The DLT40 Survey and GW 170817

Distance Less Than 40 Mpc (DLT40) is a supernova (SN) search led by David Sand (U. Arizona) and Stefano Valenti (UC Davis) that uses two, identical, 16-inch diameter PROMPT telescopes, at Cerro Tololo Inter-American Observatory (CTIO) in Chile and at a dark site in Western Australia, to observe ≈400 — 600 galaxies (within 40 Mpc) per night per site (seasonally dependent). The goal is to discover SNe only hours to days after the explosion: This is the best time to gain insight into the SN’s progenitor and explosion physics. In two years, DLT40 has discovered 18 SNe, with 10 of them discovered within 48 hours of the explosion, so far resulting in ≈4 journal publications per year.

DLT40 discovery of the first optical counterpart to a gravitational-wave event. From left to right: Last non-detection, discovery image, and difference image. DLT40 was the second of six groups to co-discover the    kilonova   , associated with GW 170817 .

DLT40 discovery of the first optical counterpart to a gravitational-wave event. From left to right: Last non-detection, discovery image, and difference image. DLT40 was the second of six groups to co-discover the kilonova, associated with GW 170817.

In addition to SNe, DLT40 can be modified to find electromagnetic counterparts to gravitational-wave (GW) events, by imaging all galaxies in each (3D) LIGO/Virgo localization and automatically comparing them to previously acquired reference images. DLT40’s co-discovery of the optical counterpart to LIGO/Virgo’s GW 170817 binary neutron star merger was presented in the LIGO multi-messenger and Hubble’s Constant papers, and more detailed analyses were presented in three additional papers. Also, since the host galaxy of GW 170817 is part of DLT40’s catalog, DLT40 was able to provide the best pre-discovery detection limit and a dense monitoring of the host galaxy for six months prior to the event with an almost daily cadence.

Construction of PROMPT-Athabasca in mid-latitude Alberta, Canada.

Construction of PROMPT-Athabasca in mid-latitude Alberta, Canada.

We are now working to significantly expand DLT40’s GW-counterpart search and discovery footprint, from the two identical PROMPT telescopes that DLT40 is currently using to a total of up to nine PROMPT telescopes spanning five dark sites in both southern and northern hemispheres. This will significantly increase the chances of early/real-time discovery, as well as the density of multiband light curves, post-discovery.

Identical systems are required to leverage DLT40’s real-time, difference-imaging pipeline. Skynet is still in possession of all five of its original, identical 16-inch diameter, f/11.2 RC Optical Systems telescopes, as well as another four identical 17-inch diameter, f/6.8 PlaneWave Instruments telescopes.

The weather is good ≈300 nights/yr at CTIO, but only ≈200 nights/yr at the other sites. It is for this reason that we are targeting two, well-separated sites in each of Australia and Canada, better ensuring near-continuous coverage in the south, as well as coverage in the north.

Absolute magnitude reached for a survey of depth V = 19 mag as a function of distance. The DLT40 SN search extends to 40 Mpc; DLT40+ GW searches will extend to 65 Mpc, corresponding to 4.3 times the volume. The peak absolute magnitude of the kilonova associated with GW 170817 is shown in red.

Absolute magnitude reached for a survey of depth V = 19 mag as a function of distance. The DLT40 SN search extends to 40 Mpc; DLT40+ GW searches will extend to 65 Mpc, corresponding to 4.3 times the volume. The peak absolute magnitude of the kilonova associated with GW 170817 is shown in red.

Successful, real-time counterpart identification requires pre-event imaging of all galaxies in the (3D) LIGO/Virgo localization (typically 30 – 50 galaxies), for image differencing. For southern-hemisphere galaxies within 40 Mpc, these are collected repeatedly as part of DLT40’s SN search. For more distant galaxies in the south, and for all galaxies in the north, we are now banking single-epoch reference images, out to 65 Mpc (this number is driven by Virgo’s anticipated sensitivity in 2018 — 2019). With filterless observations and a 45-s integration time, we typically reach r ≈ 19 — 20 mag, which is good enough to detect kilonovae similar to GW 170817’s throughout this volume. As such, we are anticipating roughly four times the event rate (roughly 4/yr).

