We present a power spectrum analysis of the ALMA Spectroscopic Survey Large Program ( ASPECS LP ) data from 84 to 115 GHz .
These data predominantly probe small-scale fluctuations ( k = 10 –100 h Mpc ^ { -1 } ) in the aggregate CO emission in galaxies at 1 \lesssim z \lesssim 4 .
We place an integral constraint on CO luminosity functions ( LFs ) in this redshift range via a direct measurement of their second moments in the three-dimensional ( 3D ) auto-power spectrum , finding a total CO shot noise power P _ { \operatorname { CO } , \operatorname { CO } } ( k _ { \operatorname { CO ( 2 - 1 ) } } ) \leq 1.9 % \times 10 ^ { 2 } \mu K ^ { 2 } ( Mpc h ^ { -1 } ) ^ { 3 } .
This upper limit ( 3 \sigma ) is consistent with the observed ASPECS CO LFs in \citet Decarli2019_3mm , but rules out a large space in the range of P _ { \operatorname { CO } , \operatorname { CO } } ( k _ { \operatorname { CO ( 2 - 1 ) } } ) inferred from these LFs , which we attribute primarily to large uncertainties in the normalization \Phi _ { * } and knee L _ { * } of the Schechter-form CO LFs at z > 2 .
Also , through power spectrum analyses of ASPECS LP data with 415 positions from galaxies with available optical spectroscopic redshifts , we find that contributions to the observed mean CO intensity and shot noise power of MUSE galaxies are largely accounted for by ASPECS blind detections , though there are \sim 20 % contributions to the CO ( 2-1 ) mean intensity due to sources previously undetected in the blind line search .
Finally , we sum the fluxes from individual blind CO detections to yield a lower limit on the mean CO surface brightness at 99 GHz of \langle T _ { \operatorname { CO } } \rangle = 0.55 \pm 0.02 \mu K , which we estimate represents 68–80 % of the total CO surface brightness at this frequency .