Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: pipe_analysis
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Sprint:DRP S21a (Dec Jan)
-
Team:Data Release Production
-
Urgent?:No
Description
The scripts in pipe_analysis are currently failing, apparently due to an API change in treecorr's GGCorrelation object:
AttributeError: 'GGCorrelation' object has no attribute 'varxi' |
Update the scripts to the new API, which now has varxip and varxim attributes, which, according to the documentation represent the variances of xip (the real part of xi+) and xim (the real part of xi-), respectively.
Attachments
Activity
Hasn't seemed to show up yet...PR is https://github.com/lsst-dm/pipe_analysis/pull/70
In the above plots, for the points where the marker is filled, the errorbars were definitely correct prior to the review. So the variation in their sizes comes from changes in processing. It is hard to have good intuition on uncertainy when the yscale is log unfortunately.
Indeed! Thanks for the review. Merged and done.
Would you mind giving this a look? I took a guess as to which of the new variance parameters should be used now that the one that was being used no longer exists. I see some differences between my current run:
and that of RC2 w_2021_02
But perhaps that is to be expected (and a desired/more correct result)?
I also took the opportunity to fix up some Matplotlib deprecation warnings in a separate commit.