The Starlink Project was a long running UK Project supporting astronomical data processing. It was shut down in 2005 but the software continued to be developed at the Joint Astronomy Centre until March 2015, and is now maintained by the East Asian Observatory. The code is open source.
Starlink News was last updated 2021-12-27.
Getting the Software
The Joint Astronomy Centre and East Asian Observatory have made a number of releases. The most recent is 2021A which can be downloaded from here where installation instructions are also provided.
Please note that there was a leap second at the end of December 2016; data taken since then will be gridded wrongly by all releases prior to 2016A. Please use 2021A (or 2018A, 2016A, or a development rsync version) to reduce JCMT data taken from this point onwards.
For a cutting edge version, you can rsync from the East Asian Observatory's build.
Citing the software
If you have used Starlink software in your research, please cite the software in your papers.
For the Starlink software package please use:
Starlink citation: Currie et al. 2014 2014ASPC..485..391C
- Starlink acknowledgment: “The Starlink software (Currie et al 2014) is currently supported by the East Asian Observatory.”
Most recent reference:
Berry et al. 2022 2022ASPC..532..559B
All of the individual packages are on the Astronomy Source Code Library, which have entries in ADS and can be cited in major astronomy journals. If a package has no other reference, please cite the ASCL entry. The following packages have a preferred reference you should use instead or as well as the ASCL entry (links are to the ADS record for the reference):
SMURF: For makemap (SCUBA-2 DR) please cite https://adsabs.harvard.edu/abs/2013MNRAS.430.2545C
For makecube (heterodyne/ACSIS DR) please cite https://adsabs.harvard.edu/abs/2015MNRAS.453...73J
The FellWalker algorithm (implemented inside Cupid's clumpfind) has its own citation: 2015A&C....10...22B Berry, D S 2015 "FellWalker-A clump identification algorithm" in Astronomy and Computing
SPLAT-VO ascl:1402.007 prefers 2014A&C.....7..108S
- SPLAT: ascl:1402.008
NDF: http://ascl.net/1411.023 prefers: 2015A&C....12..146J
PAL: http://ascl.net/1606.002 prefers: https://adsabs.harvard.edu/abs/2013ASPC..475..307J
SURF: ascl:1403.008 prefers https://adsabs.harvard.edu/abs/1998ASPC..145..216J
AST: ascl:1404.016 prefers https://adsabs.harvard.edu/abs/2016A&C....15...33B
SLALIB: ascl:1403.025 prefers https://adsabs.harvard.edu/abs/1994ASPC...61..481W
In addition, all Starlink SUNs also have bibliographic records on ADS.
The original Starlink web site has been mothballed but additional support can be found:
For questions about JCMT data reduction please see the JCMT help desk
For other questions about any of the Starlink software (e.g. installation issues, software bugs or other errors) please use the Starlink support mailing list
There is also information on the Starlink wikipedia entry.
Support Pages for Starlink Applications and Libraries
Some Starlink applications and libraries have associated support pages:
Developing and Building the software
The source code is freely available. Development and build questions can be asked on the Starlink development mailing list
The source code is directly available from a Git source code repository. Download the source code:
% git clone https://github.com/Starlink/starlink.git
You can download Git from its home page. A Git primer is available along with the Starlink Git development policies. The bulk of the collection is distributed under the GNU Public Licence, and the remainder under the old Starlink Licence.
The README file in the top level of the source code contains instructions for building. Please also see Building Starlink on macOS for some macOS tips (including a build script and list of dependencies).
Refer to the code in individual packages for more information.
We are trying to write new code in C rather than Fortran since this gives us much more flexibility in terms of threading and compilers. If you are a Fortran programmer familiar with Starlink programming some useful hints are available.
Primarily for local consumption, we also have a list of steps to take when making a release.
EAO and its plans for Starlink in the future (2018-07)
East Asian Observatory does not anticipate moving away from NDF and Starlink for its data format and astronomy software. We have therefore been working to ensure Starlink will still be fit-for-purpose over the next years as we upgrade the instrumentation at the JCMT. The current work has focused on:
- moving to HDS version 5, using the standard HDF5 file format. This
allows the larger array sizes necessary with large-format heterodyne instruments.
- Porting critical libraries and applications from F77 to C, to
remove single-thread bottlenecks and allow for long-term maintainability. ARY has been finished in the 2018A release, and NDF is currently being converted.
We would also like to acknowledge the continuing help and support provided by the wider Starlink developer and user communities, and reiterate our continued support of Starlink as an open source project.
We recognize that currently the most popular scripting language among our users is Python, and we are continuing to look at the best ways to allow easy use of NDF and Starlink from Python, and improving interoperability with the wider astronomical python community. Currently, there are existing HDS and NDF python interfaces, the PyAst package, and the starlink-pywrapper package to enable easy calling of Starlink routines from Python.
If you are interested in the history of Starlink there is a lot of documentation available.
The Starlink Bulletin was published between January 1988 and August 2004.
The Starlink Chronicle logged notable events and staff changes.
The former Atlas laboratory has a photographic record of the early days of its Starlink node, and of the Starlink inauguration. There is also an overview and summary of the genesis of Starlink.