Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://saistud.sai.msu.ru/poisk/
Дата изменения: Mon Feb 22 00:20:21 2016 Дата индексирования: Sat Apr 9 22:22:00 2016 Кодировка: Поисковые слова: п п п п п п п п п п п р п р п |
VaST [Variability Search Toolkit]A software for variable star detection
| ||||||||||
VaST is a software tool for finding variable objects on a series of astronomical images. The images (CCD frames or digitized photographic plates) must be taken with the same instrument using the same filter and saved in the FITS format. CCD images should be calibrated (dark-subtracted and flat-fielded). The input images may be shifted and rotated with respect to each other, but they have to have the same scale (arcsec/pix). The images in a series should overlap by at least ~40% to ensure successful cross-identification. VaST performs object detection and aperture photometry using SExtractor on each image, cross-matches lists of detected stars, performs magnitude calibration with respect to the first (reference) image and constructs a lightcurve for each object. The sigma-magnitude, Stetson's L variability index, Robust Median Statistic (RoMS) and other plots may be used to visually identify variable star candidates. If data permit, period-search techniques may be employed to identify periodic variable stars among the candidates. All lightcurves are constructed in arbitrary magnitudes with zero magnitude corresponding to the background level on the reference image. If there are reference stars with known magnitudes in the field of view, all magnitudes can be later converted to the standard system. Unlike software based on the image subtraction method, VaST can be used in case of unstable PSF (e.g., bad guiding or with digitized wide-field photographic images). VaST is not tied up to any external catalog and WCS (if operated in its main "variable star search" mode), so it can be used on images taken with a telephoto lens as well as with a 2.6-m class telescope. If needed, absolute astrometric calibration may be performed through an interface to the Astrometry.net code allowing one to perform automatic magnitude scale calibration and variable star identification (however, this will work only if the input images are large enough to perform blind plate solution). VaST is written in C (and partly in BASH scripting language) for GNU/Linux operating system. The latest versions are also tested on MacOS X and FreeBSD. The best practical way to run VaST under Windows is through Linux installed in a virtual machine (e.g. VirtualBox, see also a collection of pre-built VirtualBox images). The AAVSO forum thread mentions attempts to run VaST under Windows with Cygwin, but the results seem to be inconclusive. On all the supported systems VaST is able to utilize multiple processing cores to perform most of its computations in parallel. VaST is free software: you can redistribute and/or modify it under the terms of the GNU General Public License. The program is in a continual, albeit slow, state of development. You are welcome to contribute bug reports, patches and feature requests: please write me at kirx[at]scan.sai.msu.ru ScreenshotsVarious VaST subroutines (lightcurve plotter, image viewer) may be started automatically from the main program or manually by a user from a terminal. Names of the corresponding executable files are given under each screenshot. Click on images to see them in high resolution:
Video (screencast) voice comments are in Russian only, sorry... I hope to prepare an English version for quite a few years already...
Getting VaSTTo compile and use VaST you'll need:
You can find at this page the latest version of VaST (download via ftp or http, ~9.5M) along with the sample data set (ftp or http, 24M). Files described here as well as older versions of VaST are available via our anonymous FTP.You may download VaST with the command wget -c ftp://scan.sai.msu.ru/pub/software/vast/vast-latest.tar.bz2
unpack the archive containing the program
tar -xvjf vast-latest.tar.bz2
and compile it by running
cd vast-1.0rc75
make
(or 'gmake' if you are not on Linux).
At this point VaST should be ready for work.
If compilation fails, read the output carefully. Most probably it can't find some necessary libraries, header files or external programs. Please install the required programs and try to compile VaST again. If you are sure that the needed libraries and headers are installed in the system but the compiler cannot find them, try to open Makefile with a text editor and change the libraries locations to match your environment. Very old versions of VaST require PGPLOT, CFITSIO and GSL libraries to be installed in your system, but most likely you don't want to use such an old VaST version. Special note for Ubuntu users! The following installation procedure is recommended:
Special note for FreeBSD users: VaST scripts may fail to compile the internal copy of PGPLOT library. If that happens, please install PGPLOT system-wide with the command (as root): pkg install pgplot
then try to compile VaST again.
Please use gmake instead of make to compile VaST.
