Details
-
Type:
Improvement
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: ts_aos
-
Labels:
-
Story Points:5
-
Team:Telescope and Site
Description
Currently all donut stamp images passed to WEP are transposed before the Algorithm is called https://github.com/lsst-ts/ts_wep/blob/42c96f0ee9b2907e17625dee348e735f8b60b4e7/python/lsst/ts/wep/task/CalcZernikesTask.py#L235-L245
This is part of the most recent workflow, which includes
- donut detection and selection with GenerateDonutDirectDetectTask or GenerateDonutCatalogWcsTask
- cutting out postage stamp images with CutOutDonutsScienceSensorTask
- wavefront estimation with calcZernikesTask
The transpose happens inside calcZernikesTask :
wfEsti.setImg(fieldXYExtra,DefocalType.Extra, image=rotate(donutExtra.stamp_im.getImage().getArray(), eulerZExtra).T, blendOffsets=blendOffsetsExtra.tolist(), )
|
wfEsti.setImg(fieldXYIntra, DefocalType.Intra, image=rotate(donutIntra.stamp_im.getImage().getArray(), eulerZIntra).T, blendOffsets=blendOffsetsIntra.tolist(),
|
We find upon comparison to simulated data with batoid that this transpose for auxTel images is not necessary. Attached image compares the fit to a simulated batoid auxTel image where 300 nm of z8 was injected with phase screen. Both `cwfs` (old wavefront estimation code) and `wep` (current code) retrieve the wavefront correctly when provided with untransposed images. 
Attachments
Activity
A result of switching the new config `calcZernikesConfig.transposeImages` from `True` to `False`:
Merged on github
Reviewed and merged in GitHub.
Notebook comparing the fitting of batoid simulations with WEP as a task, WEP directly interacting with the algorithm, CWFS code : https://github.com/suberlak/AOS/blob/main/AOS_DM-37903_auxTel_batoid_transpose_summary.ipynb Based on zk5-10 we find that there should be no transpose done to intra/extra focal images in `CalcZernikesTask`, and the results are identical between WEP as a task and WEP run directly.