Abstract
In this talk I will discuss some of the difficulties in trying to develop theories for stellar and planetary magnetic field generation and for magnetically confined plasmas. Some of the techniques developed are very generally applicable, including theory, computational methods and machine learning.