Upon receiving a LIGO/Virgo localization, DLT40 will automatically switch from using its two southern-hemisphere PROMPT telescopes at regular priority to using all of these PROMPT telescopes at target-of-opportunity (TOO) priority, through Skynet’s Director Discretionary time. DLT40 will narrow its search to galaxies within the LIGO/Virgo localization. These galaxies will be prioritized by their position within the localization, and these priorities will be automatically updated as the localization is, and especially if we detect candidates proximal to these galaxies. Repeating observations of these galaxies will then be divided between all available DLT40 telescopes (as many as 4 — 5 at a time in the case of mid-latitude or Australian localizations), in hopes of detecting and identifying the counterpart more quickly.

Skynet returns reduced images within the minute, after which they are further processed by DLT40’s pipeline, which includes quality checks, image differencing (with HOTPANTS), candidate detection and scoring (using a machine-learning algorithm), databasing, and website generation of stamps for manual inspection. The average lag between completing an observation and visualizing candidates is 3 — 4 minutes. Once a candidate is confirmed, DLT40 will automatically trigger Skynet’s Campaign Manager (see below). DLT40 also makes the counterpart’s position and magnitude public immediately, and then triggers observations at other observatories (e.g., the 4.1-meter diameter SOAR telescope).

Combined observing plan. LIGO/Virgo localization and updates, Evryscope candidates (especially those proximal to galaxies in the (3D) LIGO/Virgo localization), and ultimately DLT40 imaging are used to identify the counterpart quickly and automatically. Non-DLT40 Skynet telescopes will additionally collect low-density, unfiltered observations pre-identification, and Skynet’s Campaign Manager will semi-autonomously collect high-density, multi-wavelength light curves post-identification. For southern-hemisphere targets, SOAR will additionally collect spectra post-identification. Dashed arrows are for manual operations.

Combined observing plan. LIGO/Virgo localization and updates, Evryscope candidates (especially those proximal to galaxies in the (3D) LIGO/Virgo localization), and ultimately DLT40 imaging are used to identify the counterpart quickly and automatically. Non-DLT40 Skynet telescopes will additionally collect low-density, unfiltered observations pre-identification, and Skynet’s Campaign Manager will semi-autonomously collect high-density, multi-wavelength light curves post-identification. For southern-hemisphere targets, SOAR will additionally collect spectra post-identification. Dashed arrows are for manual operations.

Campaign Manager

Prior to counterpart identification, the other (non-DLT40) Skynet telescopes will divide up and repeatedly observe DLT40’s prioritized list of galaxies, with observations submitted automatically by DLT40 via Skynet’s API. Unlike the DLT40 telescopes, these telescopes are heterogeneous, making automated differencing with DLT40’s reference images unattractive. However, depending on how many of these telescopes are available at any given time, they will likely collect multiple, unfiltered, pre-identification observations of the counterpart. Furthermore, Skynet periodically measures the typical limiting magnitude of each of its telescope/filter combinations, and uses this information to automatically scale exposure durations to achieve a common S/N, despite this heterogeneity.

Counterpart identification can occur in three ways: (1) from DLT40 image differencing; (2) if no DLT40 telescopes are available, from a single Evryscope candidate proximal to a galaxy in the 3D LIGO/Virgo localization; and (3) from an external group. (1) and (2) will result in DLT40 automatically activating Skynet’s Campaign Manager (CM), via our API. The CM can also be activated manually by any of us in the case of (3).

The CM is a new observing mode that we have been developing for Skynet. It is available to any Skynet user, but is most effective when coupled with target-of-opportunity (TOO) access through Skynet’s Director Discretionary time, which this program has. The CM can also be used to follow up other transient phenomena, such as gamma-ray burst (GRB) localizations; other DLT40 events, such as young SNe; other Evryscope events, such as the Proxima Centauri superflare; other externally identified and reported events; and eventually, sufficiently bright LSST events. However, GW events will remain highest priority for the next few years.

