affiliate Member every month. This month: Jacobus Dijkman, PhD Candidate at the Computational Chemistry Group of the Van't Hoff Institute for Molecular Sciences and the Amsterdam Machine Learning Lab at the Informatics Institute. Jacobus' PhD is part of the Interdisciplinary PhD program from the Data Science Centre.

How do you apply data science to your projects?

'I bridge the gap between speed and accuracy in molecular science using machine learning. Traditionally, researchers rely on computationally expensive simulations to analyze material properties for applications like carbon capture. While faster analytical models exist, their accuracy is often insufficient.

My PhD tackles this challenge by integrating machine learning with thermodynamical equations to describe molecular behavior, aiming to achieve the ideal balance of speed and precision.'

Is there a project from this past year that you are most proud of? 

'This past year, we made a significant breakthrough by being among the first to successfully integrate neural networks into a thermodynamic framework for describing molecular dynamics. This paves the way for a powerful and fast alternative to traditional molecular simulations. With our proof-of-concept, we now have a clear path to make it practically applicable, which is very exciting.'

What do you like most about being a DSC member?

'I particularly value the connections with other interdisciplinary PhD students applying machine learning in various fields. I enjoy starting conversations with people working in different disciplines and discovering that the challenges they face often boil down to familiar optimization problems.'

What is your favourite data science method?

'While massive transformer models are currently in the spotlight, I like using smaller, simpler networks that can model complex phenomena through creative training objectives.'

Are you camp Python/R/or something else?

'Python is my go-to language for machine learning due to its extensive libraries and user-friendliness. However, when dealing with molecular simulation algorithms, I rely on C for its raw speed and performance.'

Contact Jacobus Dijkman about his work

J.J.L. (Jacobus) Dijkman MSc

Faculty of Science

Van 't Hoff Institute for Molecular Sciences