Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

update: chromatic noise models & feat: hypermodel empirical distributions #242

Open
wants to merge 168 commits into
base: master
Choose a base branch
from

Conversation

jeremy-baier
Copy link

@jeremy-baier jeremy-baier commented Sep 8, 2024

This PR contains many various updates associated with chromatic noise modeling efforts over the past several years.
Notable changes include:

  • functionality for “fixed-point” analysis where one can set a chromatic signal to a max likelihood value rather than sampling in that GPs hyperparameters.
  • the ability to sample in the chromatic index of a chromatic model.
  • it also expands functionality for empirical distributions to include a hypermodel empirical distribution.
  • cleans up the hypermodel code caching stuff which is called in each iteration unnecessarily
  • move the hypermodel weights to the lnPRIOR rather than the lnLIKELIHOOD (fixes issues with PTMCMC and hm weights)
  • prints hypermodel weights to a file in out directory (model_log_weights.json)
  • adds support for the SW GP basis & kernel from Nitu et al 2024

@Hazboun6
Copy link
Member

@AaronDJohnson do you recognize the errors that are happening in the mac installs? It's a libstempo issue, so it doesn't even get to the actual unit tests.

@blarsen10
Copy link

This also now adds support for the SW GP basis & kernel from Nitu et al 2024

@Hazboun6
Copy link
Member

@blarsen10 and @jeremy-baier Can we have a look at the errors here ASAP. From what I saw it's just not passing the linting rules. It would be good to get this merged this week.

@Hazboun6
Copy link
Member

@jeremy-baier and @blarsen10 I think that the reason the tests are still failing is that GitHub CI uses the setup.py to setup the environment for the tests. That doesn't limit the scipy version yet.

Hazboun6 and others added 2 commits September 19, 2024 21:50
@AaronDJohnson
Copy link
Collaborator

@AaronDJohnson do you recognize the errors that are happening in the mac installs? It's a libstempo issue, so it doesn't even get to the actual unit tests.

Sorry for the delay on all of this. Feel free to ping me again if you don't get a reply quickly next time! I did know exactly what was happening, and I've merged a fix. If @jeremy-baier pulls from upstream/main, the Mac tests should be fixed now. The new runners only support ARM architectures, so we have to go back one version on the CI runners so that they run an x86 architecture instead.

@AaronDJohnson
Copy link
Collaborator

I've also fixed the scipy issue by updating to the new recommended function. So you can remove the maximum version number on scipy here.

@jeremy-baier jeremy-baier changed the title Chromatic noise updates update: chromatic noise models & feat: hypermodel empirical distributions Jan 10, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants