search

lizzieinvancouver / pmm

phylogenetic mixed models -- and we all learn together
2 stars 1 forks source link

does Will's code work? #10

Closed lizzieinvancouver closed 3 years ago

lizzieinvancouver commented 4 years ago

Review Will's email, check code and output more fully....

lizzieinvancouver commented 4 years ago

running sims on midge! One to check is from Nacho:

m=15; lambda=0.2 results Stan: lam=0.98 results fitcontinuous: lam=0.32

lizzieinvancouver commented 3 years ago

Answer: Yes! And Geoff pointed out that the model's sigma seems to correspond to the rate of evolution across the tree (i.e., sig2 in fastBM command). So the model returns not only lambda, but also that sig2 value and the root trait value (b_z). (Will added, "Yes, that sounds absolutely right to me - that's what it is by design." I asked whether we should interpret it too much and he said, "I don't know how lousy the tree is; I guess you should be a little cautious given it's probably not a complete taxon sample but that probably matter the most for the estimates of the root state (which are always awful anyway!)."