Free Access

Table 7.

Test set of the m12f LSR1 catalog using stars with δϖ/ϖ < 0.10.

Data set Purity ϵA
S(star)> 0.75 41% 47%
S(star)> 0.95 59% 13%

Notes. The neural networks are trained on separate m12i galaxy simulations with truth-level information and then re-trained using only observable information from m12f LSR1. The optimal cut of 0.75 yields velocity distributions most similar to the truth-level accreted distributions, but only has a purity of 41%. This is comparable to the traditional methods listed in Table 1. A stronger cut of 0.95 increases the purity, but decreases the fraction of accreted stars that are selected, and can bias the velocity distributions. This provides an estimate for the purity and efficiency expected when applied to Gaia.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.