“If you want a branch of mathematics, you develop it in the physical context, not as something separate, which you then try to apply, rather than integrating it from the beginning.”