I have been analyzing a candidate EA variable star and have had some trouble in determining its period. I have identified the star from ROTSE 3 light curves and classified it as an EA variable using data obtained from NSVS and SuperWASP. Previously, I have had trouble in fitting ROTSE 3 data with NSVS and SuperWASP, but I no longer think I have that issue.
My question is this: how do I go about determining the period of the EA star within precision limits of what VSX requires of its submissions? I am using an excel spreadsheet form of Eplot, which requires that I input a period into the spreadsheet to generate a phase plot. To even obtain this period, I had to "hand scan" light curves available in the SuperWASP archive to determine the time between two consecutive, identical points on the available light curve. My initial estimate was about 2.12-2.13 days, which I refined to 2.126-2.127 days, as this period produces, in my opinion, the clearest pattern for the phase plot (which is attached to this post). Is there a more objective way of determining the star's period? Perhaps some software I can use to accommodate the large amount of data (well over 10000 observations)? I had used IDL when I was just phasing NSVS and ROTSE 3 data individually, but SuperWASP accounts for the the majority of all observations, so I had to use Eplot to handle the amount of information.
Also, for submitting my phase plot to VSX, do I eliminate outlier observations before submitting? And is there an objective way to determine what counts as an outlier?
Thanks for your help,
Thank you for your prompt response; I took your advice and used VStar to generate a more precise period. It worked quite well, and I have attached the results.
I just have two last questions:
1. For an EA star's phase plot, the minimum brightness (the bottom of the primary eclipse) has to be at phase 0 on a phase plot. In order for this to happen, there has to be a certain epoch which can satisfy this value, correct? Can this value be the HJD date for any observation which is chosen to be the minimum of the star, or are there any restrictions on this?
2. Finally, for observations which deviate noticeably from other nearby observations, should these be counted as outliers and be removed from the plot?
Thank you very much,
SuperWASP data has that problem: some nights have different zero point than others. Given that you have so many datapoints, remove those nights.
Also, it is useful to set a limit for the error figures. E.g. deleting all points with error figures larger than 0.03 or so. It will depend on the star's magnitude, for a 9th mag. star you can delete all with errors >0.01 or 0.02 for large datasets.
This will remove much of the outliers in the light curve.
Don't forget to combine your data with NSVS data since you mentioned you were also analyzing those. This will improve the period a lot because of the increase in the time baseline.
And if you are combining data, since the above surveys are unfiltered, downloading ASAS-SN data will give you a standard V zero point that you can use to give a range (use the V data only, not g, they are different filters and both are listed in the results).
The minimum in your phase plot is displaced to the left, phase 0 should be exactly at the midle of the primary eclipse and that should be the epoch to be given in the table.
Do not forget to add all references and star identifiers in the submission.
To give multiple references you have to save the VSX submission form as "draft only" and then reopen it.
PS: I also use Excel for my plots, you can customize it as you want (especially for multiple datasets) and also adjust the period and epoch as needed (you can zoom in on the primary eclipse for that).
I have incorporated all the data from ROTSE 3, NSVS, SuperWASP, and ASAS-SN in my phase plot. I took your suggestion about the error measurements and removed all observations with errors greater than 0.03 from ASAS-SN, ROTSE 3, and NSVS observations and all observations with errors greater than 0.02 from SuperWASP observations. One thing I noticed was that the ASAS-SN data seemed to be shifted upwards in terms of magnitude from the SuperWASP, NSVS, and ROTSE3 data by approximately magnitude 0.3 (the baseline for the ASAS-SN data was lower (around 10.8) than the baseline for the SWASP, etc data (11.1)). Should I correct for this discrepency by adding 0.3 to the ASAS-SN data so it can match the SWASP data, which comprises the bulk of the total observations? I have done so and attached the resulting phase plot generated by gnuplot here.
Lastly, what exactly do you mean for the V filter data for ASAS-SN? Does the V-filter correspond to the passband section on the submission draft? If so, I have been instructed to use the R-1 passband (ROTSE-1).
Thank you very much for all your help,
You have to do the opposite. V is a standard passband, R1 are just unfiltered magnitudes calibrated with V comp stars. The same for SuperWASP. You have to shift each dataset independently to the ASAS-SN V zero point and yes, then you should select V as the passband.
And you have to adjust the period, the eclipses from each survey recorded over different years do not match. That's the good thing about combining datasets and increasing the time baseline: you can refine the period a lot.
You can't have datapoints from one survey at minimum at the same phase when there are points at maximum from the others.
All descending and ascending branches have to be adjusted using the best possible period.
A final remark, usually unfiltered magnitudes are brighter than V, the fact that V is 0.3 mag. brighter (at least that's what you said) seems to indicate this is a very blue star.
June 11 2019
I have shifted the magnitudes of the NSVS and SuperWASP sets to match the magnitude range from ASAS-SN V-filter (ROTSE 3 remains unchanged). I have also modified the period very slightly (from 2.126394 days to 2.12642 days). Combined, I think these changes make the plot somewhat better in regards to your comments.
However, I should note one thing that is very concerning about the plot and about the data is the ROTSE3 data set; it seems to not agree well with the other three I am using (NSVS, ASAS-SN, SuperWASP) in the sense that some observations just seem too dim to be where they are on the phase plot. I will try to change the period to see if a new period may make the ROTSE 3 data behave more like the other sets, but the current period of 2.12642 days seems fairly well supported by the agreement of the other three sets. Do you have any recommendations as to what I should do with this data set, if anything at all?
Hopefully, this will be the last series of questions I have for this topic for now!