Histograms

[eduardrusu]

Histograms

[drphilmarshall]

Thanks for posting the images on the wiki!

• Good question about how to decide which weights to use. I think we want to optimize our choices for accuracy, which means acceptably low bias and then maximum precision (smallest possible statistical uncertainty). So we'll know a lot more after probing the systematics - and maybe repeating our tests as best we can on the sims.
• With this working assumption that Pr(Mstar | z) = Pr(Mstar) (ie that z has been marginalised over), we are being maximally conservative: the joint Pr(Mstar,z) should give us more precision. I think this is the case for any weights that use Mstar, not just the ones that use both Mstar and z. It may not be a big concern when the photo-zs are good.

[eduardrusu]

Will close this issue, because the relevant discussion is in Rusu et al. 2017.