Lasair Cookbook ∞

Find a fast riser ∞

See the cookbook entry Finding tidal disruption events (TDE) below, and do the first step only.

Find SNII from TNS ∞

Suppose you want those objects reported to TNS (Transient Name Server) which also have been assigned tns_type as “SN II”.

In the SELECT part of the filter form, put in:

objects.diaObjectId, 
crossmatch_tns.tns_prefix, crossmatch_tns.tns_name, crossmatch_tns.type

and in the WHERE part of the filter, we put in the classification constraint and latest first, and order the results latest first.

crossmatch_tns.type = "SN II"
ORDER BY lastDiaSourceMjdTai DESC

You can select on multiple TNS types with this syntax:

crossmatch_tns.type in ["SN II", "AGN"]

Filtering on TNS status ∞

In the example above, we are concentrating on those LSST objects that are associated with TNS objects. But sometimes we make a selection on objects, and just want to know if each object is in TNS, or not. Not an exclusive filter (TNS only), it’s a column in the table (may or may not be in TNS).

In this case, you can use the special attribute objects.tns_name. If this is in your SELECT clause, there will be a column with TNS name, or blank if none.

Finding alerts from my favourite objects ∞

If your science depends on a specific population, you can make a watchlist. Click “watchlists” on the front page of Lasair.

You can check the community resources first, perhaps the watchlist has already been built. Otherwise, make a file like

<RA> | <Dec> | <name> [|<radius>]

for example

003.4835000|-18.9018056|SHBL J001355.9-185406
008.3931667|-19.3591389|KUV 00311-1938

If your watchlist is in a catalogue hosted by Vizier, there are instructions how to make this file.

Then the file can be ingested using the Lasair web: Watchlists/Create New.

Then make a filter based on the watchlist: - Filter-builder web page: go to “Filters”, then select your watchlist. - With the API there is a notebook Query objects coincident with a watchlist that shows the steps.

When you Save the filter, you can enable streaming processing, so that you get alerts as soon as possible. Lasair supports email/batch and kafka, so thet your machine can get the filtered alert stream.

Poking about in Gaia ∞

Filtering Lasair

To illustrate, lets see if we can find any outbursts amongst the Gaia catalogue of White Dwarfs. There are a million of them in GaiaWhiteDwarfsDR3. Let’s start by looking at all alerts coincident with the watchlist (in Lasair: click Filters/Create New/Select watchlist/Run Filter/Save Filter). Already included in the sample filter is objects.lastDiaSourceMjdTai > mjdnow()-365 which means there must have been a detection in the last year. So we can keep that.

Let’s require a lightcurve with more than a few detections, say 10. We can utilise the lightcurve shape attributes jump1 and jump2. All you need for the SELECT is the objectId, than you can click on the link to see the lightcurve. The WHERE section is then:

objects.lastDiaSourceMjdTai > mjdnow()-365
and objects.jump1 > 5 
and objects.nDiaSources > 10

These “jump” parameters compare the latest diaSources, in all 6 bands, with a baseline history between 70 and 10 days ago. The jump statistics are each number of sigma for two wavebands. A full explanation can be found in the jupyter notebook jumpFromMean.

Absolute magnitude ∞

Under some circumstances, it is possible to estimate the intrinsic luminosity of an astrophysical object – usually called the absolute magnitude. This estimate is only made if the Rubin object has a good chance of being associated with a galaxy whose distance is known. Below we combine flux, sky position, and redshift to estimate the peak absolute magnitude of a Rubin object. Let \(f_\nu\) be the flux in nanoJansky, and \(\delta f\) its uncertainty.

Flux is related to the apparent magnitude (\(m_{AB}\) measured in the observer frame band O. See Lasair Concepts for a conversion table.

\(m_{AB}\) = -2.5 log \(f_\nu\) + 8.9, for \(f\) in Jy

\(m_{AB}\) = -2.5 log \(f_\nu\) + 31.4, for \(f\) in nJy

Let \(M_R\) be absolute magnitude in restframe band R. It is related to apparent magnitude, distance, extinction, and k-correction.

\(M_R = m_{AB} - \mu - A_O + K_{RO}\)

Here \(\mu\) is the distance modulus, and \(A_O\) is extinction in observed band for magnitudes in the AB system. Note that the foreground extinction we apply is for the Milky Way, so this is applied to the observer frame flux and magnitude measurement.

\(K_{RO}\) is the k-correction to convert for wavelength shifting from band R to the band O. An accurate K-correction requires knowledge of the input spectrum and the redshift, along with the filter throughput function to calculate synthetic photometry. A user may want to do this for full scientific analysis but since Lasair filters are designed for rapid selection we provide the approximate k-correction only (e.g. as described in Hogg et al. 2002). For redshift \(z\), we have \(K_{RO} = -2.5 log (1+z)\). Therefore:

\(M_R = m_O - \mu - A_O + 2.5 log (1+z)\)

Note that the rest frame wavelength is shifted: the effective wavelength for \(M_R\) is:

\(\lambda_R = \lambda_O/(1+z)\)

We can compute distance modulus \(\mu\) from redshift \(z\) following Ned Wright’s cosmology calculator. For the code that computes absolute magnitude, see the milky way notebook.

