Axisymmetric orbit models of N-body merger remnants: a dependency of reconstructed mass on viewing angle

Abstract

We model mock observations of collisionless N-body disc-disc mergers with an axisymmetric orbit superposition program that has been used to model Coma ellipticals. The remnants sample representatively the shapes of disc-disc mergers including prolate, triaxial and oblate objects. The aim is to better understand how the assumption of axial symmetry affects reconstructed masses and stellar motions of systems which are intrinsically not axisymmetric, whether it leads to a bias and how such a potential bias can be recognised in models of real galaxies. The mass recovery at the half-light radius depends on viewing-angle and intrinsic shape: edge-on views allow to reconstruct total masses with an accuracy between 20% (triaxial/prolate remnants) and 3% (oblate remnant). Masses of highly flattened, face-on systems are underestimated by up to 50%. Deviations in local mass densities can be larger where remnants are strongly triaxial or prolate. Luminous M/L are sensitive to box orbits in the remnants. Box orbits cause the central kinematics to vary with viewing-angle. Reconstructed luminous M/L and central masses follow this variation. Luminous M/L are always underestimated (up to a factor of 2.5). Respective dark halos in the models can be overestimated by about the same amount, depending again on viewing angle. Reconstructed velocity anisotropies depend on viewing angle and on the orbital composition of the remnant. We construct N-body realisations of the Schwarzschild models to discuss chaotic orbits and the virial equilibrium in our models. Apparently flattened, rotating ellipticals of intermediate mass are likely close to both, axial symmetry and edge-on orientation. Our results imply that Schwarzschild models allow a reconstruction of their masses and stellar anisotropies with high accuracy. (abridged)