VStar 2.18.0 has just been released (thank you David Benn!). Since a change was made to the vstar.jnlp Web Start file, you should delete any old copies of this file that you have on your system and replace it/them with a new copy that you can get by clicking on the blue "Download VStar Now" button found on the VStar home page: https://www.aavso.org/vstar. Remember that if you save this file in a handy place, you can use it over and over again to launch VStar without having to return to the blue button each time and you will get the latest updates automatically. This happens to be one of those rare times whan a change is made to the vstar.jnlp file itself so you must download it again in order to get all of the most recent updates.
A summary of the major changes and enhancements is given in the ChangeLog which I have copied below, but I would like to point out one really interesting feature in particular - the "B-V series creator" plugin. This plugin can be used to add a B-V plot to any lightcurve containing Johnson B and V observations. It will look for B and V observations that are within a certain interval of time apart (you can specify the tolerance) then add a datapoint with the B-V value as a new series. You can then treat it like any other series or band in VStar.
To use this plugin:
- Open the "Plug-in Manager" in the "Tool" menu and install the "B-V series creator". Close VStar and re-open it again in order for the plugin to take effect.
- Request data on your favorite star with good Johnson B and V data
- Click on the "File" then "Create B-V series..." menu items
- Specify a suitable "tolerance" - anything greater than .1 days
Here is a B-V plot I just made for SN 2011fe:
This is what is in the ChangeLog:
Key Changes:
- Ticket #523 Handle transformed flag correctly in AID load
- Ticket #536 Allow exponential format in mag, time values
- Ticket #514 Remove all traces of JDBC from VStar (removed deprecated code)
- Ticket #364 Create VeLa: initial implementation of numeric expressions
- Ticket #331 Plot B-V index as a series
- Updated bands collection.
- Added WWZ and StarInfo scripting API functions.
-Sara
Thanks Sara. This release is hard to characterise: a motley collection of features and bug fixes.
I want to thank Robert Jenkins, a fellow South Australian and ASSA member, for prompting me to implement the B-V plug-in and for the testing, feedback, and encouragement he has provided. Like so many things, this had been on my TODO list for a couple of years, but his request and interest caused me to raise the priority. I know you've been having fun with it lately Robert. Enjoy!
Thanks Sara for demonstrating its use above. I'll write a short document for it soon, so it can be added to https://www.aavso.org/vstar-plugin-library. Note that a tolerance value of greater than 0 can be used; start with the default. It's most appropriate for temporally close B and V values. More later.
I'll post more follow-ups re: other items on the change log soon.
David
Hello Sara and David for the latest update to Vstar. I will definitely use it.
I have the need to determine the minima of some eclipsing binaries. Is there an algorithm implemented in Vstar to do this? I have an Excel sheet with the KK Kwee and Van Worden method mechanized, but it would be so much better to have such a tool in Vstar or another one that is recognized by the variable star community. What would the EB community like to see used for a method?
Gary
Hi Gary
VStar currently supports extrema finding for polynomial fits via the Models dialog. The user manual mentions this briefly on pages 82, 88, and 89. I've used a couple of methods in the past, currently a derivative based approach. Robert Jenkins tells me that he has been using this to good effect for eclipsing binary minima determination. Try this to see hiow it goes.
There are a number of things I want to say more about of a tutorial/FAQ nature, either in the manual, or more likely via a wiki or other online resource.
I have other related tasks on the TODO list (which only seems to grow
), including Kwee van Woerden. See these tickets:
Having access to more than one method, as with period search etc, is probably a good thing.
As with B-V, what tickets I prioritise is really affected by requests people make. Notice that each ticket has +|- "vote" buttons. Feel free to use them as well. Other than that, if you'd like to see, say, Kwee van Woerden implemented sooner rather than later, let me know.
Did you create the Kwee van Woerden spreadsheet yourself or source it from elsewhere?
Thanks.
David
Thanks for that info David. I did source the spreadsheet myself. Do you know of ones available online? I will go look, now that you mention it.
Gary
Hi Gary,
The Kwee-Van Woerden method is implemented in the software Peranso that many eclipsing binary people use for Time of Minimum determination. However this method seems to be falling out of favour.
I'm a co-author on a recent paper in OEJV and the referees required that Minima determination must not be carried out using KvW because the unrealistically small uncertainties delivered by the KvW algorithm mean those minima estimates get undue weight in statistical investigations, e.g. of period change. Polynomial fitting should be used instead.
Peranso can also fit polynomial curves to determine ToM (like VStar) and it also reports an error estimate. This is what I used for my ToM measurements in that paper. But these errors seem to me to be too large.
I'm keen to hear what software tools others use for ToM measurement. Cheers,
Mark
Hi Mark
Yes, I recall discussion about VSSS this year. I don't recall whether the problem was with particular implementations of Kwee van Woerden or the method itself.
David
Hi David,
does the VStar polynomial method provide an estimate of the error in the measured time of miniumum? Cheers,
Mark
Hi Mark
Apologies for the delay.
The resolution at which the polynomial is scrutinsed around the inflection point (using first and second derivatives) is currently 0.00001 of a JD. So, one answer is: +- 0.00001 days.
You probably had in mind something more like R2 or some other measurement of error. Happy to take guidance on this, in relation to what you have seen other tools, or just from a general approach viewpoint.
Which reminds me: I wonder whether the CHOICE course "Uncertainty about uncertainty" will ever run again? Presumably that examined error analyses from various perspectives.
David
Hi David,
I guess the +/- 0.00001 days you mentioned could be referred to a "mathematical" error value.
What I'm interested in is a "realistic" Time of Minimum error estimate which takes into account the scatter in light curve datapoints, error bars on those datapoints, and possibly the image cadence too. I have no idea how to calculate such a "realistic" ToM error but I'm sure there are standard ways that statisticians/scientists would be familiar with.
I'd like to see VStar's polynomial fit routine include such an error estimate, and I'd like to incorporate it in my Excel poly fit ToM spreadsheet.
It would be great to hear from anyone with expertise in this area.
Cheers,
Mark
Hi Mark
Yep, understood.
VStar's gives a couple of statistics for the polynomial fit itself: Aikake and Baysean Information Criteria (AIC and BIC). I implemented these because Grant Foster recommended them in his Light Curve Analysi book. RMS is provided also. See Analysis -> Models...
The polynomial fit of course also gives you a residuals dataset (as for all models) to which you can get ANOVA results via a mean series (p 41, 78, 78 etc of user manual). The current mode ANOVA plug-in is perhaps a bit nicer to use than the Info dialog approach (which will eventually be removed in favour of the plugin).
What statistic(s) does Peranso give?
I too would be happy to collaborate with anyone with greater expertise in this area.
David
Hi David, hi all,
I am usualy use for ToM the small software created by Bob Nelson - MINIMA. One can use several methods (see the screenshots).
Do not ask me how the different algorithms work
The link: http://members.shaw.ca/bob.nelson/software1.htm
also in AAVSO - Software Directory: https://www.aavso.org/software-directory
Best regards,
Velimir
Hi Velimir
Thanks for the reminder about Bob Nelson's Minima spreadsheet. I'll have a look at his error calculations etc.
Right now, VStar implements 4 of those methods: polynomial fit (actually, I see that Bob uses parabolic, so a polynomial of degree 2), Fourier fit, mean (binned in my case; not sure what Bob uses).
Do you know whether the spreadsheet uses VBA? It looks like it does from the screenshot. If so, I won't be able to run it on my Mac. I can borrow one my kids computers though.
David