Open Access

Table 4

Relative residual error of the RVs in percent after 50 training runs for the NNs using 540 000 starsim models for different time samplings and for different kinds and numbers of activity indices as input for ε Eri.

Input index uniform random1 gap1 random2 gap2 average with error
V 19.9 20.7 18.7 21.3 20.3 20.2±0.4 23.8
R 22.4 22.9 21.4 23.7 23.0 22.7±0.3 26.0
I 23.9 24.7 24.2 25.5 24.7 24.6±0.2 27.5
V-I 8.6 9.5 8.6 9.9 8.8 9.1±0.2 22.4
V-R-I 8.7 9.6 8.5 10.0 8.8 9.1±0.2 22.0
BIS 17.0 19.3 18.8 19.7 18.2 18.6±0.4 19.1
FWHM 23.1 29.1 27.0 28.5 24.9 26.5±1.0 27.1
CON 12.6 12.4 10.4 13.1 12.0 12.1±0.4 20.6
BIS-FWHM 11.9 13.9 13.4 14.0 13.0 13.2±0.3 14.5
BIS-CON 6.5 7.8 6.1 7.8 7.2 7.1±0.3 9.1
FWHM-CON 7.5 7.9 6.6 8.7 7.8 7.7±0.3 10.8
BIS-FWHM-CON 5.1 5.8 5.5 6.5 5.9 5.8±0.2 7.5
all 6 4.7 5.7 4.9 5.6 5.3 5.2±0.2 6.1
time-BIS 17.0 19.2 18.9 19.7 18.2 18.6±0.4 19.1

Notes. Column ‘average’ shows the average of all the values of the other time samplings and their standard deviation, ‘with error’ shows the RV RRE of the ‘uniform’ time sampling, where the additional RREs (see Table 2) are added to each index of the NN input data.

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.