Is there a software package that can automatically determine minima/maxima of light curves (e.g. as imported from AAVSO or TESS) and streamline the creation of O-C diagrams?
2 Likes
There are two ways to take this question.
If you are asking for software that will just take the entire data string and find all times of max or min automatically, then I believe I have seen publications where people have written their own code to do this. One of my students did something like this by using a master curve to slide along the data string, but it didnât prove as useful as we had hoped. I donât believe there is a generally available software to do this. While it can be done, I have concerns for software that can be totally a black box with little input. How much of the curve to you need for it to fit the maxima? I have had many conversations with a fellow astronomer about the wider errors we see in modern O-C diagrams for well established stars, which might be traced to automation without checking.
If you are asking for software where you can mark the each maxima (or min) and have the program accurately calculate the time for that peak, then you would likely want to use Peranso.
1 Like
I was hoping for a solution that can process an entire data set (e.g. AAVSO data across multiple years) but as you said it is not as straightforward to implement as one might have assumed. I ended up fitting a polynomial for each peak and combined the findings with data from several papers.
I have some Python codes that partially streamline a similar process, in the context of identifying eclipse timing variations by fitting individual eclipse to a cosh Gaussian (akin to generalized Gaussian) curve.
It is not a black box: you need to decide how much of the eclipse / curve to fit; determine the period, epoch of a minimum, and the parameters for the Gaussian shape. It has some semi-interactive tools to help you refine the parameters.
Once the parameters are set, it then fits individual eclipses to identify the time of the observed minima.
Sam,
How does your procedure deal with total eclipses (i.e.) those eclipses where the light curve has a âflatâ segment at mid eclipse. This happens not infrequently with EW eclipsing binaries. In addition, the âflatâ segment is often not truly horizontal.
Roy
Roy,
Assuming youâre referring to the eclipses with flat segments at mid eclipse found in some EA (not EW) type eclipsing binaries:
For the purpose of identifying eclipse midpoints / minima, the procedure should work reasonably well. I am indeed a bit uncertain about U-shaped / flat bottom eclipses, however. [*]
Fundamentally, the procedure fits an eclipse with a cosh Gaussian curve, which can cover eclipses of a wide variety of shapes (examples at the end). It probably would not work well for eclipses with asymmetric shape, however.
I am unclear about your concern that âflatâ segments are not truly horizontal. As long as itâs symmetric, it shouldnât matter much for the purpose of identifying minima (as opposed to, say, accurate modeling of the eclipses).
[*] On fitting flat bottom eclipses, I am a bit uneasy about those cases. From what I could see, the MCMC-based procedure often does not settle well (converge) for the Tau parameter (primarily affects eclipse shape) in those cases.
I donât see anything clearly wrong in the fitted results per se. Iâm just a bit uneasy, and havenât got to research or test to see if there is any real issue.
Sam
Some examples:
