The Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey ( CANDELS ) and Cluster Lensing And Supernova survey with Hubble ( CLASH ) multi-cycle treasury programs with the Hubble Space Telescope ( HST ) have provided new opportunities to probe the rate of core-collapse supernovae ( CCSNe ) at high redshift , now extending to z \approx 2.5 .
Here we use a sample of approximately 44 CCSNe to determine volumetric rates , R _ { CC } , in six redshift bins in the range 0.1 < z < 2.5 .
Together with rates from our previous HST program , and rates from the literature , we trace a more complete history of R _ { CC } ( z ) , with R _ { CC } = 0.72 \pm 0.06 yr ^ { -1 } Mpc ^ { -3 } 10 ^ { -4 } h _ { 70 } ^ { 3 } at z < 0.08 , and increasing to 3.7 ^ { +3.1 } _ { -1.6 } yr ^ { -1 } Mpc ^ { -3 } 10 ^ { -4 } h _ { 70 } ^ { 3 } to z \approx 2.0 .
The statistical precision in each bin is several factors better than than the systematic error , with significant contributions from host extinction , and average peak absolute magnitudes of the assumed luminosity functions for CCSN types .
Assuming negligible time delays from stellar formation to explosion , we find these composite CCSN rates to be in excellent agreement with cosmic star formation rate density ( SFRs ) derived largely from dust-corrected rest-frame UV emission , with a scaling factor of k = 0.0091 \pm 0.0017 M ^ { -1 } _ { \odot } , and inconsistent ( to > 95 \% confidence ) with SFRs from IR luminous galaxies , or with SFR models that include simple evolution in the initial mass function over time .
This scaling factor is expected if the fraction of the IMF contributing to CCSN progenitors is in the 8 to 50 M _ { \odot } range .
It is not supportive , however , of an upper mass limit for progenitors at < 20 M _ { \odot } .