... MS2137-23

Appendices A and B are only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... photometric

Using hyperz facilities (Bolzonella et al. 2000), see also G03.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... library

http://cernlib.web.cern.ch/cernlib/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... elliptical

Which are not approximated by elliptical lens potentials (numerically faster but leading to unphysical surface mass density at large radius).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... modeled

If we have used the former positional uncertainties of G03, the best fit model would not have been changed but we would have found a minimum $\chi^2_{\rm src}/{\rm d.o.f.}=1.64$ which is also acceptable fit.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... satisfying

It would have increased to $\chi^2_{\rm src}/{\rm d.o.f.}=1.39$ using the G03uncertainties.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...$r\sim 0.1\hbox {$^{\prime \prime }$ }$)

Which makes the issue of the fifth image worth addressing observationally.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... model

We find $\chi^2_{\rm kin}/N_{\rm bins}\sim8$ if we change $\sigma _{\rm h_4}$ by $\sigma _{\rm los}$ in Eq. (8), i.e. if we neglect the velocity bias due to non-Gaussian LOSVD. Consequently, the correction has a weak effect on the modelling.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...$\vec{R}_i \rightarrow \vec{R}_i + \vec{\delta R}_i$

Where $\vec{\delta R}$ may follow a 2D Gaussian distribution with standard deviation $\sigma_{\rm seeing}={\it FWHM}/2.35$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.