6 Oct 2017: vast-1.0rc80:

 * Added a workaround for the PV/SIP WCS convention problem that does not allow
SExtractor to take into account the distortion corrections determined by Astrometry.net
see http://adsabs.harvard.edu/abs/2012SPIE.8451E..1MS for a detailed discussion.
The workaround is to use xy2sky routine from WCSTools to convert SExtractor-derived
pixel coordinates to celestial coordinates rather than relying on SExtractor
to do this conversion.

 * The magnitude measurements are now performed in multiple apertures for each image:
ref. ap. = median(A_IMAGE) * CONST the reference aperture (the same used by previous
version of VaST), here median(A_IMAGE) is the median semi-major axis lengths among all
the sources detected at the current image, CONST = 6 (may be changed in src/vast_limits.h),
 ap. 1 = AP01 * ref. ap., by default AP01=-0.1
 ap. 2 = AP02 * ref. ap., by default AP02=0.1
 ap. 3 = AP03 * ref. ap., by default AP03=0.2
 ap. 4 = AP04 * ref. ap., by default AP04=0.3
the values of AP* parameters may be change in src/vast_limits.h. The measurements in
multiple apertures are used to reject blended sources if '--magsizefilter' is enabled
(see below) or to select for each object an aperture producing its lightcurve with
the smallest scatter. The second is an experimental option, it is disabled by default
and may be activated by specifying '--selectbestaperture' or '-3' option on the command line.
Please try and see if enabling this option actually improves the results with your data.

 * The magnitude-size filter now includes a battery of filters aimed at rejecting blended
and extended sources. For each frame VaST rejects outliers on the instrumental magnitude
vs parameter plots where the parameters are A_IMAGE, FWHM_IMAGE, mag(ref. ap.)-MAG_AUTO,
mag(ref. ap.)-mag(ap. 1), mag(ref. ap.)-mag(ap. 2), mag(ref. ap.)-mag(ap. 3),
mag(ref. ap.)-mag(ap. 4), A_IMAGE/B_IMAGE.
The filtering is now enabled by default. You may disable it
with '--nomagsizefilter' or '-2' command line options. If image contains many artifacts such as
cosmic rays or hot pixels (often found in CMOS chips) that outnumber stars at some magnitude range,
the filter may do more harm than good. If this is the case for your images, try disabling the filter
and set more conservative SExtractor source detection parameters in default.sex
Details about VaST magnitude-size filtering for each image may be visualized in gnuplot using the log files
image*.cat.magsizefilter_* (for A_IMAGE) and image*.cat.magparameter*filter_* for the remaining
filtering parameters. See SExtractor manual for the detailed description of A_IMAGE, FWHM_IMAGE, MAG_AUTO.

 * The magnitude calibration script now scales the photometric errorbars, not only the magnitudes themselves.
This makes a big difference with 'photocurve' calibration at the bright end.

 * Photometric measurements with errorbars larger than MAX_MAG_ERROR=1.086/MIN_SNR=0.362mag (for the default MIN_SNR=3)
are now ignored at all processing stages. You may change the maximum allowed error in src/vast_limits.h

 * VaST now tries to rescale the errorbars determined by SExtractor using the relation
sigma_new=sqrt( (gamma*sigma_old)^2 + epsilon^2 )
where the coefficients gamma and epsilon are determined using all non-variable stars in the current dataset.
This relation is inspired by the OGLE rescaling of photometric errors in DIA as described by
http://adsabs.harvard.edu/abs/2017MNRAS.468.2189Z
http://adsabs.harvard.edu/abs/2009MNRAS.397.1228W
VaST is not allowed to make the errors smaller than the ones reported by SExtractor,
so the parameters are forced to have the values gamma>=1.0 and epsilon^2>=0.0.
Here epsilon may be interpreted as the systematic noise floor in the current dataset.
The photometric errorbars rescaling is enabled by default. It may be disabled by running
'./vast --noerrorsrescale /path/to/images/*fit' or './vast -4 /path/to/images/*fit'
The new line in vast_summary.log indicates if the errorbars rescaling was requested or not:
 Photometric errors rescaling: YES / NO
The photometric errorbars rescaling is usually not needed in the PSF photometry mode,
but should do no harm if applied.

