We use an adaptation of the BzK _ { s } Â technique to select \sim 40,000 z \sim 2 galaxies ( to K _ { AB } = 24 ) , including \sim 5,000 passively evolving ( PE ) objects ( to K _ { AB } = 23 ) , from 2.5 deg ^ { 2 } of deep CFTH imaging .
The passive galaxy luminosity function exhibits a clear peak at R = 22 and a declining faint-end slope ( \alpha = -0.12 ^ { +0.16 } _ { -0.14 } ) , while that of star-forming galaxies is characterized by a steep faint-end slope ( \alpha = -1.43 \pm 0.02 ( systematic ) ^ { +0.05 } _ { -0.04 } ( random ) ) .
The details of the LFs are somewhat sensitive ( at < 25 % level ) to cosmic variance even in these large ( \sim 0.5 deg ^ { 2 } ) fields , with the D2 field ( located in the COSMOS field ) most discrepant from the mean .
The shape of the z \sim 2 stellar mass function of passive galaxies is remarkably similar to that at z \sim 0.9 , save for a factor of \sim 4 lower number density .
This similarity suggests that the same mechanism may be responsible for the formation of passive galaxies seen at both these epochs .
This same formation mechanism may also operate down to z \sim 0 if the local PE galaxy mass function , known to be two-component , contains two distinct galaxy populations .
This scenario is qualitatively in agreement with recent phenomenological mass-quenching models and extends them to span more than three quarters of the history of the Universe .