I just pushed another blogpost in [what is turning out to be] a series about learning how to use #pymc3. This time: how to learn the off-diagonal elements of the mass matrix. Comments welcome!
https://dfm.io/posts/pymc3-mass-matrix/ …
-
-
It's a pretty simple idea that Goodman, Weare and I have been throwing around for a few years, but there are so many reasons that it's a bad idea that I'm not sure if it'll ever see the light of day :-)
-
but basically: if you have an ensemble of NUTS chains, you can estimate the mass matrix for each walker using the complementary ensemble. We can show that this satisfies detailed balance and works well in small numbers of dimensions…
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.