What's the usage for solve_nn? And is that possible to add force information to the state?

[YunchuZhang]

As title.
I found in solve_nn it is trying to train a small network that maps states to actions.
What dose it used for?

And is that possible to add force information in the environment state?

Thanks