SVN changelog

Versions prior to GitHub

Changes in r982: (July-18-2014)

• Now supports more table formats thanks to astropy.io.ascii
  • .csv - comma separated values
  • .tab - tab separated values
  • .tex - latex table (!)
  • more upon request
• updated some of the help text, especially for --catalog
• --catalog will now try to guess a file format if it doesn't recognise the file extension (it will just assume that the first two columns are floats of ra/dec).
• forced measurements (--catalog) now always use bkg/rms data and report the local rms
• wcs_sky2pix and wcs_pix2sky replaced with wcs_world2pix and wcs_pix2world so that astropy will stop complaining so much
• fixed a bug where islands with negative ra's were being reported (H/T Joe Callingham)
• background/rms output now contains a more correct fits header

Changes in r963: (June-03-2014)

• Fixed a bug introduced in r923, whereby Aegean would only find a single component in an island (H/T Shane O'Sullivan)
• Changed format of position angle in .reg files so that they were not reported in exponential format. This would cause ds9 to not load the file.
• Running Aegean on an image, and a negative version of the same image, will now produce identical source lists with all parameters the same (or with inverted signs where appropriate).
• Running --save on an image with no beam information will no longer fail if you supply --beam parameters.

Changes in r959 (May-30-2014)

• Changes have been made to deal with images that include pixels that don't have valid world coordinates (H/T Aaron Ewall-Wice)
• ra/dec strings will be "XX:XX:XX.XX" if the fitted position is invalid some parameter values will be set to nan/null
• flags will be set for sources that have WCS problems (see below)
• updated the flags that are set by aegean
  • A new flag (W=32) has been included that indicates problems with world coordinate systems. Don't trust any fit that has this flag!

Changes in r951 (May-5-2014)

• on systems that don't support os.poll() (eg OSX), Aegean will now revert to single core mode instead of crashing
• when using the --save option, the background and rms images are now masked to contain null values wherever the input image has a null
• forced rms is now used more consistently
• ra/dec errors are now constrained to be greater than the theoretical minimum described in Condon1997
• in some situations Aegean was not making full use of all the available cores - this has been fixed
• changes have been made to the --table option
  • you can now supply a comma separated list of tables to write. Eg --table=out.vot,out.reg,out.ann (no spaces please)
  • a new format has been included - .db or .sqlite. This will create an sqlite3 database with one table per source type.
  • fixed a bug whereby VOTables had only a single character for the ra/dec strings
  • tables of all formats (except .db) now have a consistent name format - ending in _comp, _simp, _isle
  • ds9 region files for islands now show a contour around the pixels used to calculate island properties. This replaces the boxes that were previously used. The island number is printed on the brightest pixel. This is the pixel that is used to calculate the peak_flux and the position.
• many changes have been made to the --island option
  • fixed a bug where the island integrated flux was being calculated incorrectly
  • the island integrated flux is no longer multiplied by a factor η. This correction factor is reported in the relevant tables and users can apply it as desired. See the BlobCat paper for a description of η: how it is calculated and when it is appropriate.
  • as noted above: ds9 region files now show a contour around the pixels for each island

Changes in r923 (March-13-2014)

• bkg/rms estimation is now done on multiple cores if available
• Aegean will now find both positive and negative sources in an image
  • Aegean will always find both positive and negative sources however the following options allow the user to control which sources are reported
    • --nopositive - don't report sources with positive flux, default=false
    • --nonegative -don't report sources with negative flux, default=false
  • Selecting both of the above options will generate no sources!

Changes in r908: (Jan-28-2014)

• now included Aegean user guide (.pdf) in the tarball.
• all catalog and table output is now sorted before writing
• ds9 .reg files now include an island/source number annotation
  • islands will have a box drawn around them
• Kvis .ann files have the island/source number commented after each line
• fix an issue with DS9 .reg files having an incorrect position angle. (H/T Laura Driessen)
  • DS9 seems to define PA as south of east, instead of east of north
• fixed a bug that could cause fits to fail when the beam position angle is outside of [-90,90]degrees. (H/T Natasha Hurley-Walker)

Changes in r903:

• properly fixed a previous fix that caused RA's<0 to be incorrectly adjusted to RA>0 but out by 12h (H/T Andre Offringa)
• RAs that are corrected from <0 to >0 now have a correctly measured error
• update to --island
  • (still not fully implemented at the moment, subject to change)
  • the output catalog printed on screen or to file specified by --out no longer contains island sources
  • islands are only printed to output tables
    • use --table=out.vot
    • islands will be in out_isl.vot
    • components will be in out_comp.vot
• update to --catalog
  • using --table=out.vot will produce out_simp.vot which is a VO table of the forced measurements

