Anisotropy ansatz for the axisymmetric Jeans equations

The Jeans equations do not form a closed system, and to solve them a parametrization relating the velocity moments is often adopted. For axisymmetric models, a phenomenological choice (the 'b-ansatz') is widely used for the relation between the vertical (sigma(2)(z)) and radial (sigma(2)(R)) components of the velocity dispersion tensor, thus breaking their identity present in two-integral systems. However, the way in which the ansatz affects the resulting kinematical fields can be quite complicated, so that the analysis of these fields is usually performed only after numerically computing them. We present here a general procedure to study the properties of the ansatz-dependent fields v(phi)(2)& horbar;, Delta=v(phi)(2 & horbar;)-sigma(2)(z) and Delta R=v(phi)(2 & horbar;)sigma(2)(R). Specifically, the effects of the b-ansatz can be determined before solving the Jeans equations once the behaviour over the (R, z)-plane of three easy-to-build ansatz-independent functions is known. The procedure also constrains the ansatz to exclude unphysical results (as a negative v(phi)(2 & horbar;)). The method is illustrated by discussing the cases of three well-known galaxy models: the Miyamoto & Nagai and Satoh discs, and the Binney logarithmic halo, for which the regions and the constraints on the ansatz values can be determined analytically; a two-component (Miyamoto & Nagai plus logarithmic halo) model is also discussed.