brianstock-NOAA
commented
4 years ago e is defined in Egozcue (2003) at the bottom of p294 through top of p295. It is created on the R side in lines 88-94 of run_model.R, then passed to JAGS. In the comments of run_model.R:
# make 'e', an Aitchision-orthonormal basis on the S^d simplex (equation 18, page 292, Egozcue 2003)
# 'e' is used in the inverse-ILR transform (we pass 'e' to JAGS, where we do the ILR and inverse-ILR)Below that in run_model.R we also create cross and tmp.p that are used in the inverse-ILR.
PavelErnesto
Hello Brian,
I am going through the JAGS model. I am little lost on the part of inverse ILR math. I am wondering what does e means. Is it the inverse function of the natural logarithm or is it a parameter? There is not a definition of e as a vector before.
Inverse ILR math (equation 24, page 294, Egozcue 2003)
for(i in 1:N){ for(j in 1:(n.sources-1)){ cross[i,,j] <- (e[,j]^ilr.tot[i,j])/sum(e[,j]^ilr.tot[i,j]); }