Affiliation
American Association of Variable Star Observers (AAVSO)
Sun, 09/17/2017 - 15:42
What is the best method in VStar to determine the precise times for T(max or min) especially for short period (<1 day) variables?
Thanks,
Dave Cowall (cwd)
I'm not sure if it is "best", but what I do is to fit a polynomial to it. Then I examine the model for extrema. Both of these are on the Analysis menu.
Have an extema fine day,
Cliff
MINIMA V2.3 of Bob Nelson is a good tool to find minima in different ways. You can ask Bob Nelson (bob.nelson@shaw.ca) for a copy because the link in AAVSO site is broken: https://www.aavso.org/software-directory
PERSEA of Gracjan Maciejewski works also very well (gm@astri.uni.torun.pl)
Regards,
Velimir
Bob Nelson's software tools are now hosted on the Variable Stars South website (https://www.variablestarssouth.org/bob-nelson/). You can find the Minima program along with several others under the Software link. Cheers,
Mark
Hey All,
The link to Bob Nelson's software now points to the correct place. However, EBMin23 .zip downloaded just fine for me from the software directory. Is there another place where this link is broken? Also, since we do not support this software, we are not always aware when links change. If you notice that one does not work, please send an email to aavso@aavso.org and we will make sure it gets fixed.
Thanks,
Bert Pablo
Staff Astronomer
AAVSO
Hi all
Cliff mentioned the currently available approach in VStar (thanks).
Perhaps Bob Nelson's algorithms could be captured as VStar plugins? Is this software Windows-only currently?
Always happy for others to help with plugin development.
I need to set priorities because there are plenty of things for me to work on here:
https://sourceforge.net/p/vstar/bugs-and-features/?limit=250
even if it was the only thing I spent my increasingly limited spare time on.
Of course, there's plenty of room for multiple tools and approaches and I've never claimed that VStar does everything, despite efforts to the contrary.
David
So after we have an estimate of the time of a minimum or maximum how do we assign error estimates / confindence intervals to our time estimate? You always want an error estimate to go along with a value.
I thought of providing the +/- time differences that correspond to the RSM error of the fit but have no idea what confidence interval/p-value corresponding to such a measure. There must be a generally accepted way of doing this because you see error estimates associated with times of minima and maxima in the literature.
Brad Walter, WBY
Hi Brad
It's not something I've spent much time reading the literature about. For example this 2011 paper sets out two ToM uncertainty determination methods, one of which is to use the 2nd time derivative:
https://arxiv.org/pdf/1111.1142.pdf
so that the error is defined in terms of a fraction with the 1st time derivative (of a polynomial) in the numerator and the absolute value of the 2nd time derivative in the denominator.
Bob Nelson's software is, from all accounts, awesome. It's not open source however, so I can't look at it to discover its uncertainty algorithm. Perhaps it's documented in one or more papers by Bob. The papers here all seem to be applications of methods as opposed to documenting the methods themselves:
https://www.variablestarssouth.org/recent-publications-of-dr-r-h-bob-nelson
Perhaps details of algorithms are woven into one or more such papers, VSS newsletter articles or elsewhere. I have mostly heard people talk about the methods generally (KvW, polynomial, ...).
An ensemble of ToM algorithms such as Bob has created makes good sense just as using multiple period search algorithms can be a Good Thing, and could be implemented as one or more VStar plugins with the added advantage of being cross platform.
Other than that: I'm happy to take advice.
David
Unfortunately the last 1/2' to 3/4" of text in each line is cut off in this PDF every Web location that has the paper without charge references the same Cornell archive. I don't think that is a mistake by Cornell. I suspect that that is what they were given by Springer. The Postscript file is also corrupted. The alternative is to pay Springer $40 to get the paper. Springer does this kind of "stuff." Sorry $40 for a paper is outrageous.
Brad
There was apparently a problem somewhere in the internet ether. I contacted Cornell and they said the problem was I had the wrong margins set, but they were set to 8.5x11 and the right side was cut off in the browser as well as printed to PDF. I revisited the site without changing anything and the paper displayed correctly and the PDF downloaded correctly. It's one of those internet mysteries.
Brad
Hi Brad
Even when I viewed it, all the text seemed to be hard right justified, i.e. right up against the right edge, which seemed odd.
David
I downloaded it. Now, (since my virus software deleted PERANSO and I can't get it back) how do I start Vstar???
Did you download the jnlp icon or the zip file with the stand alone program and are you running Windows or something else? also do you have Jave run time environment installed?
Brad Walter, WBY
Java installed & updated. What icon? How do I get the icon?
Win7
Well, I found the icon in the Vstar folder, clicked on it, Win Photo Editor opened. Not Vstar.
Sara tells me that I have to right click the vstar.jnlp file.
See below for details right click on the download vstar.jnlp button on the Vstar overview page on the AAVSO website.
Go to AAVSO website/Data tab/Select Data Analysis/Select Vstar- Java ...
On that page is a blue rectangular button with the legend Download VStar Now
Right click the button and select "Save Link As" from the drop down list
A normal Windows "Save As" dialog will open with vstar.jnlp in the file name box
Save it to your desktop.
You will now have a VStar.jnlp Icon on your desktop. You have to have Java Runtime Environment 1.8 installed. I think if you have the full Java development package installed of the equivalent generation or later you don't need to install the JRE, but I suspect that you don't have the development package installed.
If Java is installed Double click the VStar.jnlp icon tht sownloads the app every time you use it from the AAVSO server and it removes itself when you close the program. You do not have any of the plug-ins however and you want them all. Download them using the Plugiin manager.using the directions on this page.
Brad Walter
PS: I think the default plug-in installation creates two folders in the top level of your user directory i.e. in C:\users\<your user directory name>
These folders are vstar_plugin_libs and vstar_plugins. At least that is where it put mine.
There is also a stand alone installation that doesn't need an internet connection to run. There is a link to the SourceForge page from which you can download it. put the zip file in the same directory that contains the two plug-in folders mentioned in the previous e-mail. Unzip the file to the same directory it is in. It wil create a new folder "vstar". Inside that folder at the top level you sill find a file vstar.exe.
Right click that exe file and select "send to" from the drop down list and and from the next level list select "Desktop (create shortcut) that will create a short cut to launch the stand-alone version. I don't remember if you have to reinstall the plug-ins in this version or whether it just finds them from the default location. If you changed the default installation for the plug-ins when installing them in the vstar.jnlp app, make the same changes in the stand-alone version or wyou will end up with two sets of plug-ins.
et voila
Brad Walter, WBY
PS: after you have unzipped the zip file and verify the program is working you can delete the zip file or, if you have slow internet, you can save in on an external drive with other programsyou want to reload in case of a disk crash.
Thanks Brad. Re: this:
the WebStart and standalone applications will look for plugins in the same directory.
David