We compare the performance of mass estimators for elliptical galaxies that rely on the directly observable surface brightness and velocity dispersion profiles , without invoking computationally expensive detailed modeling .
These methods recover the mass at a specific radius where the mass estimate is expected to be least sensitive to the anisotropy of stellar orbits .
One method ( ) uses the total luminosity-weighted velocity dispersion and evaluates the mass at a 3D half-light radius r _ { 1 / 2 } , i.e. , it depends on the global galaxy properties .
Another approach ( ) estimates the mass from the velocity dispersion at a radius R _ { 2 } where the surface brightness declines as R ^ { -2 } , i.e. , it depends on the local properties .
We evaluate the accuracy of the two methods for analytical models , simulated galaxies and real elliptical galaxies that have already been modeled by the Schwarzschild ’ s orbit-superposition technique .
Both estimators recover an almost unbiased circular speed estimate with a modest RMS scatter ( \lesssim 10 \% ) .
Tests on analytical models and simulated galaxies indicate that the local estimator has a smaller RMS scatter than the global one .
We show by examination of simulated galaxies that the projected velocity dispersion at R _ { 2 } could serve as a good proxy for the virial galaxy mass .
For simulated galaxies the total halo mass scales with \sigma _ { p } ( R _ { 2 } ) as \displaystyle M _ { vir } \left [ M _ { \odot } h ^ { -1 } \right ] \approx 6 \cdot 10 ^ { 12 } \left ( % \frac { \sigma _ { p } ( R _ { 2 } ) } { 200 km s ^ { -1 } } \right ) ^ { 4 } with RMS scatter \approx 40 \% .