Dynamic scheduling: target of opportunity observations of gravitational wave events

Abstract

The simultaneous detection of electromagnetic and gravitational waves from the coalescence of two neutron stars (GW170817 and GRB170817A) has ushered in a new era of ‘multimessenger’ astronomy, with electromagnetic detections spanning from gamma to radio. This great opportunity for new scientific investigations raises the issue of how the available multimessenger tools can best be integrated to constitute a powerful method to study the transient Universe in particular. To facilitate the classification of possible optical counterparts to gravitational wave events, it is important to optimize the scheduling of observations and the filtering of transients, both key elements of the follow-up process. In this work, we describe the existing workflow whereby telescope networks such as GRANDMA and GROWTH are currently scheduled; we then present modifications we have developed for the scheduling process specifically, so as to face the relevant challenges that have appeared during the latest observing run of Advanced LIGO and Advanced Virgo. We address issues with scheduling more than one epoch for multiple fields within a skymap, especially for large and disjointed localizations. This is done in two ways: by optimizing the maximum number of fields that can be scheduled and by splitting up the lobes within the skymap by right ascension to be scheduled individually. In addition, we implement the ability to take previously observed fields into consideration when rescheduling. We show the improvements that these modifications produce in making the search for optical counterparts more efficient, and we point to areas needing further improvement.

Discovery and confirmation of the shortest gamma-ray burst from a collapsar

“…We ingested the GBM localization map into the GROWTH Target-of-Opportunity (ToO) marshal, an interactive tool design to plan and schedule ToO observations for ZTF 73 . The observation plan generated by the ToO marshal relies on gwemopt [104][105][106] , a code that optimizes the telescope scheduling process for skymaps with a healpix format, like the Fermi-GBM maps. The gwemopt procedure involves slicing the skymap into predefined tiles of the size and shape of the ZTF field-of-view, determining which fields have the highest enclosed probability, and optimizing observations based on airmass and visibility windows.…”

Optical follow-up of the neutron star–black hole mergers S200105ae and S200115j

“…For S200105ae, we split the schedule into two blocks of right ascension due to the significantly displaced lobes in the skymap (see Extended Data Figure 2), with observations lasting three hours per block. We additionally utilized the "filter balancing" feature 73 , which optimizes for the number of fields that have observations scheduled in all requested filters, and employed the greedy-slew algorithm 74 for conducting our search. The ability to split the skymap in right ascension and the use of filter balancing was novel for these observations, and served to help address the previous difficulty with multi-lobed skymaps to make it possible to observe all filters requested for the scheduled fields.…”

Implications of the search for optical counterparts during the second part of the Advanced LIGO’s and Advanced Virgo’s third observing run: lessons learned for future follow-up observations

“…There are three main aspects to this scheduling: tiling, time allocation, and scheduling of the requested observations. Multitelescope, network-level observations (Coughlin et al 2019a) and improvements for scheduling in the case of multilobed maps (Almualla et al 2020) are the most recent developments in these areas. We note that GWEMOPT naturally accounts for slew and read out times based on telescope-specific configuration parameters, which are important to account for inefficiencies in either long slews or when requesting short exposure times.…”

“…for scheduling in the case of multi-lobed maps (Almualla et al 2020) are the most recent developments in these areas. We note that gwemopt naturally accounts for slew and read out times based on telescope-specific configuration parameters, which are important to account for inefficiencies in either long slews or when requesting short exposure times.…”