All Tables

Table 1: Table of the observations reduced and analysed for this work in addition to those reported by Kayser et al. (2005). The different field names in $\omega$ Cen are abbreviated by their location in the cluster (e.g om-e corresponds to east). The ending "rgb'' refers to red giant stars, "to'' to main-sequence turn-off stars, "ms'' correspondingly to upper main-sequence stars. The field of view of a MXU mask is 6 $\hbox {$.\mkern -4mu^\prime $ }$ 8 $\times$ 6 $\hbox {$.\mkern -4mu^\prime $ }$ 8.
Table 2: Summary of the stars used for validating the network model. The [Fe/H], $\log~g$ and $T_{\rm eff}$ values given are the averages from the works of several authors found in the literature (see Cayrel de Strobel et al. 2001). The uncertainties are either the standard deviations about the mean values (in the case that multiple values are listed) or the errors stated by the author (when there is only a single measurement). The last column shows the number of separate determinations in each case.
Table 3: Summary of the regularization results from neural networks with different training sets and weight decay term $\gamma$ . Increasing $\gamma$ results in smaller weights of the network and thus a more regularized solution. However, too large a value of $\gamma$ will again result in larger deviations, i.e. there is a trade-off in setting this parameter. The metallicity deviations for the standard stars are given in terms of the median values of the difference (computed value - literature value). It can be seen that training on noise-free data and validating on noisy data systematically underestimated metallicities. The results demonstrate that noise in the network inputs can help improve the regularization.
Table 4: Table of Strömgren observations of M 55. The columns show the central coordinates of the observed fields and the exposure times in the different filters. Certain parts of these fields overlap with those of the FORS2 spectroscopic observations.