Introduction

The Legacy Survey of Space and Time (LSST) is anticipated to encompass around 2 million observations spanning a decade, averaging 800 visits per night. To streamline the acquisition of these observations, we need a scheduler that is both fully automated and adaptable to weather conditions, observatory performance, and science priorities. The “Feature Based Scheduler” (FBS) scheduler module in rubin_scheduler encodes our current best approach to achieving these objectives for our all-sky multi-year survey.

These scheduling algorithms also need to be configured to best meet the LSST science goals. Short and long-term survey strategy choices must be evaluated using realistic simulations of the expected pointing history. The site_models and skybrightness_pre modules support these simulations by providing realistic weather telemetry, including seeing and cloud cover histories and pre-calculated skybrightness maps. The simulations also include a model of the observatory, providing accurate slew and settle times. Configuration of the FBS for the baseline survey strategy can be found in an example_scheduler in the scheduler module, and configuration variations for experiments in survey strategy can be found in various “sims_featureScheduler_runs*” github repositories within lsst-sims.

Simulation of pointing histories for LSST can be carried out using this software package alone. In operations, scheduling of the telescope happens within a framework of other software packages which transmit live telemetry of weather and observatory conditions as well as carry out additional safety checks for each visit and coordinate with summit databases to record relevant information, which are all necessary for the actual scheduling of Rubin Observatory. In both cases, the algorithms choosing the next visit are coming from rubin_scheduler.

When evaluating the outputs provided by rubin_scheduler, two additional software packages may be helpful: schedview which provides tools to gain further insight into the scheduler itself and rubin_sim which provides a framework for writing and applying metrics to the output pointing databases.