Jonathan Feinberg

I missed the part about how horrendously slow it was. I am suspecting it is the accumulation of evaluating `KDEMultivariateDependent` too often that is causing this. Scratch what I said...

@AnnaCraig, have had the time to test the new method?

I am not to experienced with Jensen-Shannon divergence in practice. But for most intents and purposes RMSE is usually sufficient in my experience.

Okay. Sad to hear that the z-axis is failing. Feel free to send me the files. jonathf[at]gmail.com. I definitely would want to have a look.

Okay, that is an interesting discussion. I think inclusion is the correct behavior as well. Also, I think this is an awesome addition to scipy, and I will quickly deprecate...

Exactly. Much cleaner to get the new feature with the switching of backends.

Thanks for the notice. I've actually followed the discussion for your PR. Interessting to observe the methodologies and the direction it is all taking. I will make an initial PR...

Chaospy doesn't really support unbound distributions. In most cases that doesn't matter as one can easily approximate unbound distribution by taking a large enough finite interval. For Normal I think...

Do you mean `Y = F(X1, .., X5)`, or `Y1 = F(X1), ..., Y5 = F(X5)`? A varying X wihtout varying Y does not make sense to me. Data driven...

Note that dd-pce is a big topic and depending on your reference, method might vary.