 * VaST selects candidate variables (listed in 'vast_autocandidates.log' and highlighted in blue in './find_candidates')
as objects that stand out in one or more variability index vs. magnitude plots. VaST also lists
candidate non-variable stars in 'vast_list_of_likely_constant_stars.log' -- these are well-measured
stars that do not stand out in multiple variability indexes simultaneously. The process of selecting
candidate variables and candidate constant stars may now be controlled via the parameters set in 'src/vast_limits.h'
(they follow the 'Automated selection of candidate variables' comment). See http://adsabs.harvard.edu/abs/2017MNRAS.464..274S
for a detailed discussion of variability indexes implemented in VaST.

 * ./find_candidates now deletes the list of previously-viewed lightcurves kept in vast_viewed_lightcurves.log
if started with no command line arguments ('recompute everything' mode). In earlier versions this file
was kept which caused unexpected behaviour: every time the candidate viewer was run as
 './find_candidates aa'
(in the 're-display previously-computed variability statistics' mode) it was highlighting
all the previously-viewed objects, regardless of how many times variability statistics
were recomputed with './find_candidates'

 * It is now possible to send a lightcurve to the online period search tool from
the lightcurve viewer ('./lc') by pressing 'L' key when the lightcurve viewer was
not started from the VaST directory. For example:
 cd /path/to/my_lightcurve_archive
 /home/path_to/vast-1.0rc80/lc my_lightcurve.txt

 * Implemented a more efficient procedure that should speed-up lightcurve writing
on all systems supporting OpenMP.

 * Fixed the bug that prevented VaST from saving magnitude calibration details
under some circumstances. The symptom was that vast_magnitude_calibration_details_log
was an ASCII file instead of being a directory containing the calibration details.

 * VaST should now be able to understand observation start dates written in DATE-BEG
FITS header keyword instead of the usual DATE-OBS (as recommended by the draft FITS Standard V4.0).

 * VaST may now open "slice of a data cube" i.e. an image with NAXIS = 3 and NAXIS3 = 1.
Such slice number $CUBE_SLICE may be extracted from an actual 3D data cube named $INPUT_IMAGE_CUBE
using 'fitscopy' example program from CFITSIO:
 fitscopy "$INPUT_IMAGE_CUBE[*,*,$CUBE_SLICE:$CUBE_SLICE]" \!"$OUTPUT_IMAGE_NAME"

 * VaST will now complain if it encounters a FITS image compressed by MaxIM DL.
These images have to be uncompressed in MaxIM DL before they can be processed with VaST
(or any other software that is not MaxIM DL - I love proprietary data formats).

 * Changed axis labels in './find_candidates' to indicate that the displayed values of MAD and IQR
are scaled to sigma (to ease comparison with the mag-sigma plot).

 * The new '--starmatchraius' command line option may be used to tweak the star matching radius -
the maximum distance in pixels (after applying XY coordinates transformation) between the objects
detected on two images for these objects to be assumed the same. You may need to tweak this parameter
in case you experience problems with blending.

 * Fixed the numerical overflow issue in the flat-field correction routine 'util/ccd/md'
that was causing centers of saturated stars to appear black.

 * Added support for the I band to the magnitude calibration script 'util/magnitude_calibration.sh',
as with the R band, I magnitudes are computed from APASS r and i magnitudes of the field stars
using the color transformations by Jester et al. 2005 AJ, 130, 873 http://adsabs.harvard.edu/abs/2005AJ....130..873J

 * A safe value for one of the memory-management parameters MAX_MEASUREMENTS_IN_RAM in src/vast_limits.h
is now set automatically at compile time based on the available amount of physical memory.

 * Minor bug fixes.