Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_deblender
-
Labels:None
-
Story Points:5
-
Epic Link:
-
Sprint:DRP S17-6
-
Team:Data Release Production
Description
Once DM-9644 is completed we need to create a test that calculates the amount of overlap between galaxies, basically
(Si Sj)2 / ((∑Si2) (∑Sj2))
for both the measured intensities and simulated intensities.
Attachments
Issue Links
Activity
This ticket has also included work to improve the API for the meas_deblender.sim module, which compares the simulated true values to the deblended measurements.
Fred Moolekamp
added a comment - This ticket has also included work to improve the API for the meas_deblender.sim module, which compares the simulated true values to the deblended measurements. Hi Peter, do you mind reviewing this ticket?
The main work can be seen in testSingleSim.ipynb.
In addition to implementing the test for overlapping galaxies (see the code here), I also updated the code to make it easier to compare different versions of the deblender and the simulated data using all of the tests we have developed so far.
Fred Moolekamp
added a comment - Hi Peter, do you mind reviewing this ticket?
The main work can be seen in testSingleSim.ipynb .
In addition to implementing the test for overlapping galaxies (see the code here ), I also updated the code to make it easier to compare different versions of the deblender and the simulated data using all of the tests we have developed so far. I fixed the equation in the description to match what is actually done.
Fred Moolekamp
added a comment - I fixed the equation in the description to match what is actually done. This nicely shows if the deblender works or does too little or too much.
Peter Melchior
added a comment - This nicely shows if the deblender works or does too little or too much. 
Changed to the current sprint, since this is an important test for the new deblender.