Starlink Software Collection — 2021A Release Notes
The HDS-V5 data format (which was introduced in the 2018A release and uses HDF5) is now the default output format for all NDFs and other HDS files created by Starlink. It is still possible to create output files using the old HDS-V4 format by setting the environmental variable "HDS_VERSION" to 4 before running the command to create the file. The HDS-V5 format is described in Jenness 2015 2015A&C....12..221 .
- Perl updated to Version 5.32.0.
Changes to Applications
- New wrappers for AST functions: astdecompose, astmoc, astaddpixelmask, astaddregion, astgetregdisc
- All cupid commands now support NDFs containing more than 2^32 pixels.
- Some parts of the code have been multi-threaded to improve execution speed.
The FINDCLUMPS command has a new configuration parameter called GaussClumps.AllowEdge. It controls whether clumps found by the GaussClumps algorithm that touch the edge of the array are retained or not.
- The FINDCLUMPS command has a new option, "Ellipse3", for the SHAPE parameter (which determines the shape of the clump outlines stored in the output catalogue). It provides a more robust alternative to the older options "Ellipse" and "Ellipse2".
- Several bugs have been fixed.
- The cubehelix0 colormap has been added to the quick lookups in the main panel.
- A new toolbox to display IVOA MOCs over images has been added. MOCs can also be export from the image regions toolbox.
- Various bugs in the display of ARD regions have been fixed.
- Various bugs have been fixed.
- The following commands can now handle NDFs containing over 2^31 pixels (some have been multi-threaded for increased speed): ADD, CADD, CDIV, CHANMAP, CMULT, COLLAPSE, CSUB, DIV, FITSURFACE, MFITTREND, MSTATS, MULT, NDFCOPY, NDFTRACE, NUMB, PERMAXES, STATS, SUB, SURFIT, WCSALIGN.
A new command "COMPLEX" was added to manipulate complex NDFs, e.g. to extract the real and imaginary parts, compute the modulus and argument or to combine these parts into a new complex NDF.
New command "PIXBIN" performs generalised rebining of an NDF by reading destination coordinates for each pixel from a set of associated NDFs.
ARDMASK can now process complex data.
HISCOM has a new parameter called APPNAME that specifies the application name to be stored in the new history record (previously, "HISCOM" was always used).
LINPLOT parameter XMAP now has a new option called "LRLinear", which forces the X axis to be annotated linearly increasing from left to right.
LISTSHOW can now display NDF data values at the positions in the supplied positions list.
NDFCOMPRESS parameter METHOD can now be set to "BOTH", causing the NDF to be compressed twice, once using method SCALED and then again using method DELTA.
REMQUAL has new parameter CLEAR, which allows the quality bits corresponding to the removed quality names to be cleared.
SCATTER can now display a symmetric linear fit to the points in the scatter plot.
SQORST has new parameter CENTRE, which allows the centre of the squash or stretch to be specified.
SUBSTITUTE has a new parameter called LUT, which allows multiple values to be changed simultaneously.
POLSELECT now aligns the supplied mask and catalogue in sky coordinates rather than pixel coordinates.
The output catalogue column values stored by POLVEC for pixels that have negative total intensity have been changed. Previously all output values were set bad at such pixels. Now only P (percentage polarisation) is set bad.
A new option has been added to POLVEC and POLBIN that allows P and PI (polarised intensity) values to be debiased using the "modified asymptotic estimator". See https://arxiv.org/pdf/1407.0178.pdf section 2.5.
A new command called POLEDIT has been added, which allows an existing vector catalogue to be changed in various ways. These include changing column values, adding new columns, changing the debiasing method and recalculating all columns derived from I, Q and U.
The POLPLOT command can now draw all vectors with a constant specified fixed length.
- Various bugs have been fixed.
New command MATCHBEAM smooths a 450 um SCUBA_2 map so that it has the resolution of an 850 um map.
- New command GAU2FIT determines a 2-component Gaussian PSF from a map of a compact source of known diameter (e.g. a planet).
