We present a search for fast optical transients ( \tau \sim 0.5 { hr } -1 { d } ) using repeated observations of the Pan-STARRS1 Medium-Deep Survey ( PS1/MDS ) fields .
Our search takes advantage of the consecutive g _ { P 1 } r _ { P 1 } observations ( 16.5 min in each filter ) , by requiring detections in both bands , with non-detections on preceding and subsequent nights .
We identify 19 transients brighter than 22.5 AB mag ( { S / N } \gtrsim 10 ) .
Of these , 11 events exhibit quiescent counterparts in the deep PS1/MDS templates that we identify as M4–M9 dwarfs at d \approx 0.2 - 1.2 kpc .
The remaining 8 transients lack quiescent counterparts , exhibit mild but significant astrometric shifts between the g _ { P 1 } and r _ { P 1 } images , colors of ( g - r ) _ { P 1 } \approx 0.5 - 0.8 mag , non-varying light curves , and locations near the ecliptic plane with solar elongations of about 130 deg , which are all indicative of main-belt asteroids near the stationary point of their orbits .
With identifications for all 19 transients , we place an upper limit of R _ { FOT } ( \tau \sim 0.5 { hr } ) \lesssim 0.12 deg ^ { -2 } d ^ { -1 } ( 95 \% confidence level ) on the sky-projected rate of extragalactic fast transients at \lesssim 22.5 mag , a factor of 30 - 50 times lower than previous limits ; the limit for a timescale of \sim { day } is R _ { FOT } \lesssim 2.4 \times 10 ^ { -3 } deg ^ { -2 } d ^ { -1 } .
To convert these sky-projected rates to volumetric rates , we explore the expected peak luminosities of fast optical transients powered by various mechanisms , and find that non-relativistic events are limited to M \approx - 10 mag ( M \approx - 14 mag ) for a timescale of \sim 0.5 hr ( \sim { day } ) , while relativistic sources ( e.g. , gamma-ray bursts , magnetar-powered transients ) can reach much larger luminosities .
The resulting volumetric rates are \lesssim 13 Mpc ^ { -3 } yr ^ { -1 } ( M \approx - 10 mag ) , \lesssim 0.05 Mpc ^ { -3 } yr ^ { -1 } ( M \approx - 14 mag ) and \lesssim 10 ^ { -6 } Mpc ^ { -3 } yr ^ { -1 } ( M \approx - 24 mag ) , significantly above the nova , supernova , and GRB rates , respectively , indicating that much larger surveys are required to provide meaningful constraints .
Motivated by the results of our search we discuss strategies for identifying fast optical transients in the LSST main survey , and reach the optimistic conclusion that the veil of foreground contaminants can be lifted with the survey data , without the need for expensive follow-up observations .