Natsuhiko Yoshinaga is a theoretical physicist working in active soft materials and machine learning. He started his career by the theoretical and numerical analysis of the dynamics of a polymer chain during his PhD at Kyoto University. This study is inspired by biological macromolecules. Then, he developed a theoretical model for the actin cytoskeleton, particularly focusing on stress fibres, the counterpart of muscle in a cell at Curie Institute in Paris. He studied the framework of active hydrodynamics, which is a natural extension of the theory of liquid crystals to nonequilibrium systems. We also developed the theory of thermophoresis and self-thermophoresis, and collaborated with experimental groups. At this stage, his focus was on self-propelled particles and drops which consume energy locally to realise their motion. These systems are nowadays called active matters.

After joining Tohoku University, he had more opportunities to discuss with materials scientists. He studied theoretical models of soft materials, such as the motility assay of kinesin filaments, wave patterns of Min proteins, and dynamics of drops and cell motility. Besides those studies, we extended his area and developed theoretical models for spin dynamics. At MathAM-OIL, he started to combine his theoretical and computational knowledge of soft materials, nonequilibrium statistical mechanics, nonlinear dynamics, and fluid dynamics with machine learning. He developed a method to estimate physical models from data or desired structures. Now, he is involved in several projects that unify physical modelling, machine learning, and materials science.