SPARCS WG on Redshift Measurement for Radio Continuum Surveys

Note that this group has been merged with the EMU "Redshift Estimation" Development Project

 (see http://askap.pbworks.com/w/page/44559147/Redshifts  )

Members:

• Ray Norris, CSIRO/WSU (cochair)
• Ken Duncan, Leiden (cochair)
• Minh Huynh, CSIRO
• Nick Seymour, Curtin U
• Andrew Hopkins, Macquarie Uni
• Dan Smith, U. Herts
• Mara Salvato, MPE
• Mattia Vaccari, UWC
• Duncan Farrah, Sussex
• Jim Geach, U. Herts, TBC
• Heinz Andernach, UGTO
• Matt Jarvis, Oxford
• Shea Brown, Iowa
• Kieran Luken. WSU
• Laurence Park WSU
• Alvise Raccanelli, UB
• Ely Kovetz, JHU
• Jose Bernal, UB 
• Michael Brown, Monash U
• Aaron Robotham, UWA,TBC
• Simon Driver, UWA
• Ned Taylor, Swinburne
• David Parkinsom, KASI 
• Michelle Cluver, Swinburne 
• Please email Ray Norris if you would like to join this WG, or if you think someone else should be invited

To post a message to all the list members, send email to redshifts@lists.csiro.au. Posts from non-members will be rejected.

For information on the listserver, see http://lists.csiro.au/mailman/listinfo/redshifts . Admin page is on http://lists.csiro.au/mailman/admin/redshifts

Immediate Goal

To provide a cross-fertilisation forum for the different groups working on techniques for redshift measurement for radio continuum surveys

Ultimate Goal

To construct a script where you feed in the ID of a source, and it returns

• a best estimate of redshift. In most cases, this will be derived on-the-fly  from a database or an ML algorithm based on available photometric data. In a few cases, it will be from a  published spectroscopic redshift catalog.
• the source of the redshift
• (if derived from an ML algorithm) a probability density function 

Overview:

The aim of this Working Group is to provide redshifts, or estimates of redshifts, for a large fraction of the millions of continuum sources to be discovered by next-generation radio surveys. All techniques, including spectroscopy, template fitting, and machine learning techniques will be explored.

There is no way we can get spectroscopic redshifts for the millions of sources from next-generation surveys such as EMU and LOFAR. Even accurate photometric redshifts will be really hard. Fortunately, most of our science goals dont need accurate redshifts. In some cases, even putting sources into a small number of redshift bins will be sufficient. So we are exploring new ways of estimating approximate redshifts. 

Next Videocon

• Tuesday 11  September 2018 06:00 UT = 16:00 AEST = 14:00 AWST = 07:00 UKDT = 08:00 Italy time   Agenda

Recent Papers

• Norris+2018, A Comparison of Photometric Redshift Techniques for Large Radio Surveys, PASP, submitted, pdf
• Luken+2018, PASP, Estimating redshift with k-Nearest Neighbours Regression, submitted, pdf
• Salvato+2018, The many flavours of photometric redshifts, Nature Astronomy, accepted, ads
• Smith+2017, The WEAVE-LOFAR survey, SF2a-2016, arXiv 
• Duncan+ 2018a, Photometric redshifts for the next generation of deep radio continuum surveys - I. Template fitting, MRNAS, accepted,  ads
• Duncan+ 2018b, Photometric redshifts for the next generation of deep radio continuum surveys - II. Gaussian processes and hybrid estimates, MNRAS, accepted,  ads
• Duncan+ 2018c, The LOFAR Two-metre Sky Survey -- IV. First Data Release: Photometric redshifts and rest-frame magnitudes, A&A, accepted, link

(Please add other papers here)

Activities

1) Optical Spectroscopic Redshifts

Pre 2020

Wide Area

• SDSS ( 2million redshifts)
• WEAVE-LOFAR (1 million redshifts, radio selected)
• WEAVE-QSO (optically selected) 
• Taipan  (1 million redshifts) 
• OZDES (38,000 redshifts in 2017)
• 2dF-GRS  
• What else? 

Post-2020

• WAVES (2 million redshifts) 

Deep Field

• MOONS
• WEAVE
• PFS

Post-2020

• WAVES-Deep/Udeep 

2) Radio Spectroscopic Redshifts

In the 2018-2020 period,  these will primarily come from large HI surveys such as: 

• WALLABY (500,000 redshifts)
• LADUMA
• MIGHTEE 

3) Conventional Optical Photometric Redshifts

• DES 
• 2MASS Photometric Redshift Catalogue (Bilicki et al. 2013; ~1,000,000 redshifts) 

4) Radio Photometric Redshifts

• Radio continuum SEDs are not as featureless as we once thought. Given wideband radio data (e.g. from EMU, MWA, and VLASS) maybe we can use them to estimate redshifts

5) Multiwavelength Machine-Learning Photometric Redshifts

This is a very active area. Current activities include:

• LOTSS photoz - see Duncan+ 2018c, above
• ML experiments with simulated EMU data by Luken, Norris, Park et al.
• what else? 

See also mlprojects.pbworks.com