https://pfspipe.ipmu.jp/jira This file is an XML representation of an issue en-us 8.3.4 803005 13-09-2019 https://pfspipe.ipmu.jp/jira/browse/PIPE2D-410 DRP 2-D Pipeline <p>This ticket is from the series ``Improvements to the PSF 2d modeling, April 2019 brainstorming''.</p> <p>The illumination of the pupil, especially the wide wings has been implemented poorly. At the moment, the illumination is combination of two components: 1. Just top hat with FRD Gaussian broadening and 2. ``Lorentzian'' component that should mimic the wide wings that we see in theta. The actual resulting illumination is then a linear combination of these two components. For vast majority of images the algorithm end in extreme in which there is only ``Lorentzian'' component. In addition, residuals of the defocused images often show rings showing that the illumination of the pupil was not able to capture the wings seen in the data.</p> <p>In short, modeling the illumination of the pupil should be improved.</p> PIPE2D-410

Illumination of the pupil should be improved

Task Normal Done Done ncaplar ncaplar Wed, 17 Apr 2019 13:51:13 +0000 Thu, 27 Jan 2022 14:42:31 +0000 Thu, 27 Jan 2022 14:42:31 +0000 0 2 <p>I hope that this will done, at least partially, by Brent Belland, once when is able to run my code.</p> <p>I am starting to run the code now. I'll follow with comparing some of my lab data with simulated data.</p> <p>I have analyzed the code and tested it against Cobra-mounted fiber data that I have collected previously; this analysis runs well and the FRD components fit the lab data well, as to be expected. I will continue to work on implementing methods that describe my FRD and misalignment well with minimal time impact. <span class="image-wrap" style=""><img src="https://pfspipe.ipmu.jp/jira/secure/attachment/12115/12115_SimFRD_020.png" style="border: 0px solid black" /></span></p> <p><a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=bbelland" class="user-hover" rel="bbelland">Brent Belland</a> Could you describe a bit how much playing around with the parameters which describe the illumination of the pupil in my code did you do to create this residual image? A lot? None at all?</p> <p>Relatively little parameter adjustment for this image, beyond the removal of obscurations as they aren't present in this data. I evaluated over a range of frd_sigma, but within the expected range. I did vary the lorentzian component, but beyond the correlation between lower frd_sigma requiring larger lorentzian component to describe the drop-off, I haven't found a good way to optimize fitting between the two components yet.</p> <p>Ok, lets talk a bit more tomorrow, either in the morning or when you come visit.</p> <p>This took me longer than I expected due to some of my scaling code. Running the code against a well-aligned data makes the jagged-tooth like residual mostly disappear (small dips can be seen due to pupil and background subtraction). The offset from 0 at the center has to do with my automated choice of 1.0 for the lab data combined with the curvature seen in the gaussian convolution approximation in the 2D-DRP code.  <span class="image-wrap" style=""><img src="https://pfspipe.ipmu.jp/jira/secure/attachment/12118/12118_Angfit_025.png" style="border: 0px solid black" /></span></p> <p><a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=bbelland" class="user-hover" rel="bbelland">Brent Belland</a> Brent I am only looking at this now, sorry for not looking at it immediately. Is this the same lab data as you were showing above? I.e., dd you only change alignment or it this also different data? </p> <p>I apologize for not making this clear - this is data from my alignment tests, with minimal angular misalignment. This is different data from the previous images, where I suspect angular misalignment drove some of the jagged edge residuals in the data.</p> <p>I wanted to follow up this old post with one more update, to cover a question that came up regarding if this fit was good for high FRD data rather than just from the alignment tests. I've uploaded an example figure HighFRD_test that demonstrates that, to within a few percent, the 2D-DRP is reproducing the wider profile drop-offs that I've seen in stressed, non-misaligned fibers. This particular ticket can be closed.</p> Relates PIPE2D-453 PIPE2D-433 Development Rank 0|02qpt6:508yli Story Points 6.0