Postdoctoral Researcher: How do Physical Learning Systems Learn

Postdoctoral Researcher: How do Physical Learning Systems Learn

Institute AMOLF

Amsterdam, Netherlands

From € 4.552 per month gross

We are seeking an excellent and motivated postdoctoral researcher to join our team at AMOLF, working on fundamental questions on physical self-learning systems as part of the NWO ENW‑M1 project “How do physical learning systems learn?”. The research position is intended to start in September 2026.

Physical learning is an emerging paradigm in which materials adapt their behavior through local physical rules, without digital computation. Despite rapid experimental progress, it remains poorly understood how such systems learn and what signatures learning leaves in their physical structure and energy landscape. This project aims to build the theoretical foundations of physical learning, uncovering the modes of learning available to linear and nonlinear systems, their expressiveness and capacity, and the physical imprints of learned tasks.

The postdoctoral researcher will contribute to developing this theoretical framework, with a strong focus on analytical modeling, computational methods, and the interpretation of learning signals embedded in physical structures. Recent advances in our group, including new methods for detecting learning signals in linear networks that reveal aspects of the tasks they have learned, provide a powerful conceptual starting point.

The scope of possible topics includes

• Developing theoretical tools to characterize learning modes in linear and nonlinear physical networks;
• Understanding how learning reshapes physical energy landscapes;
• Identifying physical signatures of learned tasks;
• Exploring expressiveness, capacity, and continual learning in physical systems.

This position is theoretical and computational in nature, with opportunities for collaboration with experimental groups working on physical learning in electronics, mechanics, and living flow networks (Physarum Polycephalum).

Qualifications

We seek candidates with:

• A PhD in physics, applied mathematics, materials science, mechanical engineering, computer science, or a related field;
• Strong interest in learning, adaptation, and dynamical systems in physical contexts;
• Experience with analytical andor computational modeling;
• Proficiency in numerical methods and coding (Python, JAX, MATLAB, or related tools);
• Good communication skills in English;
• Experience with complex systems, energy landscapes, physical memory, machine learning, or soft/active matter is advantageous but not required;
• We welcome applicants from diverse backgrounds and strongly encourage curiosity-driven thinkers.