The half-wave plate origin given by configuration parameter PaOff now has separate default values for 450 um and 850 um, which differ by about 4 degrees.
Fix a bug that could result in sources being smeared out in the output time-stream data.
- New parameter APP allows the application from which the configuration should be read to be specified. Previously this was fixed at "Makemap", but now for instance it can be set to "skyloop" or "calcqu".
- A new parameter called POINTING allows pointing corrections to be applied as the data is gridded into a cube.
- If an error occurs, attempt to create maps from any remaining chunks before quitting. Previously, any remaining chunks were ignored.
- A new add-on dimmconfig file called "dimmconfig_pca.lis" allows removal of multiple common modes using principal component analysis. This can produce maps with flatter backgrounds.
- Improvements to diagnostics system.
- All the above CALCQU and MAKEMAP changes are also incorporated into POL2MAP.
The default Instrumental Polarisation (IP) model has been changed from JAN2018 to AUG2019.
- New parameter WEIGHTLIM specifies the lowest weight for usable observations, in the range 0.0 to 1.0 (default 0.05).
- A History record describing the invocation of POL2MAP is now added to all output NDFs.
- Do not include lists of input NDFs in the log file by default.
- New parameter TRIM to trim noisy edges off the output coadded maps.
- Allow user control of the SNR levels used to define the AST and PCA masks.
- New parameter SMOOTH450 causes 450 um data to be smoothed to 850 um resolution.
Debiasing may now be performed using the "modified asymptotic estimator". See https://arxiv.org/pdf/1407.0178.pdf section 2.5.
- A new column called AST has been added to the output vector catalogues. It indicates which island within the AST mask (if any) contains each vector.
- If the total intensity map comes from a non-POL2 observation, degrade it by a factor of 1.35 before using it to take account of losses within POL-2.
- An error is now reported if any map supplied as input to POL2MAP has a pixel size that is different to the value specified by parameter PIXSIZE.
- All the above MAKEMAP changes are also incorporated into SKYLOOP.
- Shortmaps and bolomaps are now created if they are requested via parameter CONFIG.
- New parameter INITIALSKY allows an initial guess at the sky map to be supplied.
- Now honours the !Ast.Mapspike_Freeze configuration parameter.
- Accept raw data as input (previously input data needed to have been flatfielded first).
- The included JCMT source catalog has been updated.
- Updated to Version 3.4-1.
- Updated to Version v4.8-1.
Changes to Libraries
- Various bugs have been fixed.
- An alternative API has been added that uses 8-byte integers for functions that receive or return integer pixel indices or pixel counts.
The Plot class has a new attribute TextGapType, which adds extra control over how the axis labels and title are positioned.
The Region class has a new method, astGetRegionDisc, which returns a bounding disc for a region.
The Region class has a new method, astPointInRegion, which tests if a single point is inside a Region.
The FitsChan class has a new attribute, AltAxes, that can be used to control when to write out FITS headers describing FITS-WCS alternate axis descriptions).
The KeyMap class now supports entries holding 8-byte integer values.
- The structure of the distributed source directory has been changed.
- Simplification of compound Mappings that include linear atomic Mappings has been improved.
- File extensions are now retained in CADC style provenance headers.
- HDS_V5 is now the default output format.
- An alternative API has been added that uses 8-byte (rather than 4-byte) integers to hold pixel counts and indices, thus allowing HDS arrays with more than 2^31 pixels to be accessed by application code.
- The Fortran NDF library has been completely re-written in C.
- An alternative API has been added that uses 8-byte (rather than 4-byte) integers to hold pixel counts and indices, thus allowing NDFs with more than 2^31 pixels to be accessed by application code.
- New tuning parameters FIXDT and FIXSW allow date/time and software path within new history records to be fixed at specified values (useful for regression testing purposes).
- Support tilde ("~") expansion when using HDF_V5.
- Extra routines added to support ADAM parameters with 8-byte integer type.
- Updated to on 4.0-2.
- Updated to v3.450.
Changes to Cookbooks and Guides
- There is a known threading problem that prevents the Java application Treeview from working on HDS Version 5 (HDF5) files.