Fig. 1

Operation of Darth Fader. The number of eigentemplates to be retained is at the discretion of the user, and may depend on the distribution of spectral types in the data. We have chosen to retain 20 eigentemplates because they encompass 99% of the total eigenvalue weight. In general the number retained will be significantly less than the number of spectra in the original template catalogue. The FDR denoising procedure denoises the positive and negative halves of the spectrum independently, with positivity and negativity constraints respectively. The requirement of six features or more in the denoised spectrum effectively cleans the catalogue of galaxies likely to yield catastrophic failures in their redshift estimates. It should be noted that a “no” decision represents the termination of that spectrum from the test catalogue and our analysis. Potentially, a further analysis could be employed at this point in order to review spectra possessing 5 features, for example, to see if further information, such as the association of standard lines to these features, can yield a redshift estimate (which could be used in conjunction with an estimate obtained from cross-correlation, with corresponding redshifts likely to be correct).
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.