Is jaxopt flexible enough to implement graphical lasso?

[Chengwei94]

Hi there,

I need to rewrite graphical lasso(https://en.wikipedia.org/wiki/Graphical_lasso) for one of my projects as the sklearn version is not flexible (does not take in a matrix of regularization parameter) and too slow for my liking. I was wondering if jaxopt is flexible enough to do a reimplementation of the graphical lasso. I am neither very familiar with jax nor the optimization behind graphical lasso so I would want to be sure that it is possible before diving into it. Thanks!