Error on \(M_R\): Sum the variances of the sources of error:

\((\delta M_R)^2 = (\delta M_O)^2 + (\delta \mu)^2 + (\delta K)^2\)

The error of magnitude comes from the error in flux: \(\delta M_{AB}\) = 1.08574 \(\delta f/f\)

Note that \(\delta K\) will be small: \(\delta K \approx {2.5 \over ln 10} ({\delta z \over 1+z})\) A Taylor expansion of the K correction term yields \(\delta K \approx 1.08574 ({\delta z \over 1+z})\).

\(\delta M\) will come from the uncertainty in redshift \(\delta z\), presume there is a python function to give \(\mu\) and \(\delta\mu\) from \(z\) and \(\delta z\) for the cosmology of \(H_0\) = 67 and \(\Omega_m=0.3, \Omega_n=0.7\).

Finding tidal disruption events (TDE) ∞

The beginning of a tidal disruption event begins with the first fuel being thrown at the black hole. The flux rises quickly, and is very hot, above 8000K. The NEEDLE project will search the LSST transient stream for such events, and the following filter is a possible way to reduce the deluge of LSST alerts to something manageable.

Fast riser ∞

We want objects with a rapid initial rise, and nothing there before. We can make SQL like this:

    mjdnow() - objects.lastDiaSourceMjdTai < 7   # must be diaSource in last 7 days
AND mjdnow() - objects.firstDiaSourceMjdTai < 30 # nothing before 30 days ago
AND objects.BBBRiseRate > 0.2                    # magnitudes per day

The first two lines are about epochs of first and last diaSources in this diaObject: we want a recent brightening, and nothing there before The last line refers to the bazinBlackBody parametric fit, and here we are requiring not only that the parameter-fitting was successful, but also the rise rate is rapid – 2 mags in 10 days. For more information see (Bazin Balck Body)[core_functions/lightcurve-features.md#BBB].

Blue ∞

The LSST survey that Lasair reports makes many “paired” observation, meaning only 30 minutes apart. For most astrophysical lightcurves, this time is essentially zero, meaning that these paired diaSources can be used for a colour measurement. There is a catch however, because of the six filters of LSST, and the many combinations: u+g, u+r, g+r, r+i, i+z, z+y (LSST Cadence doc here). With the two-filter ZTF, it was easy to intuit temperature from “g-r” magnitude differences. To simplify filter building, Lasair computes the equivalent temperature by fitting a blackbody profile to whichever pair generated the flux ratio. We can use this to remove objects that are not very hot:

AND objects.latestPairColourTemp > 10          # temperature (kK) > 10,000 Kelvin

You could look for there being an earlier pair where it is also very hot:

AND objects.penultimatePairColourTemp > 10

Sherlock ∞

Sherlock knows many catalogues, and compares them (crossmatch). Tidal Disruption events take place in the centres of galaxies, and we can look first at the catalogued galaxies. At the simplest level, Sherlock reports these categories - BS and VS (known bright star and known variable star), - AGN (known Active Galactic Nucleus), - CV (known cataclysmic variable), - HPMS means High Proper Motion Star - SN and NT (in the skirts of / in the centre of a galaxy)

It is this last pair that we use to find TDEs. We could also be more restrictive and just require the ‘NT’ classification. Here is the SQL

AND sherlock_classifications.classification in ('SN', 'NT')

Absolute Magnitude ∞

Lasair makes an estimate of peak absolute magnitude for anything that Sherlock reports with a host galaxy – that is SN and NT – and therefore has a distance measurement (z or photo-z). In this case, the lightcurve is corrected for extinction and the highest flux selected, then converted to absolute magnitude.

AND absMag < -19                  # peak extinction corrected absolute magnitude

Watchlist ∞

A further cut for this type of alert might be a publication of French and Zabludoff: Identifying Tidal Disruption Events via Prior Photometric Selection of Their Preferred Hosts. The catalogue of 57,000 galaxies is available from Vizier, and there are (instructions)[core_functions/watchlists.html] for how to ingest this into Lasair.

In order to filter alerts based on a watchlist: in the filter builder web form, select “E+A Galaxies”.

Bright enough ∞

If we are to request follow-up facility for this alert, it must be bright enough, so that their spectrograph will have enough time to do the follow-up There is a translation table and formula (here)[concepts.html#lightcurve], but this means brighter than magnitude 18.9 in either g or r filter:

AND (   objects.g_psfFlux > 100000               # 100000 nJ is mag 18.9
    OR  objects.g_psfFlux > 100000)

Finally ∞

You can choose the SELECT part of the SQL query as you like, but the WHERE section is where the filter is actually implemented. Here are the clauses from above collected together for the WHERE section:

    mjdnow() - objects.lastDiaSourceMjdTai < 7
AND mjdnow() - objects.firstDiaSourceMjdTai < 30
AND objects.BBBRiseRate > 0.2
AND objects.latestPairColourTemp > 10  
AND objects.penultimatePairColourTemp > 10
AND sherlock_classifications.classification in ('SN', 'NT')
AND peakExtCorrAbsMag < -19

with the watchlist F+Z galaxies chosen.

The idea of Lasair is that you build and share filters made from the attributes listed in the schema browser. The above sections give an idea of how ideas can be converted to filters, and you should choose the sub-filters and parameters yourself.