Add GPs #632
Add GPs #632
Conversation
Codecov Report
@@ Coverage Diff @@
## main #632 +/- ##
==========================================
- Coverage 86.69% 86.68% -0.01%
==========================================
Files 37 37
Lines 2209 2208 -1
==========================================
- Hits 1915 1914 -1
Misses 294 294 see 1 file with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
| @@ -0,0 +1,799 @@ | |||
| { | |||
There was a problem hiding this comment.
later \pmb{u} can be easily confused with \mu, maybe use a letter that stand out more clearly.
Reply via ReviewNB
There was a problem hiding this comment.
I'm still unsure about the change. I understand the possible confusion, but I don't want invent a new notation. \pmb{u} is usually used in mixed models literature and so I used it./ Other letter I can think of is \pmb{b}... I'll keep thinking about it
| @@ -0,0 +1,799 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,799 @@ | |||
| { | |||
There was a problem hiding this comment.
| @@ -0,0 +1,799 @@ | |||
| { | |||
There was a problem hiding this comment.
It used to be the case that az.style.use("arviz-darkgrid") (the same for any matplotlib style) has to be defined in a different cell than the one used to import matplotlib. Otherwise only part of the style was actually applied. Not sure if this is still the case
Reply via ReviewNB
There was a problem hiding this comment.
I wasn't aware of this. Maybe related to this is that sometimes I had to use %matplotlib inline if I wanted styles to apply. Weird Jupyter magic.
But I think it's not the case anymore (at least here)
| @@ -0,0 +1,799 @@ | |||
| { | |||
There was a problem hiding this comment.
|
View / edit / reply to this conversation on ReviewNB aloctavodia commented on 2023-03-18T20:01:06Z This article demonstrates how to use Bambi
Can we link to Bill's video from PyMCON (I think is still not available...) tomicapretto commented on 2023-03-20T19:19:25Z It's still unavailable. I'll add it once we have it. |
|
View / edit / reply to this conversation on ReviewNB aloctavodia commented on 2023-03-18T20:01:07Z The default sampler is too slow? Can we use it? Just to simplify the notebook. tomicapretto commented on 2023-03-20T19:21:41Z Yeah, the default sampler is way slower. Not in all cases, but to give an example, it would take one or two minutes when |
|
View / edit / reply to this conversation on ReviewNB aloctavodia commented on 2023-03-18T20:01:08Z The concept of "basis vectors" has not been introduced, given the emphasis in this section at least a couple of sentences explaining their role should be added.
|
|
It's still unavailable. I'll add it once we have it. View entire conversation on ReviewNB |
|
Yeah, the default sampler is way slower. Not in all cases, but to give an example, it would take one or two minutes when View entire conversation on ReviewNB |
This is a work in progress and I expect this PR to be open for a considerable amount of time while I work on it.Alright, I think this is done. We now have GPs through the HSGP approximation. For more information see the notebooks in
docs/notebooks/hsgp_1d.ipynbanddocs/notebooks/hsgp_2d.ipynbas well as the tests.Before merging, we need pymc-devs/pymc#6458 to be merged in PyMC.
And before releasing, we need a new PyMC release with HSGP.