Affiliation
American Association of Variable Star Observers (AAVSO)
Sat, 05/16/2020 - 14:08
I attached a transcript of an e-mail exchange I had with Diffraction Limited suggesting a “time dithering” feature for maxim DL. Maxim has a dithering feature for telescope position which is helpful when making sky flats. However, randomizing time delays between image sequence sets would be a big help in differentiating signal from aliases when performing DCDFT and similar analysis.
If this is something that you think would be helpful for light curve analysis, It might be a good Idea for Diffraction limited to get endorsing responses to the post in their support forum. If a sufficient number of people express interest in having this new feature, its priority will most likely be elevated.
Brad Walter
File Upload
Interesting idea. I think the request would gain by being numerically quite specific, and might also gain by having some example data series and reductions with and without time dithering to show the technical benefit experimentally.
One note: Observers whose photometric data reduction is automated may much prefer dithered exposure times over dithered intervals. For us, reasonably non-uniform exposure times cause no problem, and with the advantage of eliminating waiting time between exposures. (I'm thinking of duty cycle here.)
Cheers
As to randomizing exposure times; I think I have encountered a problem with that, that I avoid by making all the exposures the same. Have not nailed down why, but I dimly recall that VPHOT doesn't seem be happy to mix 40 second and 60 second exposures, for example. Also, there must be some limit as to far far away you can be from calibration frame exposure durations before calibration errors dominate. MaxIm does have a wait between exposures that could be randomized.
Another way to two-step-time-jitter a bit and do some Low Pass Filtering at the same time is to overlap-average images.
For any scheme, there must be a mathematical proof as to what randomization roots work best for a given exposure. Maybe it depends on the ratio of signal frequency to exposure duration.
Ray
Maybe following paper (and references in it) can answer to that question:
http://articles.adsabs.harvard.edu/pdf/2009BaltA..18...83P
Best wishes,
Tõnis
For any scheme, there must be a mathematical proof as to what randomization roots work best for a given exposure
Optimization seems like too much. One need not prove that a proposed randomization scheme is optimal, one needs only demonstrate that it's better than uniform cadence, and enough better to justify the complication.