Changes in r897:

• pywcs/pyfits/ATpy have been replaced with astropy.
• the only output tables that are currently supported are VOTables, using --table=out.[vot|xml]
• Fits headers with an RA reference pixel that is <0 (which cause negative ra angles) will now have all positions translated to positive RAs.
• fixed a bug where all positions were off by one pixel in both RA and DEC
• reported errors are now scaled by the square root of the number of pixels used in the fit, to account for correlated noise signals.

Changes in r889:

• Fixed a but that could cause flux errors to be orders of magnitude larger than the measured fluxes during normal operation. Islands that are narrow (1-2 pixels wide) are now fit with a single component which is fixed to the shape of the PSF. (H/T Nadia Kudryavtseva)
• (partially) Implemented a --island option that will output the integrated flux of an entire island. NOTE the following:
  • This functionality should be considered experimental and the output catalog format may change in the future. Please give feed back if you find this functionality useful and would like to see it continued This integrated flux is calculated by summing the flux of the pixels within the island, dividing by the beam volume and adjusting by a fudge factor that accounts for the fraction of flux that is below the floodclip limit.
  • The catalog output currently has a "-1" as the component number to represent the island flux (as opposed to integers>=0 to represent individual components). The peak flux is the brightest pixel in the island, the location that of the brightest pixel. All erros are set to -1 since I don't know what to do with them yet. The a/b/pa are all 0 as they are not currently used. As noted above, this format will most likely change in future versions of Aegean.

Changes in r884:

• Fixed an error that caused the integrated flux to be wildly incorrect. (H/T Abraham Neben). Note: Integrated fluxes are calculated from peak/major/minor, and not measured/fit, and since the major/minor axes are allowed to be (slightly) smaller than the beam major/minor axes, this means that the integrated flux can sometimes be smaller than the peak flux.

Changes in r876:

• Fixed a bug that occurred when using option --cat and sources are outside of the image bounds. (H/T Nadia Kudryavtseva)

Changes in r872:

• Rewrote many functions that deal with WCS, this fixes many bugs including some that were undocumented.
• Fixed the WCS problems that were present in r837 and earlier. Most noticeably position angles are not correct for projections other than simply SIN.
• Beam objects are not stored in WCS coordinates instead of pixel coordinates. This means if you are using the --beam command line option, you need to specify major/minor/pa in degrees instead of pixels.
• Forced measurement (catalog) operation now also uses the updated WCS logic
• Many doc-strings have been updated

Changes in r837:

• Fixed bug caused by '{}' formatting for python2.6
• Added support for .reg (DS9) output files (use --table=out.reg)
• Fixed issue where .ann files were using the wrong position angle

Changes in r822:

• Fixed a bug when loading background and rms images.
• Fixed a bug that relates to "{:f}".format(numpy.int16(1)) and other stupid numpy oversights.
• It is now possible to write Kvis annotation files using the --table=out.ann option.
• It is now possible to supply/override beam parameters. Eg "--beam= 5 5 90" is the major/minor axes in pixels, and position angle in degrees. This is useful when your images don't have the BMAJ/BMIN/BPA fits keywords or they are incorrect.

Changes in r814:

• It is now possible to bypass Aegean's (awesome) finding and smart characterisation steps and just measure the flux at a given set of locations. This measurement is done by assuming that the sources are all point sources, and is the same approach that is taken in the VAST pipeline.
• Added a --catalog option that allows the user to specify the input catalog for the above changes. Catalogs can be .csv files (ra, dec decimal degrees in first two columns), or .cat files (same format as Aegean will output).

Changes in r808:

• You can now load the rms and background images from saved fits files with --rmsin and --bkgin. These files can be created with the --save_background. For large images this will save considerable time when reprocessing the same file. It is also possible that you can create these files with another algorithm.
• Fixed a bug whereby wcs headers were not being recognized for some files that were created by miriad (thanks to John Morgan for pointing this out).

Changes in r760:

• The major and minor axis of a source is now correctly reported as the beam major/minor when no fitting is done.

Changes in r706:

• Aegean now makes use of multiple cores (if available), to perform the island fitting. Use --cores=1 if you want to disable multiple cores.
• An earlier (non-parallel) version of Aegean can still be found at Aegean.tar (r687)