Using VaSTTo try VaST, download the sample data set, unpack it
wget -c ftp://scan.sai.msu.ru/pub/vast/sample_data.tar.bz2
change to the VaST installation directory and run the program
tar -xvjf sample_data.tar.bz2
cd vast-1.0rc75
After a brief computation the variability search window will be opened. Click
on any star on the sigma-magnitude plot to inspect its lightcurve. Click on
any point on the lightcurve plot to see an image from which this point
comes. The star on the image will be marked with a red cross, a red circle
around the star corresponds to the aperture used to measure this image.
You may switch from sigma to other variability indexes by pressing 'M' or 'N'
on the keyboard in the sigma-magnitude plot window.
./vast ../sample_data/*fit If you quit the program without deleting any data, you can restart it without doing all the computations again by running:
./find_candidates aa
To calibrate the magnitude scale using comparison stars with known magnitudes within the field of view use this script:
util/magnitude_calibration.sh
It will ask you to specify one or more comparison stars and their magnitudes.
For each comparison star, click on this star on the displayed image then
enter its catalog magnitude in the terminal window.
After the comparison stars have been specified, close the image window with the right mouse click and inspect the relation between instrumental and catalog magnitudes. This relation may be fitted by a linear function with the slope fixed to 1, linear function with a free slope, second degree polynome (parabola) or the Bacher et al. (2005, MNRAS, 362, 542) magnitude calibration relation a.k.a "photocurve". This option is suitable for processing digitized photographic plates. Use (P) key on your keyboard to change the fitting function. For most CCD data sets, linear function with the fixed slope provides a good fit. The data points may be weighted according to their estimated errors or weighted equally. Use (W) key to change weighting. Close the dialog with the right mouse click to apply the calibration to all stars when you are satisfied with the fit. If the image field of view is large enough to be blindly solved with Astrometry.net code (as described below), one can automatically match stars detected on the image to the UCAC4/APASS catalog by running the magnitude calibration script with a command line argument specifying the observing band:
util/magnitude_calibration.sh V
The acceptable bands are:
"C" - UCAC4 unfiltered red band,
"B" - APASS B band,
"V" - APASS V band,
"R" - APASS R band,
"r" - APASS r band,
"i" - APASS i band. Magnitude calibration against APASS can be performed
only if the field of view overlaps with the APASS survey footprint.
Note that the automated magnitude calibration will not work with the
sample data set due to its small field of view.
In order to save the current work, use the script
util/save.sh my_favorite_field_name
If the field name was not provided, the script will ask for it. To restore previous
work, use
util/load.sh my_favorite_field_name
or just copy all saved files from directory [my_favorite_field_name] back to the VaST directory.To delete all files from the previous work, you can use:
util/clean_data
There is a tool to automatically identify variable stars detected with VaST. It requires a working Internet connection at runtime. The tool may or may not work for your field depending on luck. If the field of view is greater than, say, 30' and the image quality is fine - the odds that the automatic plate solution will work for your field are pretty good. Here is how to use it... To identify the star number 00190 in the current dataset use the following command:
util/identify.sh out00190.dat 25
where "out00190.dat" is the lightcurve file of the star and "25" is the
estimated field of view size in arcminutes. It is wise to set this value a little bit smaller
than your best-guess field of view. If no field of value is given, the
default value of 40' will be assumed. Note, that the field of view
identification may take really long time (up to ~30 minutes), especially for small
fields. Please, do not interrupt the process - just wait for a result. If
the field identification failed, try to restart it with smaller estimated
field of view size. On successful identification, the script will print the
equatorial coordinates (J2000) of the star and will attempt to query popular
databases to check if this star is already known as a variable. Note, that
the coordinates returned by the identification script may be up to 10" off.
We strongly advise to use the Aladin
interactive sky atlas to verify the identification and obtain more
precise star coordinates from some astrometric catalog (e.g., USNO-B1.0).
The automatic identification function relies on tools provided by Astrometry.net. For more details visit
http://astrometry.net/. Note, that
according to the
use conditions of the index files required to run the Astrometry.net
code, you need to be ready to share the images solved using this software
with any other interested party including the Astrometry.net team. According
to the same use
conditions: "If the indexes are used in any scholarly work,
research grant proposals or other scientific or academic
publications, the Astrometry.net project must be properly cited
and acknowledged. The paper to cite is the Lang et al paper on our
bibliography page http://astrometry.net/biblio.html.
The acknowledgment text is "This research made use of tools provided
by Astrometry.net." The VaST software developers assume, that this statement
concerns the work which uses "util/identify.sh" script from the VaST
package.
