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

Pruning modifiers per sample and channel #1279

Open
alexander-held opened this issue Jan 28, 2021 · 0 comments
Open

Pruning modifiers per sample and channel #1279

alexander-held opened this issue Jan 28, 2021 · 0 comments
Labels
feat/enhancement New feature or request user request Request coming form a pyhf user

Comments

@alexander-held
Copy link
Member

Description

Modifier pruning currently removes a modifier completely from a model. It would be convenient to be able to remove modifier effects per sample or channel via the pruning API. This can be useful for example when a systematic has very different effects across phase space or between samples, and can reduce the amount of interpolations/extrapolations that need to be calculated.

Pruning is useful to speed up fits without significant impact on the results (if done correctly). See these slides (ATLAS internal) for a possible approach (@kratsg).

Is your feature request related to a problem? Please describe.

pruning currently is all-or-nothing, for more customization an external implementation is needed

Describe the solution you'd like

per-sample and per-channel modifier pruning

Describe alternatives you've considered

external implementation

Relevant Issues and Pull Requests

nothing immediately relevant, but some related issues:

Additional context

none

@matthewfeickert matthewfeickert added feat/enhancement New feature or request user request Request coming form a pyhf user labels Jan 29, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feat/enhancement New feature or request user request Request coming form a pyhf user
Projects
None yet
Development

No branches or pull requests

2 participants