3 October 2022 at 8:34 am #612800
It did went into bright outburst mode. I’ve been only able to collect 58 measurements this night due to clouds.
Attachments:8 October 2022 at 11:09 am #612950
Strange things going on… The timing of minima keeps sliding. I’ve plotted the minimum a few days ago and it was right on the observed minimum. Tonight it is about 0.2 phase off. There is very little data, as it was shot through the hole in the clouds, but the eclipse was captured. Either this is a bug in my code or the period keeps changing. But the code here is rather straightforward – I get the epoch time and simply add periods to it to plot lines. How can a period drift by 0.2 phase in such a short period of time?
if self.plot_epoch > 0 and self.plot_period > 0:
extrema = 
current_extrema = self.plot_epoch
light_curve_start = light_curve[“DATE”].min()
light_curve_end = light_curve[“DATE”].max()
while current_extrema < light_curve_end:
if light_curve_end > current_extrema > light_curve_start:
print(“Ephemeris extrema:”, current_extrema)
ax0.axvline(current_extrema – 2400000.50, color=”green”, linestyle=’–‘)
current_extrema = current_extrema + self.plot_period
- This reply was modified 4 months ago by Maxim Usatov.
Attachments:8 October 2022 at 1:34 pm #612953
Hmmm…that is very odd Max. Could it be the recent dip is a random fluctuation rather than an eclipse?8 October 2022 at 3:59 pm #612954
I don’t think so. Hopefully more data will follow.
Max14 October 2022 at 10:45 am #613045
Eclipses No(s): 94
The eclipse minima definitely has shifted or there is an error in how I am calculating minima. I am puzzled. Can O-C analysis be done in Peranso or some other software?
Attachments:14 October 2022 at 12:48 pm #613050
Period analysis can be performed using one of Peranso spectrum analysis tools. But these analyse the data globally, resulting in significant periods being identified, rather than calculating the ToMin that would be required for O-C analysis.14 October 2022 at 4:50 pm #613054
Just double-checked my minima calculation: MetroPSF provides the same ephemeris as Peranso, so looks like there is something going on with periods here. I just wonder if there’s an automated way to identify observed minima of all ~ 90 eclipses we have, or is manual the only way? Anyone brave enough to do O-C analysis? 🙂
Max15 October 2022 at 10:39 am #613057
Eclipses No(s): 95
Typical U/N/A eclipse for this state.
Attachments:16 October 2022 at 12:18 am #613059
Thinking how to automate ToM detection for the data collected so far, I have tried fitting a Gaussian function and finding its minimum – see screenshots attached for the last eclipse recorded. Unfortunately, this method is very sensitive to the data boundaries if the data is noisy. If I include more points to the left or right, I get a different curve fit. Even the Kwee-van Woerden method displays the same sensitivity in Peranso. Attached are a few examples. I was thinking to try to find ToM automatically using some predefined phase shift, say, delta phi = 0.2, from calculated minima to set search boundaries, but the choice of delta phi produces different ToM results with both Gaussian fits and the KvW.
Attachments:16 October 2022 at 2:22 pm #613064
OK, I have managed to do Gaussian fits. First, I define a delta phi search region, then identify the faintest magnitude within the search region, smooth out the noise, determine eclipse well and then fit a Gaussian curve to the well. Presenting the first O-C table. Looks like it’s linear. ANOVA period determined incorrectly?
Attachments:16 October 2022 at 5:58 pm #613067
Eclipses No(s): 96
Attachments:20 October 2022 at 8:29 am #613200
Eclipses No(s): 97
Attachments:24 October 2022 at 11:26 pm #613249
Flat O-C (ignore a few outliers) with P=0.188644 d.
Attachments:25 October 2022 at 5:32 pm #613270
Eclipses No(s): 98
Attachments:25 October 2022 at 7:05 pm #613273
Looking good Max. Any ideas about outliers? I’ve sometimes struggled to assign proper ToMs in specific eclipses, especially small ones, when the underlying light curve has been busy with bumps and humps.
26 October 2022 at 9:57 am #613279
- This reply was modified 3 months, 1 week ago by Jeremy Shears.
These are exactly the cases where the algorithm has failed to fit a Gaussian due to noise or partial data. I also need to resample the data to a finer grid, as described by KvW, to refine the result.26 October 2022 at 11:51 pm #613284
Thanks Max. I suspected this was the case. But ignoring these, the O-C is looking good 👍🏻28 October 2022 at 10:29 am #613323
Eclipses No(s): 100 (didn’t show a pic for 99th, partial data)
Attachments:28 October 2022 at 11:48 am #613325
Congratulations on the century, Max! Very nice work, indeed.28 October 2022 at 2:54 pm #613326
Thank you very much for organizing this, Jeremy!
- You must be logged in to reply to this topic.