Seth Axen

Ah, I had seen it but it wasn't clear that it was more than just a mutable replacement for `DimStack`. Very cool, I see also that it can store arbitrary...

> @sethaxen the conflicts against breaking are pretty extreme, sorry about that. > > But we should try getting it merged there for the next breaking release (which is very...

> Oh right I misunderstood. So yes its fine to merge to main. > > But: then there will be the other problem that I cant easily do the merge...

While the AdvancedVI interface seems now complete, the main issue now is that AdvancedVI expects us to be able to have a `step` function, which would here correspond to a...

Code for efficiently sampling a `QRPackedQ` can be found at https://gist.github.com/sethaxen/4d5a6d22e56794a90ece6224b8100f08. It also supports complex matrices, and for square matrices, it also supports constraining the determinant to be 1. `qr(::Matrix)`...

> Note that in order to get a uniform distribution you need to eliminate the sign ambiguity of the QR decomposition. Yes this is true and is handled by the...

> That's fascinating, I thought it wasn't possible to do compute a numerically stable reflector with non-negative β because of the division by α - β. Yeah figuring out how...

I honestly don't remember; it's been a long time. Perhaps I was trying to increase the value of `safmin` slightly to decrease the chances of `invsafmin` overflowing, but then I...

Here are some points for discussion. There are some distributions (e.g. `LKJ`) for which there is no parameter array. To support custom axes, we'd need to add an `axes` field...

> This seems nice but not as important as #1905, so I think we should get #1905 in first. Makes sense. What is still missing in #1905?