janosg / derivatives

Closed form derivatives for kernel transformations

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

To-DOs

  • Implement jacobians for the six functions in kernel_transformations.py
  • Test them againts jacobians calculated with jax. Check the autodiff cookbook to see how to take jacobians of functions. Jax differentiable versions of the functions are already implemented in kernel_transformations_jax.py. Draw many (at least 50) random inputs with different sizes to be sure that the functions do what they should.

About

Closed form derivatives for kernel transformations

License:MIT License


Languages

Language:Python 100.0%