Understanding automatic differentiation (in Julia)

If you ever wondered how automatic differentiation (AD) works under the hood and what all the jargon means, this video will walk you through the main concepts in AD and how to use it effectively and optimise your code for AD. By the end of this video, you will understand what the following terminology mean: jvp, vjp, pushforward, pullback, tangent, cotangent, forward-mode AD, reverse-mode AD, mixed mode AD, frule and rrule. I will also cover examples of defining adjoint rules for 2 fundamental operations in any partial differentiation equation solver making use of pre-allocated memory, mutation and the implicit function theorem.