Currently a student in mathematics at the University of Canterbury in Christchurch New Zealand
The JuliaML ecosystem introduces an effective way to model natural phenomena with Universal Differential Equations. UDEs enrich differential equations combining an explicitly known term with a term learned from data via a Neural Network. Here, we explore what happens when our assumptions about the known term are wrong, making use of the rich interoperability of Julia. The insight we offer will be useful to the Julia community in better understanding strengths and possible shortcomings of UDEs.