EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 594 (October 2016) A&A, 594 (2016) A17 Full HTML
Planck 2015 results
Free Access

Table 17

Comparison of component separation methods, using Gaussian FFP8 simulations.

Input map Input map + foregrounds Cleaned map
Local Equilateral Orthogonal Local Equilateral Orthogonal Local Equilateral Orthogonal
SMICA
T .................. 5.2 ± 5.8 29 ± 71 −8 ± 34 −107.0 ± 5.8 −23 ± 71 27 ± 34 7.8 ± 5.8 38 ± 71 −20 ± 34
E .................. −39 ± 30 −99 ± 133 59 ± 69 −10 ± 30 −154 ± 133 −41 ± 69 −56 ± 30 −120 ± 133 65 ± 34
T+E .................. 5.9 ± 5.1 14 ± 45 −20 ± 22 −118.0 ± 5.1 −32 ± 45 8 ± 22 8.3 ± 5.2 14 ± 45 −22 ± 22
SEVEM
T .................. 5.6 ± 5.7 32 ± 69 −8 ± 32 −113.2 ± 5.7 −8 ± 69 34 ± 32 12.7 ± 5.7 35 ± 69 −25 ± 32
E .................. −17 ± 41 −149 ± 175 28 ± 95 −14 ± 41 −171 ± 175 −44 ± 95 −22 ± 41 −120 ± 175 41 ± 95
T+E ............... 7.7 ± 5.3 12 ± 49 −37 ± 24 −126.0 ± 5.3 −29 ± 49 −57 ± 25 13.0 ± 5.3 11 ± 49 −41 ± 24
NILC
T .................. 5.1 ± 5.7 32 ± 69 −5 ± 31 −102.0 ± 5.7 −14 ± 69 32 ± 31 17.8 ± 5.7 85 ± 69 −16 ± 31
E .................. −52 ± 33 −157 ± 156 72 ± 73 −6 ± 33 −155 ± 156 −47 ± 73 −76 ± 33 −179 ± 156 113 ± 73
T+E ............... 5.7 ± 5.0 7 ± 46 −15 ± 21 −117.0 ± 5.9 −27 ± 46 12 ± 21 15.8 ± 5.0 −20 ± 46 −7 ± 21
Commander
T .................. 0.5 ± 6.2 −5 ± 73 −14 ± 36 −127.0 ± 6.2 −25 ± 73 −137 ± 36 25.6 ± 6.2 67 ± 73 −17 ± 36
E .................. −51 ± 38 −64 ± 160 93 ± 86 −10 ± 38 −153 ± 160 −45 ± 86 −70 ± 38 −78 ± 159 138 ± 86
T+E ............. 1.6 ± 5.4 −2 ± 48 −21 ± 23 −137.0 ± 5.4 −29 ± 48 13 ± 23 20.4 ± 5.4 28 ± 48 −11 ± 23

Notes. We firstly consider Gaussian, foreground-free simulations, with simulated noise for each frequency band, process them through each of the four foreground cleaning pipelines, and measure fNL for the three standard shapes (columns labelled with “Input map”). We then include foregrounds and repeat the measurement, before applying the cleaning, and including realistic noise levels for each method (columns labelled with “Input map + foregrounds”); this step is performed in order to get an idea of the level of contamination introduced by foregrounds, before cleaning. Finally, we apply the different component separation methods, and again estimate fNL from the final maps (columns labelled with “Cleaned map”). The discrepancies between fNL measured on the input map, and fNL extracted from the cleaned map, provide a figure of merit to assess how well foregrounds are subtracted by different methods. Results below have been obtained with the KSW estimator and the “cleaned map” results were also checked with the binned estimator.

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.