We use a complete and uniform sample of almost half a million galaxies from the Sloan Digital Sky Survey to characterise the distribution of stellar mass in the low-redshift Universe .
Galaxy abundances are well determined over almost four orders of magnitude in stellar mass , and are reasonably but not perfectly fit by a Schechter function with characteristic stellar mass m _ { \ast } = 6.7 \times 10 ^ { 10 } M _ { \odot } and with faint-end slope \alpha = -1.155 .
For a standard cosmology and a standard stellar Initial Mass Function , only 3.5 % of the baryons in the low-redshift Universe are locked up in stars .
The projected autocorrelation function of stellar mass is robustly and precisely determined for r _ { p } < 30 h ^ { -1 } { Mpc } .
Over the range 10 h ^ { -1 } { kpc } < r _ { p } < 10 h ^ { -1 } { Mpc } it is extremely well represented by a power law .
The corresponding three-dimensional autocorrelation function is \xi ^ { \ast } ( r ) = ( r / 6.1 h ^ { -1 } { Mpc } ) ^ { -1.84 } .
Relative to the dark matter , the bias of the stellar mass distribution is approximately constant on large scales , but varies by a factor of five for r _ { p } < 1 h ^ { -1 } { Mpc } .
This behaviour is approximately but not perfectly reproduced by current models for galaxy formation in the concordance \Lambda CDM cosmology .
Detailed comparison suggests that a fluctuation amplitude \sigma _ { 8 } \sim 0.8 is preferred to the somewhat larger value adopted in the Millennium Simulation models with which we compare our data .
This comparison also suggests that observations of stellar mass autocorrelations as a function of redshift might provide a powerful test for the nature of Dark Energy .