Entry PQNLD

[AngusWright]

Challenge Submission Porque no los dos

This is a submission discussed at the DESC Collaboration meeting. It invokes a combination of template photo-z and machine learning to assign sources to tomographic bins. Template photoz are computed with BPZ. These photo-z, along with multiband photometry, are used to train a self-organising map. Sources are then assigned to cells within the map, which are combined in the same manner as the SImpleSOM submission to create tomographic bins.

Examples of performance will be posted below.

[AngusWright]

Example stats for PQNLD:
5 bins {'SNR_ww': 252.23052841648075, 'SNR_gg': 1345.72833444563, 'SNR_3x2': 1346.5142258688065, 'FOM_3x2': 23122.107897453167} 10 bins {'SNR_ww': 254.15503568121824, 'SNR_gg ': 1633.468888855488, 'SNR_3x2': 1633.9969077616179, 'FOM_3x2': 38831.03840228895}

[AngusWright]

This method isn't optimised for SNR in any way. It is therefore reasonable to expect that the final bins do not encore the full information accessible to the method. This methods strength lies in its diagnostic power; when the assumption of perfectly representative training is broken, this method can limit degradation of the binning.