AstroQ: Automated Scheduling of Cadenced Astronomical Observations

Astronomy relies heavily on time domain observations. To maximize the scientific yield of such observations, astronomers must carefully match the observational cadence to the phenomena of interest. This presents significant scheduling challenges for observatories with multiple large programs, each with different cadence needs. To address this challenge, we developed AstroQ, an automated framework for scheduling cadenced observations. We tested this on a suite of Doppler exoplanet programs at Keck Observatory, where the algorithm powers the KPF-Community Cadence project. As a point of reference, AstroQ can determine the provably optimal ordering of 3680 observations of 200 targets -- each with its own cadence needs and accessibility constraints -- over a six month period to five minute time resolution. Schedules of this size may be constructed in ~120 seconds on modern workstation, enabling dynamic rescheduling due to weather changes, target-of-opportunity interrupts, and other needs. A key advantage of AstroQ over manual scheduling is realistic projections of program completion, savings in human effort, and elimination of human bias in balancing many programs. AstroQ is open source and may be applied to other scheduling needs, both in astronomy and beyond.