Barrier Q

[Ruby Ji]

Dear all,

I am trying to better understand how the precision matrix Q is calculated in the barrier model. The eq. 8 in the barrier paper (Bakka et al. 2019) is:

A = J - 1/8*sum(ranges[k]^2 * D[k])

 

In the Barrier.R, it is:

“A <- fem$I

  for (k in 1:xi) {

     A <- A + (ranges[k]^2 / 8.0) * fem$D[[k]]

  }”

 

I think that matrices I and J are the same thing. Why does equation 8 use a minus, while the code uses a plus? I am not a mathematician, so please forgive me if this is too basic.

Thanks,

Rujia