Smit Lunagariya
Smit Lunagariya
> I think it will be challenging because there is both linear interpolation and root finding. Yeah, I see. I was trying to write the JAX code but that fails...
This should now be possible with `overload` decorator.
I would like to work on it. Also, I think this could also be added : [https://en.wikipedia.org/wiki/Cross_entropy](https://en.wikipedia.org/wiki/Cross_entropy)
Referring to the example: [https://en.wikipedia.org/wiki/Kullback%E2%80%93Leibler_divergence#Basic_example](https://en.wikipedia.org/wiki/Kullback%E2%80%93Leibler_divergence#Basic_example) I was able to implement in sympy as following: ``` >>> from sympy.stats import * >>> from sympy import Piecewise, log, Sum, Eq >>> from...
So for KL Divergence, we can design an API as: 1. Create a function `KL_divergence` in `rv_interface.py`. 2. Check whether both rvs have the same` domain.set`, if not `raise` error....
The arguments will be two random variables. Sounds good?
Yes, they must be shifted, also, I think that the users are generally used to RVs, so could we provide such flexibility of able to use either RVs or Distributions...
@czgdp1807 Any comments on [this](https://github.com/sympy/sympy/issues/18718#issuecomment-597671337)?
ping @czgdp1807
> For the above to be possible, distribution classes should be shifted to `sympy.stats` namespace. It might be possible that if we move all the distributions to `sympy.stats` namespace, the...