Comments from Manodeep

[andrew-zentner]

i) HOD form: Zehavi 2011 had Msat convolved with . It's a choice, but might as well follow that choice to make comparison easier. And note that M1 in your equation should be labelled M1\prime -> M1 := M1\prime + M0. You should check that the MCMC run has this modulation factor in .

ii) Are the hlist files appropriate to use for HOD modelling? I think those come from running Consistent trees and most likely are not from STRICT_SO halo finding.

iii) Details about correlation matrix. Did you guys use a j/k corr matrix?

iv) The logMmin-siglogM degeneracy is mostly from number-density but is much more severe for the fainter samples. For the fainter sample, you can recover the number density by appropriately dialling the Mmin-siglogM tuple, and since bias is flat, the average bias does not change. However since bias is rapidly changing for brighter samples, you can't recover the average bias. In my case, the -19 sample 1-sigma contours extend all the way across my priors whereas the -21 sample the 1-sigma contours are actually smaller than the prior range. The 2-sigma still hits the prior range.

v) Using delta-chisqr = 1 to plot the various lines could be misleading (since the dof is not 1, delta-chisqr will be different for the 1-sigma level)

vi) What's the physical picture you guys are thinking for this galaxy assembly bias signal?

[andrew-zentner]

Only number (vi) remains to be addressed.

[andrew-zentner]

All addressed.