Hi there! My name is Jochem, and I’m currently a master’s student in mathematics at the University of Utrecht During my bachelor’s physics and mathematics, I developed a keen interest in modeling and simulations. My interests primarily involves applying ML across various fields such as physics, chemistry, biology, and finance. On this site, I’ll be sharing insights and projects on multiple of these applications. If you have questions about my work, want to suggest a project, or are interested in collaborating, please feel free to contact me.
Update:
A portion of my bachelor’s thesis is now available as a pre-print on Arxiv. The paper explores the inverse design of crystals and quasicrystals in a non-additive binary mixture of hard disks. To locate the necessary state points and physical parameters where self-assembly for (quasi)crystals occurs, we employed an inverse design technique. Essentially, we solve a black box optimization (BBO) problem using an evolutionary strategy (CMA-ES).
In this paper, we utilize two different fitness functions:
- A CNN that assigns a score of how likely the diffraction pattern, originated from a certain phase.
- A symmetry-based order parameter that leverages the symmetry encoded within the diffraction pattern.
Both fitness functions are effective in identifying state points where self-assembly occurs. We discuss the pros and cons of both models in the paper. For more information take a look at my blog post
