-
Notifications
You must be signed in to change notification settings - Fork 5
Sherlock Workflow
The settings in the settings file relating to the NED stream are:
ned stream search radius arcec: 300
first pass ned search radius arcec: 240
ned stream refresh rate in days: 90
To update the NED stream, for each transient coordinates the code does a conesearch on the tcs_helper_ned_query_history
table to see if a search has already been performed within the designated ned stream refresh rate in days
. If a match isn't found then NED is queried and the tcs_helper_ned_query_history
is updated for the transient coordinates.
The algorithm is written and modified within the sherlock.yaml
settings file. This means you can modify the algorithm without affecting anyone else's search (as long as you are working off the different transient databases).
search algorithm:
sdss qso:
angular radius arcsec: 2.0
transient classification: AGN
database table: tcs_view_qso_sdss_spect_galaxies_qsos_dr12
stop algorithm on match: False
match nearest source only: False
milliquas:
angular radius arcsec: 3.0
transient classification: AGN
database table: tcs_view_agn_milliquas_v4_5
stop algorithm on match: False
match nearest source only: False
veron:
angular radius arcsec: 2.0
transient classification: AGN
database table: tcs_view_agn_veron_v13
stop algorithm on match: False
match nearest source only: False
ned qso:
angular radius arcsec: 2.0
transient classification: AGN
...
Note, to remove a module temporarily, simply comment it out in the settings file (yaml treats lines beginning with #
as comments).
Behind the scenes there are 2 types of searches performed on the catalogues.
- Angular Separation Search
- Physical Separation Search
An example of an angular separation search looks like this in the settings file:
milliquas:
angular radius arcsec: 2.0
transient classification: AGN
database table: tcs_view_agn_milliquas_v4_5
stop algorithm on match: False
match nearest source only: False
The code performs a cone-search on database table
using the angular radius arcsec
. If matches are found the associated transient is given a transient classification
and the results are added to the tcs_cross_matches
table of the transients database. If stop algorithm on match
is true the code breaks out of the search algorithm and starts afresh with the next transient to be classified, otherwise the algorithm contines and all matches are recorded in the tcs_cross_matches
table. If match nearest source only
is true only the closest match from each catalogue query is be recorded in the tcs_cross_matches
table.
If the physical radius kpc
key is found in the conesearch module then a physical separation search is performed. First of all an angular cone-search is performed at the coordinates using a suitably large search radius. After this a further search is done on the physical distance parameters returned (distance, physical separation distance, semi-major axis length ...) for each match.
A physical match is found if:
- The transient falls within 1.5 x semi-major axis of a galaxy
- The transient is within the
physical radius kpc
of a galaxy
As before, all matches are recorded in the tcs_cross_matches
table.
If transients are found:
- within 2.0 arc of source, OR
- within 20 kpc of host galaxy AND assigned a SN classification, OR
- within 1.2 times the semi-major axis of the host AND assigned a SN classification
they are all given the same top level ranking for classification. After this catalogue weights come into effect to determine the orders of classifications. The catalogue weights are found in the [tcs_helper_catalogue_tables_info
](Crossmatch Catalogue Tables) table of the catalogues database and give an indication of the accuracy of the classifications of sources in the catalogue. For example the tcs_cat_sdss_spect_galaxies_qsos_dr12
is given a greater weight than tcs_cat_sdss_photo_stars_galaxies_dr12
as classifications of the objects based on spectral observations is more accurate than photometry alone.
Once the classifications for each individual transient are ranked, a final, ordered classification listing is given to the transient within its original database table. For example SN/VARIABLE STAR
means the the transient is most likely a SN but may also be a variable star.