A note on the graphical user interfaceThe VaST GUI is based on the PGPLOT library. The main reasons are that this library is perfectly suited for displaying and editing data and image plots and is so easy to use from programmer's standpoint. The problem is that the resulting interface may look counterintuitive. The unusual thing about the VaST interface is that it has no buttons. (There is a way to program a GUI with buttons and graphical input fields in PGPLOT, but, trust me, you'll not be impressed by its look and usability.) So, whenever a user has a choice between multiple possible actions, instead of clicking a button to execute the desired action, a user will have to physically press a key on a keyboard. The list of possible keys is always displayed by VaST at the terminal. Please have a look there to know your options. Some actions are common to all VaST windows:
Using VaST with scanned photographic platesIf you have to deal with digitized photographic materials, you'll want first to convert TIFF images (which are typically produced by scanner software) to the FITS format. It can be easily done with the tiff2fits converter. Do not forget to change your images to positive (white stars on black sky). For example:
./tiff2fits -i input.tiff output.fits
Also, you'll have to enter the information about the date and time of the
observation in fits header. This can be done, among other ways, using the
modhead program from the CFITSIO
examples page, MissFITS or edhead from the WCSTools package. You may
enter the observation date and time information in the image header by generating
the usual "DATE-OBS" and "EXPTIME" header keys, or by simply putting the
middle-of-exposure Julian Date into the "JD" key.
You will most likely need to modify the default SExtractor settings stored
in the "default.sex" file located in the VaST directory. Please use
"default.sex.beta_Cas_photoplates" and "default.sex.PHOTO" as working examples
(these files are also found in the VaST directory).
Say, you have digitized photographic images in "photo_data" directory. To run VaST on them, change to the VaST directory and run the program
cd vast-1.0rc75
The "-o" parameter tells VaST to use formulas (1) and (3) from
Bacher et al. (2005, MNRAS, 362,
542) a.k.a. "photocurve" for magnitude calibration. This technique provide slightly better
results for the aperture photometry on photographic data compared to the
second-order polynome used in VaST by default for magnitude calibration.
However, the standard calibration technique should also provide reasonably
good results (see Figure 2 in this paper).
The "-j" parameter that forces VaST to use a linear position-dependent
magnitude correction (i.e. stars on one side of the image bing
systematically fainter than on the other side) might also be useful for
photographic plates photometry.
./vast -o ../photo_data/*fit An example photographic dataset is available here. Beware, it's about 1G in size! You may find some more details in the ArXiv:1403.5334 e-print "A search for new variable stars using digitized Moscow collection plates". Search for transients with VaSTApart from the standard "variable star search mode", where a long series of images is processed and lightcurves of detected stars are explored using the sigma-magnitude plot, VaST can be also used in the specialized "transient detection mode". In this mode VaST will process a series of four images: two first-epoch (reference) images and two second-epoch images. VaST will generate an HTML report containing a list of candidate transients. The transients are defined as either objects that were not visible on the reference images, or where at least 1 mag. fainter compared to the second-epoch images. Two first-epoch images are needed because image artifacts and star blending may cause VaST to miss a star on a single reference image, so it will appear as a false transient search mode is available. If there are two reference images - chances that the program will fail to detect a star at both of them is greatly reduced. Also note that two different reference images are needed: it would be impossible to "cheat" the program by feeding the same image to it twice. An example dataset that allows one to test the transient search mode is available here. The unpacked archive will contain two sub-directories with reference and second-epoch images. The search should be conducted in two steps:
# Process images with VaST
./vast -x99 -ukf ../transient_detection_test_Ceres/reference_images/* ../transient_detection_test_Ceres/second_epoch_images/*
# Run script that will generate an HTML report
util/transients/search_for_transients_single_field.sh
Be prepared - the script will ask you to download a few catalogs needed for its work, including Tycho-2, the catalogs are more about 200M in size. The transient search report will be placed in transient_report/index.html that should be opened with your
favorite web browser. The report will contain a few false candidates and some good ones including asteroids and variable stars.
Known limitations:
Some switches that you may use with VaST"-P" — perform PSF-fitting photometry with SExtractor and PSFEx."-9" — use DS9 instead of VaST's own viewer pgfv to view FITS files. "-u" — always assume UTC time system, don't perform conversion to TT. "-k" — force VaST to ignore the "JD" keyword in FITS headers. "-x2" — accept stars with non-zero SExtractor flag: 2 - blended stars, 4 - saturated stars, 99 - accept everything. "-e" — failsafe mode: only stars detected on the reference frame will be proce |