Dwarf galaxies , among the most dark matter dominated structures of our universe , are excellent test-beds for dark matter theories .
Unfortunately , mass modelling of these systems suffers from the well documented mass-velocity anisotropy degeneracy .
For the case of spherically symmetric systems , we describe a method for non-parametric modelling of the radial and tangential velocity moments .
The method is a numerical velocity anisotropy “ inversion ” , with parametric mass models , where the radial velocity dispersion profile , \sigma _ { \mathrm { rr } } ^ { 2 } is modeled as a B-spline , and the optimization is a three step process that consists of : ( i ) an Evolutionary modelling to determine the mass model form and the best B-spline basis to represent \sigma _ { \mathrm { rr } } ^ { 2 } ; ( ii ) an optimization of the smoothing parameters ; ( iii ) a Markov chain Monte Carlo analysis to determine the physical parameters .
The mass-anisotropy degeneracy is reduced into mass model inference , irrespective of kinematics .
We test our method using synthetic data .
Our algorithm constructs the best kinematic profile and discriminates between competing dark matter models .
We apply our method to the Fornax dwarf spheroidal galaxy .
Using a King brightness profile and testing various dark matter mass models , our model inference favours a simple mass-follows-light system .
We find that the anisotropy profile of Fornax is tangential ( \beta ( r ) < 0 ) and we estimate a total mass of M _ { \text { tot } } = 1.613 ^ { +0.050 } _ { -0.075 } \times 10 ^ { 8 } \text { M } _ { \odot } , and a mass-to-light ratio of \Upsilon _ { V } = 8.93 ^ { +0.32 } _ { -0.47 } ( \text { M } _ { \odot } / \text { L } _ { \odot } ) .
The algorithm we present is a robust and computationally inexpensive method for non-parametric modelling of spherical clusters independent of the mass-anisotropy degeneracy .