Users predefine campaign prescriptions, which can then be activated via our API, or manually. A campaign prescription consists of: (1) a desired S/N per exposure or exposure block; (2) an assumed brightness in a user-selected filter at a user-selected time, post-event; (3) an assumed temporal index (both power-law and exponential options are available); (4) an assumed spectral index; and (5) a maximum acceptable exposure or exposure-block duration.

Prescriptions also consist of an initial list of telescopes and an initial list of filters for each telescope. For GW events, this will be all Skynet telescopes, and as many of UBVRI or ugriz that are available on each telescope, as well as one high-throughput option (e.g., open, clear, or luminance) per telescope, to better connect to Skynet’s pre-identification observations.

Once activated, the CM automatically runs through all prescribed filter lists on all prescribed telescopes, automatically scaling exposure or exposure-block durations from scope to scope, from filter to filter, and as a function of time, given each telescope/filter combination’s typical limiting magnitude and the assumed temporal and spectral indices, to achieve similar S/N per exposure or exposure block.

All of these images are available through Skynet in near-real time, and can be automatically calibrated, batch photometered, and custom stacked using Skynet’s Afterglow 2.0 image-analysis software. As new information is learned about the counterpart, such as its brightness at a certain time in a certain filter, or its temporal and spectral indices, the campaign prescription can be modified and the CM will automatically update upcoming exposures.

Finally, we can also discontinue use of particular telescopes, and particular filters (or add them back in), in real time. Otherwise, each telescope/filter combination automatically discontinues once the maximum acceptable exposure or exposure-block duration is exceeded.

Most of Skynet’s telescopes are currently located at CTIO, across Australia, and across North America. As the number and sizes of Skynet’s telescopes continue to increase, the CM will produce denser, and deeper, light curves.

The CM is nearly complete, with test runs using GRBs to begin prior to LIGO/Virgo’s next observing run in March or April 2019.

The densely sampled, X-ray, optical (including many Skynet/PROMPT observations), and NIR light curve of long-duration GRB 080710, and an example (but not yet formally fitted) off-axis, structured-jet model, at ≈10^16 Hz. In this particular model, the jet has a nearly classical profile, but is viewed 6 º  off axis. The model flux is obtained by integrating over the 3D emitting region, for each desired equal-arrival-time surface. The full model includes spectral evolution, such as synchrotron cooling. Although GRB 080710 is not an sGRB, we will use these data to determine what can be inferred, how accurately it can be inferred, correlations between fitted parameter values, and dependences on model assumptions (such as the density distribution of the external medium, energy injection, etc.), as fewer and fewer measurements are fitted to.

The densely sampled, X-ray, optical (including many Skynet/PROMPT observations), and NIR light curve of long-duration GRB 080710, and an example (but not yet formally fitted) off-axis, structured-jet model, at ≈10^16 Hz. In this particular model, the jet has a nearly classical profile, but is viewed 6º off axis. The model flux is obtained by integrating over the 3D emitting region, for each desired equal-arrival-time surface. The full model includes spectral evolution, such as synchrotron cooling. Although GRB 080710 is not an sGRB, we will use these data to determine what can be inferred, how accurately it can be inferred, correlations between fitted parameter values, and dependences on model assumptions (such as the density distribution of the external medium, energy injection, etc.), as fewer and fewer measurements are fitted to.

Afterglow Modeling and Hubble’s Constant

We have been developing off-axis, structured jet models of the short-duration GRB afterglows that should be associated with at least some of these GW events, to measure viewing angles. LIGO/Virgo currently has limited ability to constrain viewing angles, measuring only <36° (1-sigma) for GW 170817. This will improve as additional GW detectors come online, and ramp up to LIGO and Virgo’s sensitivities. But until then, if we can measure this for but one event, it can then be combined with the LIGO/Virgo data to obtain a “clean” measure of the distance to this event, independent of all of the calibration uncertainties, and potential systematics, of the cosmic distance ladder. This could then be used to resolve the growing “tension” between global measures of Hubble’s Constant, from the CMB and from Baryon Acoustic Oscillations, vs. higher ones from local measures, using Type Ia SNe.